SOLAR RESOURCE AND PV POTENTIAL OF THE MALDIVES 24 MONTH SOLAR RESOURCE REPORT September 2018 This report was prepared by Solargis, under contract to The World Bank. It is one of several outputs from the solar resource mapping component of the activity Energy Resource Mapping and Geospatial Planning Maldives [Project ID: P146018]. This activity is funded and supported by the Energy Sector Management Assistance Program (ESMAP), a multi-donor trust fund administered by The World Bank, under a global initiative on Renewable Energy Resource Mapping. Further details on the initiative can be obtained from the ESMAP website. This document is a final output from the above-mentioned project, and the content is the sole responsibility of the consultant authors. Users are strongly advised to exercise caution when utilizing the information and data contained, as this may include preliminary data and/or findings, and the document has not been subject to full peer review. Final outputs from this project will be marked as such, and any improved or validated solar resource data will be incorporated into the Global Solar Atlas. Copyright © 2018 THE WORLD BANK Washington DC 20433 Telephone: +1-202-473-1000 Internet: www.worldbank.org The World Bank, comprising the International Bank for Reconstruction and Development (IBRD) and the International Development Association (IDA), is the commissioning agent and copyright holder for this publication. However, this work is a product of the consultants listed, and not of World Bank staff. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work and accept no responsibility for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for non-commercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: +1-202-522-2625; e-mail: pubrights@worldbank.org. Furthermore, the ESMAP Program Manager would appreciate receiving a copy of the publication that uses this publication for its source sent in care of the address above, or to esmap@worldbank.org. Annual Solar Resource Report for solar meteorological stations after completion of 24 months of measurements Republic of Maldives Reference No. 129-07/2018 Date: 26 September 2018 Customer Consultant World Bank Solargis s.r.o. Energy Sector Management Assistance Program Contact: Mr. Marcel Suri Contact: Mr. Sandeep Kohli Mytna 48, 811 07 Bratislava, Slovakia 1818 H St NW, Washington DC, 20433, USA Phone +421 2 4319 1708 Phone: +1-202-361-0033 E-mail: marcel.suri@solargis.com E-mail: skohli@worldbank.org http://solargis.com http://www.esmap.org/RE_Mapping Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/201824 month TABLE OF CONTENTS Table of contents .................................................................................................................................................. 4 Acronyms .............................................................................................................................................................. 6 Glossary ................................................................................................................................................................ 7 1 Introduction .................................................................................................................................................. 9 1.1 Background ....................................................................................................................................... 9 1.1 Delivered data sets ............................................................................................................................ 9 1.2 Information included in this report ...................................................................................................10 2 Position of solar meteorological sites ........................................................................................................11 3 Ground measurements in Maldives .............................................................................................................13 3.1 Instruments and measured parameters ...........................................................................................13 3.2 Station operation and instruments’ field verification ........................................................................14 3.3 Quality control of measured solar resource data ..............................................................................17 3.3.1 Hanimaadhoo Airport..........................................................................................................17 3.3.2 Hulhulé Airport ....................................................................................................................21 3.3.3 Kadhdhoo Airport ................................................................................................................25 3.3.4 Gan Airport..........................................................................................................................29 3.4 Recommendations on the operation and maintenance of the sites .................................................33 3.5 Instruments re-calibration at the end of measurement campaign ....................................................33 4 Solar resource model data ..........................................................................................................................36 4.1 Solar model ......................................................................................................................................36 4.2 Site adaptation of the solar model − method ...................................................................................37 4.3 Results of the model adaptation at four sites ...................................................................................39 5 Meteorological model data..........................................................................................................................46 5.1 Meteorological model ......................................................................................................................46 5.2 Validation of meteorological data ....................................................................................................46 5.2.1 Air temperature at 2 metres ................................................................................................47 5.2.2 Relative humidity.................................................................................................................49 5.2.3 Wind speed and wind direction at 10 metres ......................................................................52 5.2.4 Atmospheric pressure.........................................................................................................54 5.3 Uncertainty of meteorological model data ........................................................................................54 6 Solar resource: uncertainty of longterm estimates .....................................................................................56 6.1 Uncertainty of solar resource yearly estimate ..................................................................................56 6.2 Uncertainty due to interannual variability of solar radiation ..............................................................57 6.3 Combined uncertainty ......................................................................................................................59 7 Time series and Typical Meteorological Year data .....................................................................................62 7.1 Delivered data sets ...........................................................................................................................62 7.2 TMY method.....................................................................................................................................63 7.3 Results .............................................................................................................................................63 8 Conclusions .................................................................................................................................................68 Annex 1: Site related data statistics ....................................................................................................................69 © 2018 Solargis page 4 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 2424 months of measurements Solargis reference No. 129-07/2018 Yearly summaries of solar and meteorological parameters.........................................................................69 Monthly summaries of solar and meteorological parameters .....................................................................71 Frequency of occurrence of GHI and DNI daily model values for a period 1999 to 2017 .............................73 Frequency of occurrence of GHI and DNI 30-minute model values for a period 1999 to 2017 .....................77 Frequency of occurrence of GHI and DNI measured and model values .......................................................81 List of figures .......................................................................................................................................................87 List of tables ........................................................................................................................................................89 References ...........................................................................................................................................................91 Support information .............................................................................................................................................93 Background on Solargis ................................................................................. Error! Bookmark not defined. Legal information........................................................................................... Error! Bookmark not defined. © 2018 Solargis page 5 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 2424 months of measurements Solargis reference No. 129-07/2018 ACRONYMS AP Atmospheric Pressure CFSR Climate Forecast System Reanalysis. The meteorological model operated by the US service NOAA (National Oceanic and Atmospheric Administration) CFS v2 Climate Forecast System Version 2 CFSv2 model is the operational extension of the CFSR (NOAA, NCEP) DIF Diffuse Horizontal Irradiation, if integrated solar energy is assumed. Diffuse Horizontal Irradiance, if solar power values are discussed DNI Direct Normal Irradiation, if integrated solar energy is assumed. Direct Normal Irradiance, if solar power values are discussed. GFS Global Forecast System. The meteorological model operated by the US service NOAA (National Oceanic and Atmospheric Administration) GHI Global Horizontal Irradiation, if integrated solar energy is assumed. Global Horizontal Irradiance, if solar power values are discussed. MACC Monitoring Atmospheric Composition and Climate – meteorological model operated by the European service ECMWF (European Centre for Medium-Range Weather Forecasts) Meteosat IODC (MFG) Meteosat satellites operated by EUMETSAT organization. MFG: Meteosat First Generation. For this report the data from the Meteosat IODC are used. The satellite is positioned over the Indian Ocean. PWAT Precipitable water (water vaour) QC Quality control RH Relative Humidity at 2 metres TEMP Air Temperature at 2 metres WD Wind Direction at 10 metres WS Wind Speed at 10 metres © 2018 Solargis page 6 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 GLOSSARY Aerosols Small solid or liquid particles suspended in air, for example desert sand or soil particles, sea salts, burning biomass, pollen, industrial and traffic pollution. All-sky irradiance The amount of solar radiation reaching the Earth's surface is mainly determined by Earth-Sun geometry (the position of a point on the Earth's surface relative to the Sun which is determined by latitude, the time of year and the time of day) and the atmospheric conditions (the level of cloud cover and the optical transparency of atmosphere). All-sky irradiance is computed with all factors taken into account Bias Represents systematic deviation (over- or underestimation) and it is determined by systematic or seasonal issues in cloud identification algorithms, coarse resolution and regional imperfections of atmospheric data (aerosols, water vapour), terrain, sun position, satellite viewing angle, microclimate effects, high mountains, etc. Clear-sky irradiance The clear sky irradiance is calculated similarly to all-sky irradiance but without taking into account the impact of cloud cover. Long-term average Average value of selected parameter (GHI, DNI, etc.) based on multiyear historical time series. Long-term averages provide a basic overview of solar resource availability and its seasonal variability. P50 value Best estimate or median value represents 50% probability of exceedance. For annual and monthly solar irradiation summaries it is close to average, since multiyear distribution of solar radiation resembles normal distribution. P90 value Conservative estimate, assuming 90% probability of exceedance (with the 90% probability the value should be exceeded). When assuming normal distribution, the P90 value is also a lower boundary of the 80% probability of occurrence. P90 value can be calculated by subtracting uncertainty from the P50 value. In this report, we apply a simplified assumption of normal distribution of yearly values. Root Mean Square Represents spread of deviations given by random discrepancies between measured and Deviation (RMSD) modelled data and is calculated according to this formula: 2 ∑ =1( − ) = √ On the modelling side, this could be low accuracy of cloud estimate (e.g. intermediate clouds), under/over estimation of atmospheric input data, terrain, microclimate and other effects, which are not captured by the model. Part of this discrepancy is natural - as satellite monitors large area (of approx. 3.3 x 3.4 km for MSG satellite), while sensor sees only micro area of approx. 1 sq. centimetre. On the measurement side, the discrepancy may be determined by accuracy/quality and errors of the instrument, pollution of the detector, misalignment, data loggers, insufficient quality control, etc. Solar irradiance Solar power (instantaneous energy) falling on a unit area per unit time [W/m 2]. Solar resource or solar radiation is used when considering both irradiance and irradiation. Solar irradiation Amount of solar energy falling on a unit area over a stated time interval [Wh/m 2 or kWh/m2]. Uncertainty Is a parameter characterizing the possible dispersion of the values attributed to an of estimate, Uest estimated irradiance/irradiation values. In this report, uncertainty assessment of the solar resource model estimate is based on a detailed understanding of the achievable accuracy of the solar radiation model and its data inputs (satellite, atmospheric and other data), which is confronted by an extensive data validation experience. The second source of uncertainty is ground measurements. Their quality depends on accuracy of instruments, their maintenance and data quality control. Third contribution to the © 2018 Solargis page 7 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 uncertainty is from the site adaptation method where ground-measured and satellite- based data are correlated. © 2018 Solargis page 8 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 1 INTRODUCTION 1.1 Background This report is prepared within Phase 2 of the project Renewable Energy Resource Mapping for the Republic of the Maldives. This part of the project focuses on solar resource mapping and measurement services as part of a technical assistance in the renewable energy development implemented by the World Bank in Maldives. It is being undertaken in close coordination with the Ministry of Environment and Energy (MEE) of Maldives, the World Bank’s primary country counterpart for this project. This project is funded by the Energy Sector Management Assistance Program (ESMAP) and Asia Sustainable and Alternative Energy Program (ASTAE), both administered by the World Bank and supported by bilateral donors. This report summarizes results of two years of the measuring campaign by the project ESMAP at four solar meteorological stations, installed in Maldives as part of the World Bank’s ESMAP mission in Maldives. This report describes delivery of site-specific data, site adaptation of solar model, data uncertainty and statistical summary. This report accompanies delivery of site-specific solar resource and meteorological data for four sites, where solar meteorological stations have been operated. As a result of high-quality operation of the meteorological sits, and site adaptation the Solargis model, reliable historical time series and TMY data is computed. The delivered time series and Typical Meteorological Year data is ready for use for bankable evaluation of solar energy projects. The measurements at four sites, located in Maldives, are provided by Suntrace company (Germany) with assistance of local company Renewable Energy Maldives (REM). The model data for the same sites and related works, together with this report are supplied by Solargis (Slovakia). 1.1 Delivered data sets The site specific data, provided as part of this delivery, include: • Solar and meteorological ground measurements, after second level quality assessment (first level delivered by Suntrace), representing period of minimum two years 12/2015 – 12/2017 (extended until 03/2018 and 04/2018) • Time series, representing last 19 years (01/1999 to 12/2017) • Typical Meteorological Year data, also representing last 19 calendar years (01/1999 to 12/2017) The data is delivered in formats ready to use in energy simulation software. This report provides detailed insight of the methodologies and reuslts. Table 1.1 Delivered data characteristics Feature Time coverage Primary time step Delivered files Measurements Dec 2015 to Dec 2017* 1 minute Quality controlled measurements – (Suntrace) * Planned two-year period extended until 1- minute March 2018 on sites where threshold 95% of quality data was not reached Model data – original Jan 1999 to Dec 2017 30 minutes Time series – hourly (Solargis) Time series – monthly Time series – yearly Model data – site adapted Jan 1999 to Dec 2017 30 minutes Time series – hourly (Solargis) Time series – monthly Time series – yearly Model data – site adapted Jan 1999 to Dec 2017 hourly Typical Meteorological Year P50 (Solargis) Typical Meteorological Year P90 © 2018 Solargis page 9 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 1.2 Parameters in the delivered site-adapted time series and TMY data (hourly time step) Parameter Acronym Unit TS TMY P50 TMY P90 Global horizontal irradiance GHI W/m2 X X X Direct normal irradiance DNI W/m2 X X X Diffuse horizontal irradiance DIF W/m2 X X X Global tilted irradiance (at optimum angle) GHI W/m2 X - - Solar azimuth SA ° X X X Solar elevation SE ° X X X Air temperature at 2 metres TEMP °C X X X Wind speed at 10 metres WS m/s X X X Wind direction at 10 metres WD ° X X X Relative humidity RH % X X X Air Pressure AP hPa X X X Precipitable Water PWAT kg/m2 X X X 1.2 Information included in this report This report presents: • Solar resource and meteorological measurements after 24 months of operation o Review and quality check of the measured data o Calibration procedures and results o List and explanation of the occurred disturbances and failures • Comparison of the measurements to the model and uncertainty analysis o Comparison of solar and meteo measurements with the model data o Site adaptation of satellite data based on ground measurements and uncertainty estimate o Estimate of data uncertainty • Data analysis (measured vs. modelled) o Monthly summaries of solar and meteorological parameters captured at the site o Variability of measured solar parameters o Frequency of occurrence of GHI and DNI 1-minute and 30-minute values o Frequency of occurrence of GHI and DNI 1-minute and 30-minute ramps © 2018 Solargis page 10 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 2 POSITION OF SOLAR METEOROLOGICAL SITES In Maldives, four measuring stations were installed within the ESMAP Solar initiative. All of them have been located at the airports within the premises of Maldives Meteorological Service (Figure 2.1, Table 2.1). Figure 2.1: Position of solar meteorological stations in Maldives Map of global horizontal irradiation in the background © 2018 Solargis page 11 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 2.1 Overview information on the solar meteorological stations installed in Maldives Site name Latitude Longitude Altitude (airport) Site ID [º] [º] [m a.s.l.] Measurement station host Hanimaadhoo MVHAQ 6.7482° 73.1696° 2 Hanimaadhoo International Airport Hulhulé MVMLE 4.1927° 73.5281° 2 Male International Airport Kadhdhoo MVKDO 1.8599° 73.5203° 2 Kadhdhoo Airport Gan MVGAN -0.6911° 73.1599° 2 Gan International Airport Besides the good geographical distribution within the territory of Maldives, the locations fit well to the population centres, where solar installations can be primarily deployed. In addition to geographical and socio-economic criteria, the sites fulfil the criteria for the operation and maintenance of the solar measuring stations: • Availability of free horizon, • Availability of GSM networks, • Availability of local work force for maintenance, • Easy to access and high level of security During the two years of the measurements, the measured data was analysed and harmonized with the objective to acquire reference solar radiation data for reducing the uncertainty of the model Chapter 6). The quality data from four Tier-2 meteorological stations are available for this assessment (Chapter 3). Position and detailed information about measurement sites is available also on Global Solar Atlas website: http://globalsolaratlas.info/?c=4.850154,73.146973,6&e=1 © 2018 Solargis page 12 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 3 GROUND MEASUREMENTS IN MALDIVES 3.1 Instruments and measured parameters Basic information about measurements sites is in Table 3.1. Solar parameters at all stations are measured by high (SR20, for measurements of GHI) and medium (RSP, for GHI, DNI and DIF) accuracy equipment (Table 3.2 and 3.3). The measurement campaign in Maldives has been performed by Suntrace GmbH company, Germany. Table 3.1 Overview information on measurement stations operated in the region Site name Latitude Longitude Altitude Station type ID (airport) Site ID [º] [º] [m a.s.l.] Installation date 1 Hanimaadhoo MVHAQ 6.7482° 73.1696° 2 TIER 2 10 Dec 2015 2 Hulhulé MVMLE 4.1927° 73.5281° 2 TIER 2 08 Dec 2015 3 Kadhdhoo MVKDO 1.8599° 73.5203° 2 TIER 2 14 Dec 2015 4 Gan MVGAN -0.6911° 73.1599° 2 TIER 2 13 Dec 2015 Table 3.2 Instruments installed at the solar meteorological stations Site name GHI 1 GHI 2 DNI DIF Temp RH WS AP Hanimaadhoo SR20 - T1 RSP 4G RSP 4G RSP 4G DKRF 400 DKRF 400 First Class - Hulhulé SR20 - T1 RSP 4G RSP 4G RSP 4G DKRF 400 DKRF 400 First Class - Kadhdhoo SR20 - T1 RSP 4G RSP 4G RSP 4G DKRF 400 DKRF 400 First Class - Gan SR20 - T1 RSP 4G RSP 4G RSP 4G DKRF 400 DKRF 400 First Class - Table 3.3 Technical parameters and accuracy class of the instruments Parameter Instrument Type Manufacturer Uncertainty GHI 1 Secondary standard pyranometer SR20 - T1 Hukseflux < ±2.0 % (daily) GHI 2 Rotating Shadowband Irradiometer RSP 4G Reichert GmbH Indicatively ±5 % DIF Rotating Shadowband Irradiometer RSP 4G Reichert GmbH Indicatively ±8 % DNI Rotating Shadowband Irradiometer RSP 4G Reichert GmbH Indicatively ±5 % TEMP Temperature probe DKRF 400 Driesen und Kern ± 1.5 ºC RH Relative humidity probe DKRF 400 Driesen und Kern ± 3.5 % RH WS Wind speed sensor (WS at 3.0 m height) First Class Thies Clima < ±1 % AP Barometric pressure sensor Not provided - - - Data logger Blueberry Wilmers GmbH - © 2018 Solargis page 13 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 3.2 Station operation and instruments’ field verification In this report, a complete set of data from of planned 2-year measurement campaign is analysed. As the measurement stations have been installed during December 2015, the regular envisaged period for the data analysis starts in January 2016 and ends in December 2017. However, the limit of 95% of high quality data availability within 2-years period was not fulfilled at three stations (Tab. 3.5). Therefore, the measurement campaign was extended for three to four months more (in Kadhdhoo and Gan until April 2018) to reach this criterion. Overview of the data availability (recovery rate), time step and measured parameters is shown in Table 3.4 to 3.6. Table 3.4 Overview information on solar meteorological stations operating in the region Site name(airport) Site ID Measurement period Primary time step Hanimaadhoo MVHAQ 11 Dec 2015 to 31 Mar 2018 1 minute Hulhulé MVMLE 09 Dec 2015 to 31 Mar 2018 1 minute Kadhdhoo MVKDO 15 Dec 2015 to 30 Apr 2018 1 minute Gan MVGAN 14 Dec 2015 to 30 Apr 2018 1 minute Table 3.5 shows data recovery statistics for 2-years period for each station. In this statistics, only serious technical issues (missing data for a longer period, erroneous data - initial installation problem or shading ring problem) are accounted. Short-term and other operational issues (surrounding shading, morning instrument dew, etc.) are not included. Data loss column represents amount of missing data or data excluded during quality control process due to technical issues. Percentage share is calculated from daytime values and days represent cumulative amount of missing data (one day may be composed from several shorter missing data periods). Influenced days column represents number of days with fully or partially missing data or days excluded by quality control process. It has to be noted that acceptance criteria were not achieved only for DNI and DIF measured by RSR4 instrument mainly due to shadow ring issue. The GHI measurements for all station fulfilled the requirements. Table 3.5 Data recovery statistics of the two-year measurement campaign period Maldives Data loss* Influenced days** Acceptance criteria measurement period Erroneous + missing data Total Length of individual periods DNI, DIF GHI 01/2016 - 12/2017 [%] [days] Description [days] (days) 95% 10+ days 95% 10+ days Hanimaadhoo 3.6 26 shading ring problem 92 24, 68 OK FAIL OK OK Hulhule 7.4 54 shading ring problem and missing data 156 2, 66, 88 FAIL FAIL OK OK Kadhdoo 10 73 shading ring problem 172 64, 108 FAIL FAIL OK OK Gan 15 110 shading ring problem (DNI, DIF) and missing data 249 14, 31, 35, 4, 91, 74 FAIL FAIL OK OK The measurement campaign continued for three to four months more, and measurements were used together with the initial period (December 2015) to re-evaluate the 2-years recovery rate (availability) criteria for DNI and DIF (Table 3.6). First, the data was quality controlled; the measurements affected by serious technical issues were excluded. The most influenced were the measurements from Gan station, where shading ring issue reduced amount of available DNI and DIF data from the extension period, the GHI data was not affected. The valid data from the extension period was added to the valid data from the 2-years period and the 2-years recovery statistics was recalculated for each station. The results (Table 3.6) show that prolonged measurement campaign helped to significantly exceed required 95% recovery rate at all four stations. © 2018 Solargis page 14 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 3.6 Data recovery statistics of the measurement campaign period including prolongation Two years period (2016-2017) Measurements extension (2015, 2018) Recovery rate (Availability) Acceptance Excluded* Missing to 95% threshold Length Excluded* Valid two years extension two years + extension criteria [%] [%] [days] [%] [days] [%] [%] [%] 95% Hanimaadhoo 3.6 1.4 111 0.0 111 96.4 15.2 111.6 OK Hulhule 7.4 -2.4 113 6.5 106 92.6 14.5 107.1 OK Kadhdoo 10.0 -5.0 136 1.2 134 90.0 18.4 108.4 OK Gan 15.0 -10.0 137 30.2 96 85.0 13.1 98.1 OK * Excluded due to technical issues Table 3.7 Period of measurements analysed in this report Year, month 2015 2016 2017 2018 Parameter 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Hanimaadhoo Hulhulé Kadhdhoo Gan During the measurement campaign, local staff of Maldives Meteorological Service, which was trained by Suntrace, participated on inspection and cleaning of instruments. The local partner Renewable Energy Maldives performed detailed visits and station maintenance after 6 months of meteorological stations operation. Field verification of instruments was performed by Suntrace GmbH after first year of operation. In particular, one day long comparative measurements of solar radiation parameters has been cross checked with the instruments from the spare parts to proof that sensitivities (calibration constants) remained stable within the instrument specifications,. According to the results of filed verification of the instruments, the calibration coefficients at Gan, Kadhdhoo and Hulhulé meteorological stations has been updated. Detailed results and discussion is supplied in Annual maintenance visit reports delivered separately for each meteorological station in February 2017. Table 3.8 Meteorological stations maintenance and instruments field verification Hanimaadhoo - MVHAQ Comments and issues Station type Tier 2 • Shading of the pyranometer by wind mast during some minutes in the afternoon in May 2016 – July 2016 and in Instruments cleaning interval Average: 1.0 June 2017. GHI values are replaced by GHI values from [days] Longest: 2 the RSP. • TEMP and RH sensor malfunction - on 14.06.2017 the Verification visits date 31 May 2016 stopped measuring RH and on 22.11.2017 also stopped 23 Nov 2016 measuring TEMP. The missing values of RH were 23 Jun 2017 replaced by measurements from the MMS 3 to 10 April 2018 meteorological measurement station (distance approx. 10 m) in period June 2017 – March 2018 and missing Instruments field verification GHI – reference SR 20 – T1 values of TEMP were replaced in period November 2017 DHI 2 – reference RSP 4G – March 2018. DIF – reference RSP 4G DNI – reference RSP 4G Hulhulé - MVMLE Comments and issues Station type Tier 2 • Incorrect measurements during noontime in period 15.05.2016 – 01.06.2016 until RSR shadowband Instruments cleaning interval Average: 3.9 adaptation [days] Longest: 31 • Shading of the pyranometer from wind mast during some minutes in the afternoon in May 2016 – July 2016 and Verification visits date 1 Jun 2016 June – August 2017. GHI values are replaced by GHI 22 Nov 2016 values from the RSP. 29 Jun 2017 • No station cleaning in July 2016 3 to 10 April 2018 © 2018 Solargis page 15 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Instruments field verification GHI – reference SR 20 – T1 • Malfunction of TEMP and RH sensor in 1 – 22 November DHI 2 – reference RSP 4G 2016. The missing values were replaced by 10-minute DIF – reference RSP 4G measurements from the MMS meteorological DNI – reference RSP 4G measurement station (distance approx. 2 m) interpolated to 1-minute resolution. • Replacement of RSR on 29.06.2017 by new calibrated unit, then incorrect measurements during noontime until shadowband adaptation (20.07.2017). Kadhdhoo - MVKDO Comments and issues Station type Tier 2 • The pyranometer is influenced by shading from the wind mast during some minutes in the afternoon in period Instruments cleaning interval Average: 1.1 June – July 2016 and June – July 2017. These values [days] Longest: 3 are replaced by GHI values from the RSI. • Malfunction of TEMP and RH sensor in February – April Verification visits date 30 May 2016 2018. The missing values were replaced measurements 27 Nov 2016 from the MMS meteorological station (distance approx. 24 Jun 2017 10 m). 3 to 10 April 2018 • solar radiation sensors with void calibration February 2018 – April 2018 Instruments field verification GHI – reference SR 20 – T1 DHI 2 – reference RSP 4G DIF – reference RSP 4G DNI – reference RSP 4G Gan - MVGAN Comments and issues Station type Tier 2 • Incorrect measurements during noontime in period 14.12.2015 – 13.01.2016, then in May 2016, June 2017 Instruments cleaning interval Average: 1.2 – July 2017 and November 2017 – January 2018, due to [days] Longest: 5 sun position. • Shading of the pyranometer from wind mast in Verification visits date 28 May 2016 December 2015 – March 2016, then in October 2016 – 25 Nov 2016 December 2016, March 2017 and November 2017 – 25 June 2017 March 2018. GHI values are replaced by GHI values from 3 to 10 April 2018 the RSP. • Malfunction of TEMP and RH sensor in 01.09.2016 – Instruments field verification GHI – reference SR 20 – T1 13.10.2016 and 23.04.2017 - 27.04.2017. The missing DHI 2 – reference RSP 4G values were replaced by 10-minute measurements from DIF – reference RSP 4G the MMS meteorological measurement station (distance DNI – reference RSP 4G approx. 10 m) interpolated to 1-minute resolution. • RSR measurements outage due to low battery voltage in period from 16.12.2016 to 19.12.2016 • February 2018 – April 2018 solar radiation sensors with void calibration At the end of measurement campaign, the re-calibration of instruments and handing-over of measurement stations to local authority is planned. These tasks will be performed by Suntrace, procedures are described in detail in Chapter 3.5. © 2018 Solargis page 16 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 3.3 Quality control of measured solar resource data Prior to correlation with satellite-based solar data, the ground-measured solar radiation was quality-controlled by Solargis. Quality control (QC) was based on methods defined in SERI QC procedures and Younes et al. [1, 2] and the in-house developed tests. The ground measurements were inspected also visually, mainly for identification of shading and other error patterns such as RSP shading ring malfunction. Figures 3.1 to 3.12 show results of quality control for individual stations. The colours in Figures 3.1, 3.3, 3.6 and 3.9 indicate the following flags: • Blue: data excluded by visual inspection - mainly shading, shading ring issues • Green: data passed all tests • Grey: sun below horizon • White and brown strips: missing data • Red: GHI, DNI and DIF consistency problem • Violet and cyan: physical limit problem. The data records not passing the quality control test were flagged and excluded from the further processing (Tables 3.10, 3.13, 3.16 and 3.19). The results show relatively small amount of excluded GHI data readings, mainly due to local shading occurring in the early morning and late afternoon. The amount of excluded DNI data samples is considerably higher, mainly due to shadowband issues and shading. 3.3.1 Hanimaadhoo Airport Table 3.9 Occurrence of data readings for Hanimaadhoo meteorological station Data availability GHI SR20 GHI, DNI RSP Sun below horizon 602,004 49.7% 602,004 49.7% Sun above horizon 610,415 50.3% 610,415 50.3% Total data readings 1,212,419 100.0% 1,212,419 100.0% Table 3.10 Excluded ground measurements after quality control (Sun above horizon) in Hanimaadhoo Occurrence of data samples (Sun above horizon) Type of test GHI SR20 DNI RSP GHI RSP Physical limits test 431 0.1% 0 0.0% 2,631 0.4% Consistency test (GHI – DNI – DIF) - - 149 0.0% 149 0.0% Visual test (incorrect data) 43,072 7.1% 62,451 10.2% 41,536 6.8% Other (non valid data) 98 0.0% 166 0.0% 166 0.0% Total excluded data samples 43,601 7.1% 62,766 10.3% 44,482 7.3% Total samples 610,415 100.0% 610,415 100.0% 610,415 100.0% © 2018 Solargis page 17 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 © 2018 Solargis page 18 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 3.1 Results of DNI and GHI quality control in Hanimaadhoo. Green – data passing all tests; grey – sun below horizon; red – consistency issue, violet and cyan – physical limit, blue excluded by visual inspection. Top: DNI (RSP); middle: GHI (RSP); bottom: GHI (SR20) Main findings: • Shadowband malfunction. Due to location of measurement station close to the Equator the rotating shadowband did not in some periods sufficiently shade the sensor. This insufficient shading results in incorrect DNI and DIF measurements. This problem was identified for period from 3 to 26 June 2016 and 8 May to 14 July 2017 around the noon. Approximately 3.1% of DNI and DIF measurements were excluded. • A systematic difference between GHI measurements from secondary standard pyranometer SR20-T1 and RSP (Figure 3.4). GHI from SR20 is in average higher by 5.3% than GHI from RSP. In the noon time, it can exceed 7%. The overall difference in 2016 was 4.5% while in 2017 it was 5.7%. • Late afternoon shading from surrounding objects. Since February 2017 also early morning shading. • GHI measurements from SR20 were replaced by data from RSP in case of shading from wind mast (see Table 3.8). © 2018 Solargis page 19 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 3.2 Systematic difference between GHI from SR20 and RSP − Hanimaadhoo. Table 3.11 Quality control summary – Hanimaadhoo Indicator Flag Description Station description, metadata Installation report available Instrument accuracy 1x Secondary standard pyranometer SR20-T1 (GHI) 1x Rotating Shadowband Radiometer RSP 4G(GHI, DIF, DNI) Instrument calibration Instruments were calibrated Data structure Clear Cleaning and maintenance Cleaning log available information Diligent cleaning Time reference Correct and clear time reference Quality control complexity RSP G4 data, full QC comparison of GHI from RSP and SR20 Quality control results Except for period of ca 3 months (noontime), dues to shadowband malfunction Early morning and late afternoon shading Inconsistency between of RSP and SR20 measurements – increasing in time Availability of high quality data More than 27 months, DNI ca 26 months due to shadowband malfunction Other issues Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 20 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 3.3.2 Hulhulé Airport Table 3.12 Occurrence of data readings for Hulhulé meteorological station Data availability GHI SR20 GHI, DNI RSP Sun below horizon 603,061 49.6% 603,061 49.6% Sun above horizon 612,238 50.4% 612,238 50.4% Total data readings 1,215,299 100.0% 1,215,299 100.0% Table 3.13 Excluded ground measurements after quality control (Sun above horizon) in Hulhulé Occurrence of data samples (Sun above horizon) Type of test GHI SR20 DNI RSP GHI RSP Physical limits test 498 0.1% 3 0.0% 2,540 0.4% Consistency test (GHI – DNI – DIF) - - 576 0.1% 576 0.1% Visual test (incorrect data) 29,425 4.8% 62,727 10.2% 33,313 5.4% Other (non valid data) 101 0.0% 11,356 1.9% 910 0.1% Total excluded data samples 30,024 4.9% 74,662 12.2% 37,339 6.1% Total samples 612,238 100.0% 612,238 100.0% 612,238 100.0% © 2018 Solargis page 21 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 3.3 Results of DNI and GHI quality control in Hulhulé. Green – data passing all tests; grey – sun below horizon; red – consistency issue, cyan and violet – physical limit issue, blue excluded by visual inspection; brown – missing data. Top: DNI (RSP); middle: GHI (RSP); bottom: GHI (SR20) © 2018 Solargis page 22 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Main findings: • Shadowband malfunction (Figure 3.4). This issue was identified for a period from 15 May to 19 July 2016 and from 29 April to 25 July 2017 around the noon. Affected data readings in May 2016 and July 2017 were excluded by data provider, remaining affected readings were excluded by quality control. Approximately 5.6% of DNI and DIF measurements were excluded in total. • A systematic difference between GHI measurements from secondary standard pyranometer SR20-T1 and RSP (Figure 3.5). GHI from SR20 is in average higher by 7.4% than GHI from RSP. In the noontime the difference can exceed 10%. The overall difference in 2016 was 6.8% while in 2017 it was 8.1%. • Late afternoon shading from surrounding objects • GHI measurements from SR20 were replaced by data from RSP in case of shading from wind mast (see Table 3.8). • Malfunction of RSP instrument (GHI, DNI, DIF) in the end of March 2018. Figure 3.4 Effect of RSP shadowband issues – drop of DNI in Hulhulé. Blue: DNI RSP; yellow: GHI SR20; red: GHI RSP; green: DIF RSP; dashed: theoretical clear-sky profile Figure 3.5 Systematic difference between GHI from SR20 and RSP - Hulhulé. © 2018 Solargis page 23 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 3.14 Quality control summary – Hulhulé Indicator Flag Description Station description, metadata Installation report available Instrument accuracy 1x Secondary standard pyranometer SR20-T1 (GHI) 1x Rotating Shadowband Radiometer RSP 4G (GHI, DIF, DNI) Instrument calibration Instruments were calibrated Data structure Clear Cleaning and maintenance Cleaning log available information No cleaning in July 2016 Time reference Correct and clear time reference Quality control complexity RSP G4 data, full QC comparison of GHI from RSP and SR20 Quality control results Shadowband malfunction (ca 5 months noontime) Inconsistency between of RSP and SR20 measurements – increasing in time Early morning and late afternoon shading Malfunction of RSP in March 2018 Availability of high-quality data More than 27 months, DNI ca 25 months due to shadowband malfunction Other issues Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 24 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 3.3.3 Kadhdhoo Airport Table 3.15 Occurrence of data readings for Kadhdhoo meteorological station Data availability GHI SR20 GHI, DNI RSP Sun below horizon 619,617 49.6% 619,617 49.6% Sun above horizon 630,242 50.4% 630,242 50.4% Total data readings 1,249,859 100.0% 1,249,859 100.0% Table 3.16 Excluded ground measurements after quality control (Sun above horizon) in Kadhdhoo Occurrence of data samples (Sun above horizon) Type of test GHI SR20 DNI RSP GHI RSP Physical limits test 803 0.1% 0 0.0% 2,260 0.4% Consistency test (GHI – DNI – DIF) - - 190 0.0% 190 0.0% Visual test (incorrect data) 55,844 8.9% 110,959 17.6% 54,448 8.6% Other (non valid data) 48 0.0% 102 0.0% 101 0.0% Total excluded data samples 56,695 9.0% 111,251 17.7% 56,999 9.0% Total samples 630,242 100.0% 630,242 100.0% 630,242 100.0% © 2018 Solargis page 25 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 3.6 Results of DNI and GHI quality control − Kadhdhoo. Green – data passing all tests; grey – sun below horizon; red – consistency issue, violet – physical limit issue, blue excluded by visual inspection. Top: DNI (RSP); middle: GHI (RSP); bottom: GHI (SR20) © 2018 Solargis page 26 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Main findings: • Shadowband malfunction (Figure 3.7). This issue was identified for a period from 25 May to 27 July 2016, from 2 May to 17 August 2017 and 25 April to 30 April 2018 around the noon. Affected data readings were excluded by quality control. Approximately 9.0% of DNI and DIF measurements were excluded in total. • A systematic difference between GHI measurements from secondary standard pyranometer SR20-T1 and RSP (Figure 3.8). GHI from SR20 is in average higher by 4.9% than GHI from RSP. In the noon time it can exceed 7%. The overall difference in 2016 was 4.5% while in 2017 it was 5.0%. • Late afternoon and early morning shading from surrounding objects • GHI measurements from SR20 were replaced by data from RSP in case of shading from wind mast (see Table 3.8). Figure 3.7 Effect of RSP shadowband issues – drop of DNI in Kadhdhoo. Blue: DNI RSP; yellow: GHI SR20; red: GHI RSP; green: DIF RSP; dashed: theoretical clear-sky profile Figure 3.8 Systematic difference between GHI from SR20 and RSP − Kadhdhoo. © 2018 Solargis page 27 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 3.17 Quality control summary – Kadhdhoo Indicator Flag Description Station description, metadata Installation report available Instrument accuracy 1x Secondary standard pyranometer SR20-T1 (GHI) 1x Rotating Shadowband Radiometer RSP 4G (GHI, DIF, DNI) Instrument calibration Instruments were calibrated up to January 2018, then 2 months were measured with void calibration Data structure Clear Cleaning and maintenance Cleaning log available information Diligent cleaning Time reference Correct and clear time reference Quality control complexity RSP G4 data, full QC comparison of GHI from RSP and SR20 Quality control results Shadowband malfunction (ca 5.5 months noontime) Early morning and late afternoon shading Inconsistency between of RSP and SR20 measurements – increasing in time Availability of high-quality data More than 28 months, DNI ca 25 months due to shadowband malfunction Other issues Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 28 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 3.3.4 Gan Airport Table 3.18 Occurrence of data readings for Gan meteorological station Data availability GHI SR20 GHI, DNI RSP Sun below horizon 619 976 49.5% 619 976 49.5% Sun above horizon 631 323 50.5% 631 323 50.5% Total data readings 1 251 299 100.0% 1 251 299 100.0% Table 3.19 Excluded ground measurements after quality control (Sun above horizon) in Gan Occurrence of data samples (Sun above horizon) Type of test GHI SR20 DNI RSP GHI RSP Physical limits test 753 0.1% 0 0.0% 2 387 0.4% Consistency test (GHI – DNI – DIF) - - 234 0.0% 234 0.0% Visual test (incorrect data) 0 0.0% 47 487 7.5% 16 0.0% Other (non valid data) 98 0.0% 63 418 10.0% 328 0.1% Total excluded data samples 851 0.1% 111 139 17.6% 2 965 0.5% Total samples 631 323 100.0% 631 323 100.0% 631 323 100.0% © 2018 Solargis page 29 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 3.9 Results of DNI and GHI quality control − Gan. Green – data passing all tests; grey – sun below horizon; red – consistency issue, violet – physical limit issue, blue excluded by visual inspection; brown – missing data. Top: DNI (RSP); middle: GHI (RSP); bottom: GHI (SR20) © 2018 Solargis page 30 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Main findings: • Shadowband malfunction (Figure 3.10). This issue was identified for several periods (from 14 December 2015 to 12 January 2016; from 1 May to 31 May 2016, from 22 October to 25 November 2016, from 17 April to 17 July 2017, from 18 October 2017 to 6 March 2018) around the noon. Affected data readings were excluded partly by data provider and partly by quality control. Approximately 17.5% of DNI and DIF measurements were excluded in total. • A systematic difference between GHI measurements from secondary standard pyranometer SR20-T1 and RSP (Figure 3.11). GHI from SR20 is in average higher by 5.3% than GHI from RSP. In the noon time it can exceed 7%. The overall difference in 2016 was 4.7% while in 2017 it was 5.8%. • GHI measurements from SR20 were replaced by data from RSP in case of shading from wind mast (Figure 3.12). Figure 3.10 Effect of RSP shadowband issues – drop of DNI in Gan. Blue: DNI RSP; yellow: GHI SR20; red: GHI RSP; green: DIF RSP; dashed: theoretical clear-sky profile Figure 3.11 Systematic difference between GHI from SR20 and RSP - Gan. © 2018 Solargis page 31 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 3.12 Replacement of SR20 GHI by RSP GHI due to wind mast shading in Gan. Blue: DNI RSP; red: GHI RSP yellow: GHI SR20; green: DIF RSP; dashed: theoretical clear-sky profile Table 3.20 Quality control summary – Gan Indicator Flag Description Station description, metadata Installation report available Instrument accuracy 1x Secondary standard pyranometer SR20-T1 (GHI) 1x Rotating Shadowband Radiometer RSP 4G (GHI, DIF, DNI) Instrument calibration Instruments were calibrated up to January 2018, then 2 months measured with void calibration Data structure Clear Cleaning and maintenance Cleaning log available information Diligent cleaning Time reference Correct and clear time reference Quality control complexity RSP G4 data, full QC comparison of GHI from RSP and SR20 Quality control results Shadowband malfunction (ca 10.5 months noontime) Inconsistency between of RSP and SR20 measurements – increasing in time Availability of high-quality data More than 28.5 months for GHI, more than 23.5 months DNI and DIF (reduction due to shadowband malfunction) Other issues Legend: Quality flag Very good Good Sufficient Problematic Insufficient Not specified © 2018 Solargis page 32 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 3.4 Recommendations on the operation and maintenance of the sites Based on the results of quality control (Table 3.11, 3.14, 3.17 and 3.20), we conclude that the solar radiation measurements come from the high (SR20) and medium (RSP) accuracy equipment that is professionally operated and maintained. Some issues were identified during the data quality control of the whole period of ground measurement campaign: • Incorrect DNI and DIF measurements due to shadowband malfunction. These data were flagged and excluded from further processing. Most influenced are measurements from Gan station, where ca 17.5% of DNI and DIF data was excluded due to this issue for full period of measurements (15% for 2-years period). • Higher systematic difference between GHI from SR20 and RSP (up to 7.4% in Hulhulé). The difference between the instruments increasing within the second year at all stations (6.8% for 2016 and 8.1% for 2017 in Hulhulé). This is known effect that should relate mainly to GHI instrument calibration. Due to this issue, the GHI from RSP was not considered for site-adaptation. The effect of this issue on the measured DNI is unknown as the data from higher accuracy reference instrument was not available. It is expected that the effect will not be extremely high, because both initial measurements (GHI and DIF) used for the calculation of the DNI are affected in the same way and should at the end in some extent eliminate each other. Unfortunately there is no direct proof for this behaviour. Therefore, a higher uncertainty of DNI measurements should be considered. For future operation and maintenance of the measuring stations by local provider we recommend: • Adopt measures to avoid the shadowband problems, where possible. • Identify and fix source of systematic difference between SR20 and RSP 3.5 Instruments re-calibration at the end of measurement campaign Methodology At the end of 2-years measurement campaign all solar sensors should be re-calibrated upon the hand-over to a local organization. Suntrace originally proposed a “round robin" re -calibration strategy using the one set of spare sensors: one thermopile pyranometer and one RSI, which both still hold valid calibration certificates as they have not been exposed to sunlight over the 2 years of operations and thus, according to the manufacturer, can be still regarded as newly calibrated. After exchange at the first site (MVMLE – Male International Airport), the 2 used sensors were returned to Europe for re-calibration. The re-calibrated sensors should have been then installed at site no. 2. This procedure should have been repeated at all other stations, until the sensors of all 4 stations were re-calibrated. Reasons for improved re-calibration: • When proposing the re-calibration plan for the ESMAP measurement campaign in Maldives in 2014, Suntrace assumed that following the findings of [24] a duration of 4 weeks for re-calibrating the RSI instruments provides sufficient quality. In the course of the project DLR published a paper [25], which comes to the conclusion that the long term-outdoor calibration at PSA should last at least 60 days and during winter 90 days are needed to reach good results. This causes that the calibration of the RSI instruments at PSA takes about 2 to 3 times longer than originally planned. • Latest findings on the quality of remote calibration of RSI using PSA as the reference station [26] indicate relatively large differences for sites in more humid regions. An RSI instrument calibrated at PSA and operated in India shows significant differences compared to the reference station operated at the same site. These systematic differences are likely due to much higher water vapor at the site in India compared to PSA site, which is located in an arid climate. As all 4 sites in Maldives are in humid tropical climate, we have to assume that the remote calibration at PSA may be also leading to significant deviations there. The “round robin” re-calibration actually started in July 2017 with exchanging the RSI and SR20 sensor with the spare parts at MVMLE. The dismantled RSI was then sent for re-calibration to Spain (PSA calibration) and the SR20 to the Netherlands. The round-trip transport time turned out to be 6 weeks longer than planned due to customs organization etc. © 2018 Solargis page 33 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 The duration for shipping and re-calibration of these sensors turns out to be far too lengthy - one swap was taking 3 months instead of only 6 weeks as originally planned due to administration required from different parties. Therefore, and for the increase of accuracy of the RSI calibration, Suntrace is proposing an improved re-calibration method. This proposed new time schedule of improved re-calibration is shown in Table 3.21. This time schedule may change, depending on the decisions taken. Summary of proposed re-calibration and check of solar radiation sensors (Table 3.22): • Thermopile pyranometers: all 5 pyranometers will be exchanged by a new or refurbished/re-calibrated pyranometers of same type or better. • RSI: Instead of shipping to Europe and performing a long-term calibration against a reference station (3 months in summer or 6 months in winter) re-calibrated thermopile pyranometers at each site will be used and applied an in-situ calibration. This calibration (improved calibration based on [27]) has the advantage that it is more site- and time-specific, as it considers the actual aerosol and water vapor conditions at the site, whereas the long-term calibration against a reference station would be optimized to the local climate of that location. Especially in tropical regions this can lead to significant offsets. The advanced calibration method takes the local weather conditions of the last weeks and the measurements of the precise thermopile pyranometer into account to calibrate the RSI. In order to achieve this, all 5 data loggers will be additionally exchanged with new data loggers including the in-situ calibration algorithm. • Silicon photodiode: Check of proper horizontal levelling within specs, comparison with thermopile pyranometer during the past 2 years and after replacement against a newly calibrated one. Calculation of degradation and in case of significant sensitivity changes derivation and implementation of a new calibration constant. Re-calibration and check of auxiliary meteorological sensors: • Temperature/humidity probe: Exchange of the sensor head against a new one. • Radiation shield: Cleaning of the surface to guarantee continued low albedo for minimizing errors due to heating up under strong sunlight. • Barometric pressure: Comparison of the barometric pressure during the site visit with another barometer of equal of better quality. • Wind speed: Anemometer function will be checked during site visit. • Wind direction: Wind vane readings will be verified by compass bearings during site visit. Table 3.21 Updated re-calibration schedule proposed by Suntrace Period Main tasks 04.06. – 10.06.2018 Receive python prototype algorithm 11.06. – 15.07.2018 Validation and documentation of algorithm Install test station on roof of university building in Hamburg 16.07. – 29.07.2018 Implement C-code into data logger of clone station 30.07. – 26.08.2018 Test run 27.08. – 09.09.2018 Verify proper operation in data logger 10.09. – 23.09.2018 Shipping to Maldives 24.09. – 07.10.2018 Installations on all 4 Islands in Maldives 08.10. – 14.10.2018 Hand-over of stations to MMS and final workshop © 2018 Solargis page 34 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 3.22 Overview of instruments to be re-calibrated and/or exchanged with new/refurbished instruments No. Site Code Instrument Manufacturer Model Serial No. Requires re- To be calibration exchanged 1 MVMLE Thermo-hygro Driesen und Kern DKRF 400 K04496 Yes Yes sensor 2 MVMLE Data Logger Wilmers Blueberry 1125 No Yes Messtechnik GmbH COMPACT 3 MVHAQ RSI Reichert GmbH RSP 4G 14-11 Yes No 4 MVHAQ Pyranometer Hukseflux SR20-T1 4485 Yes Yes 5 MVHAQ Thermo-hygro Driesen und Kern DKRF 400 K04495 Yes Yes sensor 6 MVHAQ Data Logger Wilmers Blueberry 1124 No Yes Messtechnik GmbH COMPACT 7 MVKDO RSI Reichert GmbH RSP 4G 14-07 Yes No 8 MVKDO Pyranometer Hukseflux SR20-T1 4383 Yes Yes 9 MVKDO Thermo-hygro Driesen und Kern DKRF 400 K04498 Yes Yes sensor 10 MVKDO Data Logger Wilmers Blueberry 1122 No Yes Messtechnik GmbH COMPACT 11 MVGAN RSI Reichert GmbH RSP 4G 14-03 Yes No 12 MVGAN Pyranometer Hukseflux SR20-T1 4384 Yes Yes 13 MVGAN Thermo-hygro Driesen und Kern DKRF 400 K04497 Yes Yes sensor 14 MVGAN Data Logger Wilmers Blueberry 1121 No Yes Messtechnik GmbH COMPACT Comments: Additionally, one re-calibrated RSI, pyranometer, thermo-hygro sensor and one new data logger will be provided as spare parts. © 2018 Solargis page 35 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 4 SOLAR RESOURCE MODEL DATA 4.1 Solar model Solar radiation is calculated by Solargis model, which is parameterized by a set of inputs characterizing the cloud transmittance, state of the atmosphere and terrain conditions. A comprehensive overview of the Solargis model is made available in the recent book publication [3]. The methodology is also described in [4, 5]. The related uncertainty and requirements for bankability are discussed in [6, 7]. In Solargis approach, the clear-sky irradiance is calculated by the simplified SOLIS model [8]. This model allows fast calculation of clear-sky irradiance from the set of input parameters. Sun position is deterministic parameter, and it is described by the algorithms with satisfactory accuracy. Stochastic variability of clear-sky atmospheric conditions is determined by changing concentrations of atmospheric constituents, namely aerosols, water vapour and ozone. Global atmospheric data, representing these constituents, are routinely calculated by world atmospheric data centres: • In Solargis, the new generation aerosol data set representing Atmospheric Optical Depth (AOD) is used. The calculation accuracy is strongly determined by quality of aerosols, especially for cloudless conditions. The aerosol data implemented by MACC-II/CAMS and MERRA-2 projects are used [9, 10]. • Water vapour is also highly variable in space and time, but it has lower impact on the values of solar radiation, compared to aerosols. The GFS and CFSR databases (NOAA NCEP) are used in Solargis, and the data represent the daily variability from 1999 to the present time [11, 12, 13]. • Ozone absorbs solar radiation at wavelengths shorter than 0.3 µm, thus having negligible influence on the broadband solar radiation. The clouds are the most influencing factor, modulating clear-sky irradiance. Effect of clouds is calculated from the satellite data in the form of a cloud index (cloud transmittance). The cloud index is derived by relating radiance recorded by the satellite in spectral channels and surface albedo to the cloud optical properties. In Solargis, the modified calculation scheme of Cano has been adopted to retrieve cloud optical properties from the satellite data [14]. To calculate all-sky irradiance in each time step, the clear-sky global horizontal irradiance is coupled with cloud index. Direct Normal Irradiance (DNI) is calculated from Global Horizontal Irradiance (GHI) using modified Dirindex model [15]. Diffuse irradiance for tilted surfaces, which is calculated by Perez model [16]. The calculation procedure includes also terrain disaggregation, the spatial resolution is enhanced with use of the digital terrain model to 250 meters [17]. Solargis model version 2.1 has been used. Table 4.1 summarizes technical parameters of the model inputs and of the primary data outputs. Table 4.1 Input data used in the Solargis and related GHI and DNI outputs for Maldives Inputs into the Solargis model Source Time Original Approx. grid of input data representation time step resolution Cloud index Meteosat MFG IODC 1999 to 2016 30 minutes 2.8 x 3.2 km Meteosat MSG IODC 2017 to date 15 minutes 3.1 x 3.5 km satellites (EUMETSAT) Atmospheric optical depth MACC-II/CAMS* 2003 to date 3 hours 75 km and 125 km (aerosols)* (ECMWF) MERRA-2 (NASA) 1999 to 2002 1 hour 50 km Water vapour CFSR/GFS 1999 to date 1 hour 35 and 55 km (NOAA) Elevation and horizon SRTM-3 - - 250 m (SRTM) Solargis primary data outputs - 1999 to date 30 minutes 250 m (GHI and DNI) * Aerosol data for 2003-2012 come from the reanalysis database; the data representing years 2013-present are derived from near- real time operational model © 2018 Solargis page 36 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 4.1: Sites with solar meteorological stations used for site adaptation of Solargis model in Maldives 4.2 Site adaptation of the solar model − method This chapter describes accuracy improvement of the delivered model time series for four sites. This improvement has been achieved by site adaptation of the model, based on the use of 2+ years of local measurements. The fundamental difference between a satellite observation and a ground measurement is that signal received by the satellite radiometer integrates a large area, while a ground station represents a pinpoint measurement. This results in a mismatch when comparing instantaneous values from these two observation instruments, mainly during intermittent cloudy weather and changing aerosol load. Nearly half of the hourly Root Mean Square Deviation (RMSD) for GHI and DNI can be attributed to this mismatch (value at sub-pixel scale), which is also known as the “nugget effect” [18]. The satellite pixel is not capable describing the inter-pixel variability in complex regions, where within one pixel diverse natural conditions mix-up (e.g. fog in narrow valleys or along the coast). In addition, the coarse spatial resolution of atmospheric databases such as aerosols or water vapour is not capable to describe local patterns of the state of atmosphere. These features can be seen in the satellite DNI data by increased bias due to imperfect description of aerosol load. Satellite data have inherent inaccuracies, which have certain degree of geographical and time variability. © 2018 Solargis page 37 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Especially DNI is strongly sensitive to variability of cloud information, aerosols, water vapour, and terrain shading. The relation between uncertainty of global and direct irradiance is nonlinear. Often, a negligible error in global irradiance may have high counterpart in the direct irradiance component. The solar energy projects require representative and accurate GHI and DNI time series. The satellite-derived databases are used to describe long-term solar resource for a specific site. However, their problem when compared to the high-quality ground measurements is a slightly higher bias and partial disagreement of frequency distribution functions, which may limit their potential to record the occurrence of extreme situations (e.g. very low atmospheric turbidity resulting in a high DNI and GHI). A solution is to correlate satellite-derived data with ground measurements to understand the source of discrepancy and subsequently to improve the accuracy of the resulting time series. The Solargis satellite-derived data are correlated with ground measurement data with two objectives: • Improvement of the overall bias (removal of systematic deviations) • Improvement of the fit of the frequency distribution of values. Limited spatial and temporal resolution of the input data and simplified nature of the models results in the occurrence of systematic and random deviations of the model outputs when compared to the ground observations. The deviations in the satellite-computed data, which have systematic nature, can be reduced by site adaptation or regional adaptation methods. In this report we apply site-adaptation of the solar model, i.e. adapting the model for the local conditions represented by on-site measurements. The terminology related to the procedure improving accuracy of the satellite data is not harmonized, and various terms are used: • Correlation of ground measurements and satellite-based data; • Calibration of the satellite model (its inputs and parameters); • Site adaptation or regional adaptation of satellite based data. The term site adaptation is more general and best explains the concept of adapting the satellite-based model (by correlation, calibration, fitting and recalculation) to the ground measured data. Site adaptation aims to adapt the characteristics of the satellite-based time series to the site-specific conditions described by local measurements. Three conditions are important for successful adaptation of the satellite-based model: 1. High quality DNI and GHI ground measurements for at least 12 months must be available; optimally data for 2 or 3 years should be used; 2. High quality satellite data must be used, with consistent quality over the whole period of data; 3. There must be identified a systematic difference between both data sources. Systematic difference can be measured by two characteristics: • Bias (offset) • Systematic deviation in the distribution of hourly or daily values (in the histogram) Systematic difference can be stable over the year or it can slightly change seasonally for certain meteorological conditions (e.g. typical cloud formation during a day, seasonal air pollution). The data analysis should distinguish systematic differences from those arising at occasional events, such as extreme sand storms or forest fires. The episodically-occurring differences may mislead the results of adaptation, especially if short period of ground measurements is only available. If one of the three above-mentioned conditions is not fulfilled, site adaptation (regional adaptation) will not provide the expected results. On the contrary, such an attempt may provide worse results. For the quantitative assessment of the accuracy enhancement procedures, the following metrics is used: • Metrics based on the comparison of all pairs of the hourly daytime data values: Mean Bias, and Root Mean Square Deviation (RMSD), histogram, in an absolute and relative form (divided by the daytime mean DNI values); • Metrics based on the difference of the cumulative distribution functions: KSI (Kolmogorov-Smirnov test Integral) [19] The normalized KSI is defined as an integral of absolute differences of two cumulative distribution functions D normalized by the integral of critical value acritical: © 2018 Solargis page 38 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 , where critical value depends on the number of the data pairs N. As the KSI value is dependent on the size of the sample, the KSI measure may be used only for the relative comparison of fit of cumulative distribution of irradiance values. For the accuracy enhancement of solar resource parameters in this study, a combination of two methods was used. First, systematic deviations due to influence of aerosols were partially removed. Afterwards, to improve the distribution of values, the fitting of cumulative frequency distribution curves of ground measurements and satellite data was used. The site-adaptation procedure first identifies the sources of discrepancies by comparing the ground-measured data with Solargis model data, for the period of the overlap between both data sets. Based on this analysis, correction coefficients to improve the fit between the measured and the model Solargis data are developed. In the second step, these coefficients are used for the adaptation of the full (19 years) time series. The satellite data is available in 15 and 30-minute time step and the ground measurements in 1-minute time step. To partially remove the conceptual difference of point and satellite pixel measurements, prior to site adaptation, all the measures are calculated using aggregated data in the hourly time step. The adaptation was based on measured DNI data from RSP 4G instrument and GHI data from the secondary standard SR20-T1 instrument. GHI measured by the RSP 4G was not used because of higher uncertainty of the measurements (Chapter 3.3). 4.3 Results of the model adaptation at four sites The original Solargis data show a regional pattern of overestimation, compared to the ground measurements – mainly for DNI. The model adaptation allowed removing large part of mismatch between satellite-based data and ground measurements. The GHI fits the ground measurements very well. Tables 4.2 to 4.5 summarize results of the site-adaptation for all solar measuring stations. Table 4.2 Direct Normal Irradiance: bias and KSI before and after model site-adaptation Meteo station Original DNI data DNI after site adaptation Bias Bias KSI Bias Bias KSI [kWh/m2] [%] [-] [kWh/m2] [%] [-] Hanimaadhoo 19 5.3 104 0 0.0 56 Hulhulé 31 8.4 149 0 0.0 82 Kadhdhoo 29 7.5 148 0 0.0 94 Gan 33 8.4 159 0 0.0 110 Mean 28.0 7.4 140 0.0 0.0 85.5 Standard deviation 6.2 1.5 0.0 0.0 © 2018 Solargis page 39 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 4.3 Global Horizontal Irradiance: bias and KSI before and after model site-adaptation Meteo station Original GHI data GHI after site adaptation Bias Bias KSI Bias Bias KSI [kWh/m2] [%] [-] [kWh/m2] [%] [-] Hanimaadhoo 3 0.7 49 0 0.0 38 Hulhulé 0 0.0 50 0 0.0 42 Kadhdhoo 3 0.7 52 0 0.0 45 Gan 7 1.4 67 0 0.0 44 Mean 3.3 0.7 55 0.0 0.0 42 Standard deviation 2.9 0.5 0.0 0.0 Table 4.4 Direct Normal Irradiance: RMSD before and after model site-adaptation Meteo station RMSD of original DNI data RMSD of DNI after site adaptation Hourly Daily Monthly Hourly Daily Monthly [%] [%] [%] [%] [%] [%] Hanimaadhoo 31.8 17.7 6.6 31.6 17.1 4.3 Hulhulé 35.2 20.0 9.1 34.2 18.6 3.5 Kadhdhoo 35.6 19.3 8.1 34.9 18.3 2.9 Gan 35.8 20.1 9.3 35.0 18.7 4.3 Mean 34.6 19.3 8.3 33.9 18.2 3.7 Table 4.5 Global Horizontal Irradiance: RMSD before and after model site-adaptation Meteo station RMSD of original GHI data RMSD of GHI after site adaptation Hourly Daily Monthly Hourly Daily Monthly [%] [%] [%] [%] [%] [%] Hanimaadhoo 15.3 6.6 2.3 15.3 6.6 2.2 Hulhulé 16.4 7.2 1.8 16.5 7.2 1.8 Kadhdhoo 16.7 7.1 1.3 16.7 7.1 1.2 Gan 16.2 7.2 2.2 16.1 7.0 1.7 Mean 16.2 7.0 1.9 16.1 7.0 1.7 As a result, at the level of individual measurement sites in Maldives, the mean bias of the adapted values was reduced to zero. The values of RMSD and KSI accuracy parameters are also reduced, especially for DNI. The effect of the site adaptation is presented in a detail for all sites (Figures 4.2 to 4.5). The changes of GHI are very small, as the original data have good fit to ground measurements. The changes of DNI are more significant. © 2018 Solargis page 40 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Hanimaadhoo: Original DNI Hanimaadhoo: DNI after adaptation Hanimaadhoo: Original GHI Hanimaadhoo: GHI after adaptation Figure 4.2: Correction of DNI and GHI hourly values for Hanimaadhoo. Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. © 2018 Solargis page 41 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Hulhulé: Original DNI Hulhulé: DNI after adaptation Hulhulé: Original GHI Hulhulé: GHI after adaptation Figure 4.3: Correction of DNI and GHI hourly values for Hulhulé Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. © 2018 Solargis page 42 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Kadhdhoo: Original DNI Kadhdhoo: DNI after adaptation Kadhdhoo: Original GHI Kadhdhoo: GHI after adaptation Figure 4.4: Correction of DNI and GHI hourly values for Kadhdhoo. Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. © 2018 Solargis page 43 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Gan: Original DNI Gan: DNI after adaptation Gan: Original GHI Gan: GHI after adaptation Figure 4.5: Correction of DNI and GHI hourly values for Gan. Left: original Solargis data, right: site-adapted Solargis data. The X-axis represents the measured data and the Y-axis represents the satellite-derived data. The change of model GHI and DNI after adaptation is presented on an example of Hanimaadhoo (Figure 4.6). The change for GHI is negligible, the adapted DNI is lower than original one. The other sites are very similar (Table 4.6). The site-adapted model values better represent the geographical characteristics of DNI and GHI solar resource and they also improve the distribution and match of hourly values. The measurements show that the model performs well in the region, and these results significantly improve the confidence about the reliability of the measured and modelled solar resource data in Maldives. © 2018 Solargis page 44 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 4.6: Comparison of Solargis original and site-adapted data for Hanimaadhoo site. Left: DNI; Right: GHI; Data represent years 1999 to 2017. Table 4.6 Comparison of long-term average of yearly summaries of original and site-adapted values Meteo station DNI annual values GHI annual values Original Site-adapted Difference Original Site-adapted Difference [kWh/m2] [kWh/m2] [%] [kWh/m2] [kWh/m2] [%] Hanimaadhoo 1524 1440 -5.5 2030 2018 -0.6 Hulhulé 1619 1485 -8.3 2051 2052 +0.1 Kadhdhoo 1685 1555 -7.7 2058 2046 -0.6 Gan 1748 1609 -8.0 2076 2048 -1.4 © 2018 Solargis page 45 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 5 METEOROLOGICAL MODEL DATA 5.1 Meteorological model For the territory of Maldives, the last 19 years of the Solargis model-based meteorological data is derived from the regional models. The meteorological data in Solargis database is derived from the two data sources: CFSR and CFSv2, with original characteristics specified in Table 5.1. Table 5.1 Original source of Solargis meteorological data: models CFSR and CFSv2. Climate Forecast System Reanalysis Climate Forecast System (CFSR) (CFSv2) Time period 1999 to 2010 2011 to the present time Original spatial resolution 30 x 35 km 19 x 22 km Original time resolution 1 hour 1 hour Table 5.2 shows meteorological parameters available in Solargis, their specifications and it also indicates, which of them have been delivered within this study. In general, the original spatial resolution of the models is enhanced to 1 km for air temperature and relative humidity by spatial disaggregation and use of the Digital Elevation Model SRTM-3; for Maldives this correction has no effect. The spatial resolution of other parameters is unchanged. Table 5.2 Solargis meteorological parameters delivered within this project Time Spatial Data Data Meteorological parameter Acronym Unit resolution representation delivered validated Air temperature at 2 metres TEMP °C 60 minute 1 km Yes Yes (dry bulb temperature) Relative humidity at 2 metres RH % 60 minute 1 km Yes Yes Wind speed at 10 metres WS m/s2 60 minute Original model Yes Yes Wind direction at 10 metres WD ° 60 minute Original model Yes Yes Atmospheric pressure AP hPa 60 minute Original model Yes Yes Precipitable water PWAT 60 minute Original model Yes No Important note: meteorological parameters are derived from the numerical weather model outputs and they have lower spatial and temporal resolution. Thus, they do not represent the same accuracy as the solar resource data. Especially wind speed data has higher uncertainty, and it provides only overview information for solar energy projects. Thus, the local microclimate of the site may deviate from the values derived from the global databases. 5.2 Validation of meteorological data The validation procedure was carried out to compare the modelled data with ground-measurements from the 4 meteorological stations installed within the ESMAP project: Hanimaadhoo, Hulhulé, Kadhdhoo and Gan. In general, the data from the meteorological models represent larger area, and it is not capable to represent accurately the local microclimate created by small land mass of the islands in the atolls of the Maldives. © 2018 Solargis page 46 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 5.2.1 Air temperature at 2 metres Air temperature is derived from the model outputs of both CSFR and CSFv2 meteorological models and recalculated at the spatial resolution of 1 km (Table 5.3 and Figures 5.1 to 5.4). Considering spatial and time interpolation, the deviation of the modelled values to the ground observations for hourly values can occasionally reach several degrees of Celsius. Figures 5.1 to 5.4 show graphical representation of the model values accuracy at the meteorological stations. In general, the model matches the ground measurements well. Main issue identified is underestimation of daily temperature amplitude. This is caused by relatively small land mass of the islands (in comparison to the pixel size of the meteorological model). Model air temperature is driven mainly by the air temperature over the ocean, where the daily amplitude is lower than temperature amplitude at individual islands. The limitations of the meteorological models are seen only partially, as the air temperature is changing only a little across the seasons and across the day time. Table 5.3 Air temperature at 2 m: accuracy indicators of the model outputs [ºC]. CFSv2 model Meteorological station Bias Bias Bias Bias Bias RMSD RMSD RMSD mean min max nigh-time day-time hourly daily monthly Hanimaadhoo -0.9 0.7 -2.5 0.1 -1.7 1.8 1.4 1.0 Hulhulé -0.9 0.3 -2.3 -0.3 -1.6 1.5 1.1 0.9 Kadhdhoo -0.5 1.2 -2.3 0.4 -1.4 1.6 0.9 0.6 Gan -0.4 1.2 -1.9 0.4 -1.3 1.4 0.8 0.5 Figure 5.1: Scatterplots of air temperature at 2 m at Hanimaadhoo meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements © 2018 Solargis page 47 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 5.2: Scatterplots of air temperature at 2 m at Hulhulé meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements Figure 5.3: Scatterplots of air temperature at 2 m at Kadhdhoo meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements © 2018 Solargis page 48 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 5.4: Scatterplots of air temperature at 2 m at Gan meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis) Blue: day-time, Black: night-time measurements 5.2.2 Relative humidity Relative humidity is calculated from the specific humidity, air pressure and the site-adapted air temperature. The comparison of the model values with ground measurements at all 4 meteorological stations is shown in Table 5.4 and Figures 5.5 to 5.8. The issue identified is amplitude reduction of daily relative humidity. Otherwise, relative humidity is quite stable across the seasons and day time. Table 5.4 Relative humidity: accuracy indicators of the model outputs [%]. CFSv2 model Meteorological station Bias Bias Bias Bias Bias RMSD RMSD RMSD mean min max nigh-time day-time hourly daily monthly Hanimaadhoo -5 0 -11 -10 -2 11 13 8 Hulhulé -5 0 -9 -7 -3 7 6 5 Kadhdhoo -7 -2 -12 -10 -4 9 8 7 Gan -7 -2 -12 -10 -4 9 8 7 © 2018 Solargis page 49 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 5.5: Scatterplots of relative humidity at 2 m at Hanimaadhoo meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. Figure 5.6: Scatterplots of relative humidity at 2 m at Hulhulé meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. © 2018 Solargis page 50 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 5.7: Scatterplots of relative humidity at 2 m at Kadhdhoo meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. Figure 5.8: Scatterplots of relative humidity at 2 m at Gan meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time measurements. © 2018 Solargis page 51 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 5.2.3 Wind speed and wind direction at 10 metres Wind speed and direction values delivered within Solargis data represent the height at 10 meters and they are calculated from the CFSR and CFSv2 models, from 10 m wind u- and v- components. The spatial resolution is kept as in the original data. Wind measurements take place at the height of approx. 3 metres, while the model data represent values at height of 10 m above ground. Comparison of the modelled wind speed with ground measurements is shown in Table 5.5 and Figures 5.9 to 5.12. The model values are higher compared to the wind conditions measured at the meteorological stations. Similarly to relative humidity, the data representation for wind speed and wind direction strongly depends on the local conditions; therefore the model values are only indicative and better characterize a larger region rather than the local microclimate. The important source of systematic difference is different height of the installed sensor (3 metres above ground), compared to the model assumptions (10 metres). Table 5.5 Wind speed: accuracy indicators of the model outputs [m/s]. CFSR and CFSv2 models Meteorological station Bias Bias Bias Bias Bias RMSD RMSD RMSD mean min max nigh-time day-time hourly daily monthly Hanimaadhoo 2.8 2.4 3.1 3.0 2.6 3.3 3.2 3.1 Hulhulé 2.4 2.2 2.4 2.5 2.2 2.9 2.7 2.6 Kadhdhoo 2.8 2.4 3.1 3.0 2.7 3.4 3.2 3.0 Gan 1.5 1.4 1.4 1.8 1.1 2.1 1.8 1.5 Figure 5.9: Scatterplots of wind speed at Hanimaadhoo meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m height and model data at 10 m) © 2018 Solargis page 52 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 5.10: Scatterplots of wind speed at Hulhulé meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m height and model data at 10 m) Figure 5.11: Scatterplots of wind speed at Kadhdhoo meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m height and model data at 10 m) © 2018 Solargis page 53 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 5.12: Scatterplots of wind speed at Gan meteorological station. Measured values (horizontal axis) and meteorological model values (vertical axis). Blue: day-time, black: night-time. (observations at 3 m heigth and model data at 10 m) 5.2.4 Atmospheric pressure It was found that atmospheric pressure from the model fits well measured data with a very small bias for all stations, not exceeding 2 hPa. 5.3 Uncertainty of meteorological model data The meteorological parameters are derived from two very similar numerical meteorological models covering periods from 1999 to 2010 (CFSR model) and 2011 to 2017 (CFSv2). Considering the comparison results, the uncertainty of the estimate for the main meteorological parameters is summarised in Table 5.6. The uncertainty expresses 80% probability of occurrence, so that it can be used for calculation of 90% probability of exceedance values. It was found that the modelled air temperature fits reasonably well the measured data though (logically) due to the spatial resolution, the model daily amplitude is significantly smaller than amplitude of measured data. Similarly to air temperature, relative humidity from the model exhibits much smaller daily amplitude than measured relative humidity. Wind speed data, obtained from the meteorological model, represents an area of larger region and this data is smoothed, in comparison to the point measurements collected at the meteorological sites. For all stations at Maldives wind speed is overestimated by the meteorological model. The reason is coarse resolution of the model and different elevation (observations at approx. 3 m height and the model data representing 10 m) © 2018 Solargis page 54 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 5.6 Expected uncertainty of modelled meteorological parameters at the project site. Unit Annual Monthly Hourly Air temperature at 2 m °C ±1.0 ±1.0 ±2.0 Relative Humidity at 2 m % ±8 ±8 ±15 Average wind speed at 10 m m/s ±3.0 ±3.5 ±4.0 © 2018 Solargis page 55 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 6 SOLAR RESOURCE: UNCERTAINTY OF LONGTERM ESTIMATES 6.1 Uncertainty of solar resource yearly estimate The uncertainty of site-adapted satellite-based GHI and DNI is determined by the uncertainty of the model and of the ground measurements [7], more specifically it depends on: 1. Parameterization and adaptation of numerical models integrated in Solargis for the given data inputs and their ability to generate accurate results for various geographical and time-variable conditions: • Data inputs into Solargis model (accuracy of satellite data, aerosols and water vapour and Digital Terrain Model). • Clear-sky model and its capability to properly characterize various states of the atmosphere • Simulation accuracy of the satellite model and cloud transmittance algorithms, being able to properly distinguish different types of desert surface, clouds, fog, but also snow and ice. • Diffuse and direct decomposition models • Site adaptation methods. 2. Uncertainty of the ground-measurements, which is determined by: • Accuracy of the instruments • Maintenance practices, including sensor cleaning, calibration • Data post-processing and quality control procedures. Accuracy statistics, such as bias and RMSD (Chapter 4.3) characterize accuracy of Solargis model in the given validation points, relative to the ground measurements. The validation statistics is affected by local geography and by quality and reliability of the ground-measured data. Therefore the validation statistics only indicates performance of the model in the region. Solargis model uncertainty is compared to the high-quality data measured by the meteorological instruments. Representativeness of such data comparison (satellite and ground-measured) is determined by the precision of the measuring instruments, the maintenance and operational practices, and by quality control of the measured data – in other words, by the measurement accuracy achieved at each meteorological station. From the user’s perspective, the information about the model uncertainty has probabilistic nature. It generalizes the validation accuracy and it has to be considered at different confidence levels. The expert estimate of the calculation uncertainty in this report assumes 80% probability of occurrence of values, so that it is straightforward to calculate P90 exceedance values. The solar model uncertainty is discussed in Chapters 4 and 6.1. The main findings are summarized in Table 6.1. The site-adaptation procedure reduced uncertainty of estimate of all parameters. Chapter 6.3 evaluates combined uncertainty, in which also interannual variability is included (Chapter 6.2). Table 6.1 Uncertainty of the model estimates for original and site-adapted annual long-term values (Considers 80% probability of occurrence) Uncertainty of long-term Acronym Uncertainty of the original Uncertainty of the Solargis model after site annual values Solargis model adaptation After 1st year After 2nd year Global Horizontal Irradiation GHI ±5.0% ±3.5% ±3.0% Direct Normal Irradiance DNI ±11.0% ±6.0% ±6.0% The reduction of the model uncertainty was significant already after site adaptation using one year of ground measurements; and it was further reduced for GHI after second year of measurement campaign (Table 6.1). A higher uncertainty of DNI, measured with lower accuracy RSP instrument, is also a result of some quality issues as well as higher temporal instability of the instrument (Chapter 3.3). As a result, the uncertainty of site-adapted DNI after second year of measurements remained unchanged. In case of both GHI and DNI parameters, the site adaptation has increased confidence in the model data values. © 2018 Solargis page 56 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 6.2 Uncertainty due to interannual variability of solar radiation Weather changes in cycles and has also stochastic nature. Therefore annual solar radiation in each year can deviate from the long-term average in the range of few percent. The estimation of the interannual variability below shows the magnitude of this change. The uncertainty of GHI and DNI prediction is highest if only one single year is considered, but when averaged for a longer period, weather oscillations even out and approximate to the long- term average. In this report, the interannual variability is calculated from the unbiased standard deviation stdev of GHI and DNI over 19 years, considering, in the long-term, the normal distribution of the annual sums for n years, where xi is any particular year and ̅ is longterm yearly average. Due to the limited number of years of available data, for the calculation we apply simplified assumption of normal distribution of yearly values: 1 = √ ∑ =1( − ̅ ) 2 −1 Tables 6.2 and 6.3 show GHI and DNI values that are to be exceeded at P90 for a consecutive number of years. The variability (varn) for a number of years (n) is calculated from the unbiased standard deviation (stdev): = √ The uncertainty, which characterises 80% probability of occurrence (Uvar), is calculated from the variability (varn), multiplying it with 1.28155: = 1.28155 The lower boundary (negative value) of uncertainty represents 90% probability of exceedance, and it is used for calculating the P90 value. Table 6.2 Annual GHI that should be exceeded with 90% probability in the period of 1 to 10 (25) years Hanimaadhoo│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 1.3 0.9 0.8 0.7 0.6 0.5 0.5 0.5 0.4 0.4 0.3 Uncertainty P90 [±%] 1.7 1.2 1.0 0.8 0.7 0.7 0.6 0.6 0.6 0.5 0.3 Minimum GHI P90 1984 1994 1999 2001 2003 2004 2005 2006 2007 2007 2011 Hulhulé│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 1.5 1.1 0.9 0.8 0.7 0.6 0.6 0.5 0.5 0.5 0.3 Uncertainty P90 [±%] 2.0 1.4 1.1 1.0 0.9 0.8 0.7 0.7 0.7 0.6 0.4 Minimum GHI P90 2012 2023 2029 2032 2034 2035 2037 2038 2039 2039 2044 Kadhdhoo│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 2.1 1.5 1.2 1.0 0.9 0.8 0.8 0.7 0.7 0.6 0.4 Uncertainty P90 [±%] 2.6 1.9 1.5 1.3 1.2 1.1 1.0 0.9 0.9 0.8 0.5 Minimum GHI P90 1992 2007 2014 2019 2021 2024 2025 2026 2028 2029 2035 Gan│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 2.0 1.4 1.2 1.0 0.9 0.8 0.8 0.7 0.7 0.6 0.4 Uncertainty P90 [±%] 2.6 1.8 1.5 1.3 1.2 1.1 1.0 0.9 0.9 0.8 0.5 Minimum GHI P90 1994 2010 2017 2021 2024 2026 2027 2029 2030 2031 2037 © 2018 Solargis page 57 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 6.3 Annual DNI that should be exceeded with 90% probability in the period of 1 to 10 (25) years. Hanimaadhoo│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 3.6 2.5 2.1 1.8 1.6 1.5 1.4 1.3 1.2 1.1 0.7 Uncertainty P90 [±%] 4.6 3.3 2.7 2.3 2.1 1.9 1.7 1.6 1.5 1.5 0.9 Minimum DNI P90 1373 1393 1401 1406 1410 1412 1414 1416 1417 1419 1426 Hulhulé│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 4.1 2.9 2.3 2.0 1.8 1.7 1.5 1.4 1.4 1.3 0.8 Uncertainty P90 [±%] 5.2 3.7 3.0 2.6 2.3 2.1 2.0 1.8 1.7 1.6 1.0 Minimum DNI P90 1408 1430 1440 1446 1450 1453 1456 1458 1459 1460 1469 Kadhdhoo│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 4.5 3.2 2.6 2.2 2.0 1.8 1.7 1.6 1.5 1.4 0.9 Uncertainty P90 [±%] 5.8 4.1 3.3 2.9 2.6 2.4 2.2 2.0 1.9 1.8 1.2 Minimum DNI P90 1466 1492 1503 1510 1515 1519 1521 1523 1525 1527 1537 Gan│Years 1 2 3 4 5 6 7 8 9 10 25 Variability [±%] 4.3 3.0 2.5 2.1 1.9 1.7 1.6 1.5 1.4 1.4 0.9 Uncertainty P90 [±%] 5.5 3.9 3.2 2.7 2.5 2.2 2.1 1.9 1.8 1.7 1.1 Minimum DNI P90 1520 1546 1558 1564 1569 1573 1575 1577 1579 1581 1591 We can interpret the above Table 6.2 and 6.3 on the example of Hulhulé site: i. GHI interannual variability at P90 of 2.0% has to be considered for any single year in Hulhulé. In other words, assuming that the long-term average is 2052 kWh/m2, it is expected (with 90% probability) that annual GHI exceeds, at any single year, the value of 2012 kWh/m2. ii. Within a period of three consecutive years, it is expected at P90 that annual average of GHI exceeds value of 2029 kWh/m2; iii. For a period of 25 years, it is expected at 90% probability that due to interannual variability the estimate of the long-term annual DNI average will deviate within the range of ±1.0% in Hulhulé. Thus assuming that the estimate of the long-term average is 1485 kWh/m2, it can be expected at P90 that due to variability of weather, it should be at least 1469 kWh/m2. It is to be underlined that prediction of the future irradiation is based on the analysis of the recent historical data (period 1999 to 2017). Future weather changes may include man-induced or natural events such as volcano eruptions, which may have impact on this prediction. Based on the existing scientific knowledge [20, 21], an effect of extreme volcano eruptions, with an emission of large amount of stratospheric aerosols, can be estimated on the example of Pinatubo event in 1991 (the second largest volcano eruption in 20th century). It can be expected that in such a case, the annual DNI in the affected year can decrease by approx. 16% or more, compared to the long-term average, still influencing another two consecutive years. In the same way, the volcano eruption of the comparable size may reduce long-term average estimate of DNI by about 4%. The decrease of GHI is much lower; the annual value in the particular year of eruption could be reduced by about 2% compared to the long-term average. © 2018 Solargis page 58 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 6.3 Combined uncertainty In this Chapter, the combined uncertainty of the annual GHI and DNI values is quantified. Taking into account uncertainties of both types of data (satellite and ground measured), the combined effect of two components of the uncertainty of the site-adapted GHI and DNI values has to be considered. 1. Uncertainty of the estimate (Uest) of the annual solar resource values, which is ±3. 0% for GHI and ±6.0% for DNI (Chapter 6.1); 2. Interannual variability (Uvar) in any particular year, due to changing weather. In four Maldivian sites, it varies from ±1.7% to ±2.6% for GHI and from ±4.6% to ±5.8% for DNI. The uncertainty due to weather variability decreases over the time with square root of the number of years (Chapter 6.2). The two above-mentioned uncertainties combine in Uc (see Glossary), which represents a conservative expectation of the minimum GHI and DNI assuming various number of years N (Tables 6.4 and 6.5). Considering a simplified assumption of normal distribution of the annual values, probability of exceedance can be calculated at different confidence levels. GHI and DNI minimum annual values expected for combined uncertainty in any single year are shown on Figure 6.2 and 6.3. Table 6.4 Combined probability of exceedance of annual GHI for uncertainty of the estimate ±3. 0%. Nr. of Uncertainty Interanual Combined Expected minimum│Hanimaadhoo years of estimate variability uncertainty [kWh/m2] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 3.0 1.7 3.4 2144 2107 2087 2055 2018 1982 1949 1929 1892 5 3.0 0.7 3.1 2131 2098 2081 2051 2018 1985 1956 1938 1905 10 3.0 0.5 3.0 2130 2097 2080 2051 2018 1986 1957 1939 1907 25 3.0 0.3 3.0 2129 2096 2079 2050 2018 1986 1957 1940 1908 Nr. of Uncertainty Interanual Combined Expected minimum│Hulhulé years of estimate variability uncertainty [kWh/m2] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 3.0 2.0 3.6 2185 2146 2126 2091 2052 2013 1978 1958 1918 5 3.0 0.9 3.1 2168 2134 2116 2086 2052 2018 1988 1970 1936 10 3.0 0.6 3.1 2166 2133 2115 2085 2052 2019 1989 1971 1938 25 3.0 0.4 3.0 2165 2132 2114 2085 2052 2019 1990 1972 1939 Nr. of Uncertainty Interanual Combined Expected minimum│Kadhdhoo years of estimate variability uncertainty [kWh/m2] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 3.0 2.6 4.0 2194 2150 2127 2088 2046 2003 1964 1941 1897 5 3.0 1.2 3.2 2165 2130 2111 2080 2046 2011 1980 1961 1926 10 3.0 0.8 3.1 2161 2127 2109 2079 2046 2012 1982 1964 1930 25 3.0 0.5 3.0 2159 2125 2108 2078 2046 2013 1983 1966 1932 Nr. of Uncertainty Interanual Combined Expected minimum │Gan years of estimate variability uncertainty [kWh/m2] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 3.0 2.6 4.0 2195 2152 2129 2090 2048 2005 1966 1943 1900 5 3.0 1.2 3.2 2167 2132 2113 2082 2048 2013 1982 1963 1928 10 3.0 0.8 3.1 2163 2129 2111 2081 2048 2014 1984 1966 1932 25 3.0 0.5 3.0 2161 2128 2110 2080 2048 2015 1985 1968 1934 © 2018 Solargis page 59 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 6.5 Combined probability of exceedance of annual DNI for uncertainty of the estimate ±6.0%. Nr. of Uncertainty Interanual Combined Expected minimum│Hanimaadhoo years of estimate variability uncertainty 2 [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 6.0 4.6 7.6 1637 1579 1549 1497 1440 1382 1331 1300 1242 5 6.0 2.1 6.3 1605 1557 1531 1488 1440 1392 1348 1322 1274 10 6.0 1.5 6.2 1601 1554 1528 1486 1440 1393 1351 1326 1278 25 6.0 0.9 6.1 1598 1552 1527 1486 1440 1394 1352 1327 1281 Nr. of Uncertainty Interanual Combined Expected minimum│Hulhulé years of estimate variability uncertainty 2 [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 6.0 5.2 7.9 1699 1636 1603 1547 1485 1423 1367 1333 1271 5 6.0 2.3 6.4 1659 1608 1581 1535 1485 1435 1389 1362 1311 10 6.0 1.6 6.2 1653 1604 1577 1534 1485 1436 1393 1366 1317 25 6.0 1.0 6.1 1649 1601 1575 1533 1485 1437 1395 1369 1321 Nr. of Uncertainty Interanual Combined Expected minimum│Kadhdhoo years of estimate variability uncertainty 2 [kWh/m ] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 6.0 5.8 8.3 1790 1721 1684 1623 1555 1487 1426 1389 1320 5 6.0 2.6 6.5 1739 1685 1657 1609 1555 1502 1454 1425 1371 10 6.0 1.8 6.3 1732 1680 1653 1606 1555 1504 1458 1430 1378 25 6.0 1.2 6.1 1728 1677 1650 1605 1555 1505 1460 1433 1383 Nr. of Uncertainty Interanual Combined Expected minimum│Gan years of estimate variability uncertainty [kWh/m2] N [±%] N years [±%] P90 [±%] P01 P05 P10 P25 P50 P75 P90 P95 P99 1 6.0 5.5 8.1 1846 1776 1739 1677 1609 1540 1478 1441 1371 5 6.0 2.5 6.5 1798 1742 1713 1663 1609 1554 1504 1475 1419 10 6.0 1.7 6.2 1791 1738 1709 1661 1609 1556 1508 1480 1426 25 6.0 1.1 6.1 1787 1735 1707 1660 1609 1557 1510 1483 1430 This analysis is based on the data representing a history of year 1999 to 2017, and on the expert extrapolation of the related weather variability. This report may not reflect possible man-induced climate change or occurrence of extreme events such as large volcano eruptions in the future (see the last paragraph in Chapter 6.2). Graphical visualisation of Tables 6.4 and 6.5 on the example of the Hulhulé Airport is shown in Figures 6.1 and 6.2, where the expected probabilities of exceedance (different Pxx scenarios) are drawn on the cumulative distribution curve showing yearly GHI and DNI values. © 2018 Solargis page 60 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 100 P99: 1918 P95: 1958 GHI Value at Pxx P50 P90: 1978 P75 P90 90 P95 P99 80 P75: 2013 70 60 P50: 2052 Pxx 50 40 30 20 10 0 1850 1950 2050 2150 2250 GHI [kWh/m2] Figure 6.1: Expected Pxx values for GHI at Hulhulé Airport 100 P99: 1271 P95: 1333 DNI Value at Pxx P50 P90: 1367 P75 P90 90 P95 P99 80 P75: 1423 70 60 P50: 1485 Pxx 50 40 30 20 10 0 1150 1250 1350 1450 1550 1650 1750 DNI [kWh/m2] Figure 6.2: Expected Pxx values for DNI at Hulhulé Airport © 2018 Solargis page 61 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 7 TIME SERIES AND TYPICAL METEOROLOGICAL YEAR DATA 7.1 Delivered data sets This report is accompanied by data sets delivered individually for position of each of four solar meteorological stations in Maldives. The data include (Tables 7.1 and 7.2): • Solar and meteorological measurements, after second level quality assessment (first level was delivered by Suntrace) representing an planned minimum of 24 months of the measuring campaign; • Time series, representing last 19+ years; • Typical Meteorological Year data, representing last 19 years. The data is delivered in formats ready to use in energy simulation software. This report provides detailed insight of the methodologies and results. Table 7.1 Delivered data characteristics Feature Time coverage Primary time step Delivered files Ground measurements Dec 2015 to Mar 2018* 1 minute Quality controlled measurements (Suntrace) 1- minute time step Model data Jan 1999 to Dec 2017 30 minutes Time series – hourly original model Time series – monthly (Solargis) Time series – yearly Model data Jan 1999 to Dec 2017 30 minutes Time series – hourly site adapted model Time series – monthly (Solargis) Time series – yearly Model data Jan 1999 to Dec 2017 hourly Typical Meteorological Year P50 site adapted model Typical Meteorological Year P90 (Solargis) * April 2018 at Kadhdhoo and Gan airport site Table 7.2 Parameters in the delivered site-adapted time series and TMY data (hourly time step) Parameter Acronym Unit TS TMY P50 TMY P90 Global horizontal irradiance GHI W/m2 X X X Direct normal irradiance DNI W/m2 X X X Diffuse horizontal irradiance DIF W/m2 X X X Global tilted irradiance (at optimum angle) GHI W/m2 X - - Solar azimuth SA ° X X X Solar elevation SE ° X X X Air temperature at 2 metres TEMP °C X X X Wind speed at 10 metres WS m/s X X X Wind direction at 10 metres WD ° X X X Relative humidity RH % X X X Air Pressure AP hPa X X X Precipitable Water PWAT kg/m2 X X X © 2018 Solargis page 62 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 7.2 TMY method The Typical Meteorological Year (TMY) data sets are delivered, together with Solargis time series data and this report. TMY contains hourly data derived from the time series covering complete 19 years (1999 to 2017). The data history of 19 years is compressed into one year (Figure 7.1 to 7.3) following two criteria: • Minimum difference between statistical characteristics (annual average, monthly averages) of TMY and long-term time series. This criterion is given about 80% weighting. • Maximum similarity of monthly Cumulative Distribution Functions (CDF) of TMY and full-time series, so that occurrence of typical hourly values is well represented for each month. This criterion is given about 20% weighting. TMY P50 data set is constructed on the monthly basis. For each month the long-term average monthly value and cumulative distribution for each parameters is calculated: Direct Normal Irradiance (DNI), Global Horizontal Irradiance (GHI), Diffuse Horizontal Irradiance (DIF) and Air Temperature (TEMP). Following the monthly data for each individual year from the set of 19 years are compared to the long-term parameters. The monthly data from the year, which resembles the long-term parameters more closely, is selected. The procedure is repeated for all 12 months, and the TMY is constructed by concatenating the selected months into one artificial (but representative) year. The method for calculation P90 data set is based on the TMY P50 method. It has been modified in a way of how a candidate month is selected. The search for set of twelve candidates is repeated in iteration until a condition of minimization of difference between annual P90 value and annual average of new TMY is reached (instead of minimization of differences in monthly means and CDFs, as applied in P50 case). Once the selection converges to minimum difference, the TMY is created by concatenation of selected months. The P90 annual values are calculated for each confidence limit − from the combined uncertainty of estimate and inter -annual variability, which can occur in any year (Chapter 6.3). To derive TMY that fits specific needs of the selected energy application the different weights are given to individual parameters – thus highlighting important properties. In solar energy applications, the higher importance is given to GHI and DNI. In assembling TMY P50, the values of DNI, GHI, DIF and TEMP are only considered, where the weights are set as follows: 0.9 is given to DNI, 0.3 to GHI, 0.02 to diffuse horizontal irradiance, and 0.07 to air temperature (divided by the total of 1.29). To derive solar resource parameters with an hourly time step, the original satellite data with time resolution of 30- minutes were aggregated by time integration. The meteorological parameters are available in the original 1-hourly time step. The TMY datasets were constructed from solar radiation and meteorological data (Chapters 4 and 5). Time zone was adjusted to Maldives Time MVT (UTC +05:00). More about the Solargis TMY method in [22]. 7.3 Results Two data sets are derived from the Solargis historical time series for the four sites: P50 and P90. In graphs and tables below we show the values for Hulhulé meteorological site, in order to present the methodology of TMY data calculation. Important note: Due to the inherent features of the underlying methods, monthly values in TMY data set in Tables 7.3 to 7.6 do not fit to the values generated from full time series (Figures 7.1 to 7.4). Table 7.3 Monthly and yearly long-term GHI averages as calculated from time series and from TMY representing P50, and P90 cases at Hulhulé site Global Horizontal 1 2 3 4 5 6 7 8 9 10 11 12 Year Irradiation [kWh/m2] Time series (19 years) 178 182 207 185 166 157 162 169 163 177 152 155 2052 TMY for P50 case 180 175 206 186 164 158 165 172 156 175 160 155 2052 TMY for P90 case 173 178 208 176 158 153 159 170 156 169 124 152 1978 © 2018 Solargis page 63 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 7.4 Monthly and yearly long-term DNI averages as calculated from time series and from TMY representing P50, and P90 cases at Hulhulé site Direct Normal 1 2 3 4 5 6 7 8 9 10 11 12 Year Irradiation [kWh/m2] Time series (19 years) 140 147 163 143 117 105 104 107 102 131 110 117 1485 TMY for P50 case 141 148 162 142 118 105 103 107 101 128 112 117 1485 TMY for P90 case 133 143 162 134 101 99 101 103 93 117 76 105 1367 Table 7.5 Monthly and yearly long-term DIF averages as calculated from time series and from TMY representing P50, and P90 cases at Hulhulé site Direct Normal 1 2 3 4 5 6 7 8 9 10 11 12 Year Irradiation [kWh/m2] Time series (19 years) 79 73 82 76 79 78 83 86 84 79 73 75 946 TMY for P50 case 81 68 82 79 76 80 87 89 78 79 80 77 956 TMY for P90 case 77 71 82 74 80 77 81 88 81 79 69 78 937 Table 7.6 Monthly and yearly long-term TEMP averages as calculated from time series and from TMY representing P50, and P90 cases at Hulhulé site Air temperature [°C] 1 2 3 4 5 6 7 8 9 10 11 12 Year Time series (19 years) 27.5 27.6 28.2 28.7 28.7 28.4 28.2 28.0 27.9 27.9 27.7 27.5 28.0 TMY for P50 case 27.6 27.5 28.2 28.2 28.4 28.4 28.0 28.3 27.9 27.9 28.2 28.2 28.1 TMY for P90 case 27.5 27.5 28.2 28.7 28.6 28.6 28.1 28.1 27.8 28.0 27.7 27.5 28.0 As an example of interpretation of the tables above, the TMY data sets for P50 and P90 can be described as: 1. P50 TMY data set represents, for each month, the average climate conditions and the most representative cumulative distribution function, therefore extreme situations (e.g. extremely cloudy weather) are not represented in this dataset. The long-term annual summary of GHI and DNI are considered as the most critical parameters to consider, and in this data set P50 GHI value is 2052 kWh/m2 and DNI value is 1485 kWh/m2. 2. P90 TMY data set represents for each month the climate conditions, which after summarization GHI and DNI for the whole year results in the value equal or close to P90 derived by the analysis of uncertainty of the estimate and of the interannual variability for any single year (Chapter 6.3). Thus TMY for P90 represents generally a conservative estimate, i.e. a year with the long-term value of GHI of 1978 kWh/m2 and DNI of 1367 kWh/m2. 250 200 GHI [kWh/m 2 ] 150 100 50 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Time series TMY P50 TMY P90 Figure 7.1: GHI monthly values derived from time series and TMY P50 and P90 at Hulhulé site. © 2018 Solargis page 64 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 250 200 DNI [kWh/m 2 ] 150 100 50 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Time series TMY P50 TMY P90 Figure 7.2: DNI monthly values derived from time series and TMY P50 and P90 at Hulhulé site. 250 200 DIF [kWh/m 2 ] 150 100 50 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Time series TMY P50 TMY P90 Figure 7.3: DIF monthly values derived from time series and TMY P50 and P90 at Hulhulé site 30.0 29.0 TEMP [ C] 28.0 27.0 26.0 25.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Time series TMY P50 TMY P90 Figure 7.4: TEMP monthly values derived from time series and TMY P50 and P90 at Hulhulé site It is important to note that the data reduction in the TMY data set is not possible without loss of information contained in the original multiyear time series. Therefore time series data are considered as the most accurate reference suitable for the statistical analysis of solar resource and meteorological parameters of the sites. © 2018 Solargis page 65 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 7.5: Seasonal profile of GHI, DNI and DIF for Typical Meteorological Year P50 Hulhulé site: X-axis – day of the year; Y-axis – solar irradiance W/m2 © 2018 Solargis page 66 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure 7.6: Snapshot of Typical Meteorological Year for P50 for Hulhulé © 2018 Solargis page 67 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 8 CONCLUSIONS This report accompanies delivery of site-specific solar resource and meteorological data for four sites, where solar meteorological stations have been installed during December 2015 and operated originally for period of 24 months (until December 2017, when extended until March (April in Kadhdhoo and Gan) 2018. Comparing to the results obtained after period of first 12 months of measurement campaign, the data uncertainty has been reduced for GHI and remain unchanged for DNI due to some measurements issues. The measured data is used in site-adaptation of the Solargis model. As a result, reliable historical time series and TMY data is computed and provided in formats ready to use in standard photovoltaic energy simulation software. The reliability of the delivered data is high and it can be used for energy evaluation of any project in Maldives and also for bankable assessment. © 2018 Solargis page 68 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 ANNEX 1: SITE RELATED DATA STATISTICS Yearly summaries of solar and meteorological parameters Statistics for site adapted model yearly values representing 19 years (1999 to 2017). 6.0 2192 Average annual sum of Global Horizontal Irradiation [kWh/m2] Average daily sum of Global Horizontal Irradiation [kWh/m2] 5.5 2009 5.0 1826 4.5 1644 4.0 1461 3.5 1278 3.0 1096 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Hanimaadhoo 1.7% Hulhulé 2.0% Kadhdhoo 2.6% Gan 2.6% Figure I: Interannual variability of site-adapted yearly GHI [kWh/m2]. Annual averages (avg, solid line) and standard deviation (value behind the names of sites). 6.0 2192 Average daily sum of Direct Normal Irradiation [kWh/m 2] Average annual sum of Direct Normal Irradiation [kWh/m 2] 5.5 2009 5.0 1826 4.5 1644 4.0 1461 3.5 1278 3.0 1096 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Hanimaadhoo 4.6% Hulhulé 5.2% Kadhdhoo 5.8% Gan 5.5% Figure II: Interannual variability of site-adapted yearly DNI [kWh/m2]. Annual averages (avg, solid line) and standard deviation (value behind the names of sites). © 2018 Solargis page 69 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 30.0 29.0 Monthly air temperature [°C] 28.0 27.0 26.0 25.0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Hanimaadhoo Hulhulé Kadhdhoo Gan Figure III: Interannual variability of yearly TEMP [C]. Annual averages (avg, solid line). © 2018 Solargis page 70 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Monthly summaries of solar and meteorological parameters The graphs compare (site-adapted) monthly model time series compared to long-term averages. 8.0 8.0 7.0 Hanimaadhoo Hulhulé 7.0 Daily sums of GHI [kWh/m2] Daily sums of GHI [kWh/m2] 6.0 6.0 5.0 5.0 4.0 4.0 3.0 3.0 2.0 2.0 1.0 1.0 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 8.0 8.0 7.0 Kadhdhoo Gan 7.0 Daily sums of GHI [kWh/m2] Daily sums of GHI [kWh/m2] 6.0 6.0 5.0 5.0 4.0 4.0 3.0 3.0 2.0 2.0 1.0 1.0 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year Figure IV: GHI monthly averages [kWh/m2]. Monthly average shown as solid line; min/max monthly values as as boundary lines; last 24 months in red. 8.0 8.0 7.0 Hanimaadhoo 7.0 Hulhulé Daily sums of DNI [kWh/m2] 6.0 6.0 Daily sums of DNI [kWh/m2] 5.0 5.0 4.0 4.0 3.0 3.0 2.0 2.0 1.0 1.0 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 8.0 8.0 7.0 Kadhdhoo 7.0 Gan Daily sums of DNI [kWh/m2] Daily sums of DNI [kWh/m2] 6.0 6.0 5.0 5.0 4.0 4.0 3.0 3.0 2.0 2.0 1.0 1.0 0.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year Figure V: DNI monthly averages [kWh/m2]. Monthly average shown as solid line; min/max monthly values as boundary lines; last 24 months shown in red. © 2018 Solargis page 71 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 30.0 30.0 29.0 Hanimaadhoo 29.0 Hulhulé Monthly air temperature [°C] Monthly air temperature [°C] 28.0 28.0 27.0 27.0 26.0 26.0 25.0 25.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year 30.0 30.0 29.0 Kadhdhoo 29.0 Gan Monthly air temperature [°C] Monthly air temperature [°C] 28.0 28.0 27.0 27.0 26.0 26.0 25.0 25.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec max-min LTA Last Year max-min LTA Last Year Figure VI: TEMP monthly averages [C]. Monthly average shown as solid line; min/max monthly values as boundary lines; last 24 months shown in red. © 2018 Solargis page 72 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Frequency of occurrence of GHI and DNI daily model values for a period 1999 to 2017 The histograms below show occurrence statistics of daily values derived from the satellite-based time series for GHI and DNI. The time covered in the graphs below is 19 complete calendar years (1999 to 2017). The occurrence is calculated separately for each month. 25 12 12 25 12 25 January February March Percentage of days 10 20 10 20 10 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 25 12 25 12 25 April May June Percentage of days 10 10 20 10 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 25 12 12 25 12 25 July August September Percentage of days 10 20 10 20 10 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 25 12 25 12 12 25 October November December Percentage of days 10 10 10 20 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 5 5 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure VII: Histograms of daily summaries of Global Horizontal Irradiation in Hanimaadhoo. 12 25 25 12 12 25 January February March Percentage of days 10 20 10 20 10 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 25 12 25 12 25 April May June Percentage of days 10 10 20 10 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 25 25 12 12 25 July August September Percentage of days 10 20 10 20 10 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 25 12 12 25 25 12 October November December Percentage of days 10 10 10 20 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 5 5 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure VIII: Histograms of daily summaries of Global Horizontal Irradiation in Hulhulé. © 2018 Solargis page 73 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 12 25 12 25 25 12 January February March Percentage of days 10 20 10 20 10 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 25 12 25 12 12 25 April May June Percentage of days 10 10 20 10 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 25 12 25 25 12 July August September Percentage of days 10 20 10 20 10 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 25 12 12 25 25 12 October November December Percentage of days 10 10 10 20 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 5 5 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure IX: Histograms of daily summaries of Global Horizontal Irradiation in Kadhdhoo. 12 25 12 25 25 12 January February March Percentage of days 10 10 10 20 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 5 5 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 25 12 25 12 12 25 April May June Percentage of days 10 10 10 20 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 5 5 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 12 25 12 25 25 12 July August September Percentage of days 10 20 10 20 10 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 2 5 2 5 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 25 12 12 25 25 12 October November December Percentage of days 10 10 10 20 20 20 8 8 8 15 15 15 6 6 6 10 10 10 4 4 4 5 5 5 2 2 2 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Daily sum of GHI [kWh/m2] Figure X: Histograms of daily summaries of Global Horizontal Irradiation in Gan. Figures VII to X show histograms of daily GHI summaries for each month as calculated from Solargis time series representing the years 1999 to 2017. The distribution of daily values is not symmetric: median is drawn by the vertical line, and percentiles P10, P25, and P75, and P90 are displayed with dark grey and light grey colour bands, respectively. The percentiles P10 and P90 show 80% occurrence of daily values within each month and percentiles P25 and P75 show 50% occurrence. © 2018 Solargis page 74 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 10 10 10 January February March Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 April May June Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 July August September Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 October November December Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XI: Histograms of daily summaries of Direct Normal Irradiation in Hanimaadhoo. 10 10 10 January February March Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 April May June Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 July August September Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 October November December Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XII: Histograms of daily summaries of Direct Normal Irradiation in Hulhulé. © 2018 Solargis page 75 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 10 10 10 January February March Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 April May June Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 July August September Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 October November December Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XIII: Histograms of daily summaries of Direct Normal Irradiation in Kadhdhoo. 10 10 10 January February March Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 April May June Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 July August September Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10 10 10 October November December Percentage of days 7.5 7.5 7.5 5 5 5 2.5 2.5 2.5 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Daily sum of DNI [kWh/m2] Figure XIV: Histograms of daily summaries of Direct Normal Irradiation in Gan. Figures XI to XIV show histograms of daily DNI summaries for each month as calculated from Solargis time series representing the years 1999 to 2017. The distribution of daily values is not symmetric: median is drawn by the vertical line, and percentiles P10, P25, and P75, and P90 are displayed with dark grey and light grey colour bands, respectively. The percentiles P10 and P90 show 80% occurrence of daily values within each month and percentiles P25 and P75 show 50% occurrence. © 2018 Solargis page 76 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Frequency of occurrence of GHI and DNI 30-minute model values for a period 1999 to 2017 The histograms below show occurrence statistics of 30-minute values derived from the satellite-based time series for GHI and DNI. The time covered in the graphs below is 19 complete calendar years (1999 to 2017). The occurrence is calculated separately for each month. Figure XV: Histograms and cumulative distribution function of 30-minute GHI in Hanimaadhoo Figure XVI: Histograms and cumulative distribution function of 30-minute GHI in Hulhulé © 2018 Solargis page 77 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure XVII: Histograms and cumulative distribution function of 30-minute GHI in Kadhdhoo Figure XVIII: Histograms and cumulative distribution function of 30-minute GHI in Gan Figures XV to XVIII show monthly histograms (bars) and cumulative distribution (line) of 30-minute GHI values, calculated from Solargis time series. The values represent the occurrence of GHI values within 50 W/m2 bins, ranging from 0 to 1200 W/m2. © 2018 Solargis page 78 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure XIX: Histograms and cumulative distribution function of 30-minute DNI in Hanimaadhoo Figure XX: Histograms and cumulative distribution function of 30-minute DNI in Hulhulé © 2018 Solargis page 79 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure XXI: Histograms and cumulative distribution function of 30-minute DNI in Kadhdhoo Figure XXII: Histograms and cumulative distribution function of 30-minute DNI in Gan Figures XIX to XXII show monthly histograms (bars) and cumulative distribution (line) of 30-minute DNI values, calculated from Solargis time series. The values represent the occurrence of DNI values within 50 W/m2 bins, ranging from 0 to 1200 W/m2. © 2018 Solargis page 80 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Frequency of occurrence of GHI and DNI measured and model values representing two years of ground measurements Figures XXIII to XXX show histograms comparing the measured values with the model GHI and DNI data. The period covered in these histogram is last 24-months (two full years of data, i.e. from 1 Jan 2016 to 31 Dec 2017): • 1-minute measured vs. 30-min satellite-based model values • 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values • Daily measured (aggregated from 1-min) vs. daily satellite-based model values Aggregation process deals with the missing values in the ground measurement in three steps: 1. Only those 1-minute measured data values that passed through quality control (Chapter 3.3) is taken into account (satellite time series does not have gaps.); 2. Aggregation of 1-minute measured data values into 30-minute slots (equivalent to satellite time slots) is applied if more than 15 valid data-points is available, otherwise the 30-minute data slot is ignored in further statistical comparison; 3. Daily aggregation of measured data represents the same 30-minute time slots in a day (passing through the two steps above), as those in the satellite-based data. Incorrect data slots found in the measurements are excluded in both the measured and model data. Figure XXIII: Measured vs. satellite-based GHI values in Hanimaadhoo 1-minute measured vs. 30-min satellite-based values. 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXIV: Measured vs. satellite-based GHI values in Hulhulé 1-minute measured vs. 30-min satellite-based values. 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values © 2018 Solargis page 81 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure XXV: Measured vs. satellite-based GHI values in Kadhdhoo 1-minute measured vs. 30-min satellite-based values. 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXVI: Measured vs. satellite-based GHI values in Gan 1-minute measured vs. 30-min satellite-based values. 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXVII: Measured vs. satellite-based DNI values in Hanimaadhoo 1-minute measured vs. 30-min satellite-based values. 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXVIII: Measured vs. satellite-based DNI values in Hulhulé 1-minute measured vs. 30-min satellite-based values. 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values © 2018 Solargis page 82 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure XXIX: Measured vs. satellite-based DNI values in Kadhdhoo 1-minute measured vs. 30-min satellite-based values. 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values Figure XXX: Measured vs. satellite-based DNI values in Gan 1-minute measured vs. 30-min satellite-based values. 30-minute measured (aggregated from 1-min) vs. 30-min satellite-based values Daily measured (aggregated from 1-min) vs. daily satellite-based values © 2018 Solargis page 83 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Frequency of occurrence of GHI and DNI ramps Figures XXXI to XXXVIII show histograms of instantaneous changes (ramps) calculated from the measurements and compared to the instantaneous changes calculated for the model data. Figures show both negative (-) and positive (+) changes. Two versions for GHI and DNI are shown: • Ramps calculated from 1-minute measured values compared to ramps calculated from 30-minute satellite-based data (figure on the left) • Ramps calculated from 30-minute aggregated valid measurement compared to ramps calculated from 30-minute satellite-based data (figure on the right). Occurrence of gaps in the measurements is managed in the same way as described about in this Chapter: 1. For measurements, only those 1-minute data values (measurements) that passed through quality control (Chapter 3.3) is taken into account (satellite time series does not have gaps.); 2. For measurements, the aggregation (averaging) of 1-minute measured data values into 30-minute slots (equivalent to satellite time slots) is applied if more than 15 valid data-points is available, otherwise the 30-minute data slot is ignored in further statistical comparison; Figure XXXI: 1-minute and 30-minute GHI ramps (measured and satellite data) at Hanimaadhoo. Figure XXXII: 1-minute and 30-minute GHI ramps (measured and satellite data) at Hulhulé © 2018 Solargis page 84 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure XXXIII: 1-minute and 30-minute GHI ramps (measured and satellite data) at Kadhdhoo Figure XXXIV: 1-minute and 30-minute GHI ramps (measured and satellite data) at Gan Figure XXXV: 1-minute and 30-minute DNI ramps (measured and satellite data) at Hanimaadhoo © 2018 Solargis page 85 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure XXXVI: 1-minute and 30-minute DNI ramps (measured and satellite data) at Hulhulé Figure XXXVII: 1-minute and 30-minute DNI ramps (measured and satellite data) at Kadhdhoo Figure XXXVIII: 1-minute and 30-minute DNI ramps (measured and satellite data) at Gan © 2018 Solargis page 86 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 LIST OF FIGURES Figure 2.1: Position of solar meteorological stations in Maldives ....................................................................... 11 Figure 3.1 Results of DNI and GHI quality control in Hanimaadhoo. ................................................................... 19 Figure 3.2 Systematic difference between GHI from SR20 and RSP − Hanimaadhoo. ........................................ 20 Figure 3.3 Results of DNI and GHI quality control in Hulhulé. .............................................................................. 22 Figure 3.4 Effect of RSP shadowband issues – drop of DNI in Hulhulé. .............................................................. 23 Figure 3.5 Systematic difference between GHI from SR20 and RSP - Hulhulé..................................................... 23 Figure 3.6 Results of DNI and GHI quality control − Kadhdhoo. .......................................................................... 26 Figure 3.7 Effect of RSP shadowband issues – drop of DNI in Kadhdhoo. .......................................................... 27 Figure 3.8 Systematic difference between GHI from SR20 and RSP − Kadhdhoo. .............................................. 27 Figure 3.9 Results of DNI and GHI quality control − Gan. .................................................................................... 30 Figure 3.10 Effect of RSP shadowband issues – drop of DNI in Gan. ................................................................. 31 Figure 3.11 Systematic difference between GHI from SR20 and RSP - Gan. ....................................................... 31 Figure 3.12 Replacement of SR20 GHI by RSP GHI due to wind mast shading in Gan. ........................................ 32 Figure 4.1: Sites with solar meteorological stations used for site adaptation of Solargis model in Maldives ..... 37 Figure 4.2: Correction of DNI and GHI hourly values for Hanimaadhoo. .............................................................. 41 Figure 4.3: Correction of DNI and GHI hourly values for Hulhulé ......................................................................... 42 Figure 4.4: Correction of DNI and GHI hourly values for Kadhdhoo. .................................................................... 43 Figure 4.5: Correction of DNI and GHI hourly values for Gan. .............................................................................. 44 Figure 4.6: Comparison of Solargis original and site-adapted data for Hanimaadhoo site.................................. 45 Figure 5.1: Scatterplots of air temperature at 2 m at Hanimaadhoo meteorological station. .............................. 47 Figure 5.2: Scatterplots of air temperature at 2 m at Hulhulé meteorological station. ........................................ 48 Figure 5.3: Scatterplots of air temperature at 2 m at Kadhdhoo meteorological station. .................................... 48 Figure 5.4: Scatterplots of air temperature at 2 m at Gan meteorological station. .............................................. 49 Figure 5.5: Scatterplots of relative humidity at 2 m at Hanimaadhoo meteorological station. ............................ 50 Figure 5.6: Scatterplots of relative humidity at 2 m at Hulhulé meteorological station. ....................................... 50 Figure 5.7: Scatterplots of relative humidity at 2 m at Kadhdhoo meteorological station. .................................. 51 Figure 5.8: Scatterplots of relative humidity at 2 m at Gan meteorological station. ............................................ 51 Figure 5.9: Scatterplots of wind speed at Hanimaadhoo meteorological station. ............................................... 52 Figure 5.10: Scatterplots of wind speed at Hulhulé meteorological station. ....................................................... 53 Figure 5.11: Scatterplots of wind speed at Kadhdhoo meteorological station. ................................................... 53 Figure 5.12: Scatterplots of wind speed at Gan meteorological station. ............................................................. 54 Figure 6.1: Expected Pxx values for GHI at Hulhulé Airport ................................................................................. 61 Figure 6.2: Expected Pxx values for DNI at Hulhulé Airport ................................................................................. 61 Figure 7.1: GHI monthly values derived from time series and TMY P50 and P90 ................................................ 64 Figure 7.2: DNI monthly values derived from time series and TMY P50 and P90 ................................................ 65 Figure 7.3: DIF monthly values derived from time series and TMY P50 and P90 ................................................. 65 Figure 7.4: TEMP monthly values derived from time series and TMY P50 and P90 ............................................ 65 Figure 7.5: Seasonal profile of GHI, DNI and DIF for Typical Meteorological Year P50 ........................................ 66 Figure 7.6: Snapshot of Typical Meteorological Year for P50 for Hulhulé ........................................................... 67 Figure I: Interannual variability of site-adapted yearly GHI [kWh/m2]. .................................................................. 69 © 2018 Solargis page 87 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Figure II: Interannual variability of site-adapted yearly DNI [kWh/m2]. ................................................................. 69 Figure III: Interannual variability of yearly TEMP [C]. .......................................................................................... 70 Figure IV: GHI monthly averages [kWh/m2]. ........................................................................................................ 71 Figure V: DNI monthly averages [kWh/m2]........................................................................................................... 71 Figure VI: TEMP monthly averages [C]. .............................................................................................................. 72 Figure VII: Histograms of daily summaries of Global Horizontal Irradiation in Hanimaadhoo. ............................ 73 Figure VIII: Histograms of daily summaries of Global Horizontal Irradiation in Hulhulé. ..................................... 73 Figure IX: Histograms of daily summaries of Global Horizontal Irradiation in Kadhdhoo. ................................... 74 Figure X: Histograms of daily summaries of Global Horizontal Irradiation in Gan. .............................................. 74 Figure XI: Histograms of daily summaries of Direct Normal Irradiation in Hanimaadhoo. .................................. 75 Figure XII: Histograms of daily summaries of Direct Normal Irradiation in Hulhulé. ............................................ 75 Figure XIII: Histograms of daily summaries of Direct Normal Irradiation in Kadhdhoo. ....................................... 76 Figure XIV: Histograms of daily summaries of Direct Normal Irradiation in Gan. ................................................ 76 Figure XV: Histograms and cumulative distribution function of 30-minute GHI in Hanimaadhoo ....................... 77 Figure XVI: Histograms and cumulative distribution function of 30-minute GHI in Hulhulé ................................. 77 Figure XVII: Histograms and cumulative distribution function of 30-minute GHI in Kadhdhoo............................ 78 Figure XVIII: Histograms and cumulative distribution function of 30-minute GHI in Gan .................................... 78 Figure XIX: Histograms and cumulative distribution function of 30-minute DNI in Hanimaadhoo....................... 79 Figure XX: Histograms and cumulative distribution function of 30-minute DNI in Hulhulé .................................. 79 Figure XXI: Histograms and cumulative distribution function of 30-minute DNI in Kadhdhoo ............................. 80 Figure XXII: Histograms and cumulative distribution function of 30-minute DNI in Gan...................................... 80 Figure XXIII: Measured vs. satellite-based GHI values in Hanimaadhoo ............................................................. 81 Figure XXIV: Measured vs. satellite-based GHI values in Hulhulé........................................................................ 81 Figure XXV: Measured vs. satellite-based GHI values in Kadhdhoo..................................................................... 82 Figure XXVI: Measured vs. satellite-based GHI values in Gan ............................................................................. 82 Figure XXVII: Measured vs. satellite-based DNI values in Hanimaadhoo ............................................................ 82 Figure XXVIII: Measured vs. satellite-based DNI values in Hulhulé...................................................................... 82 Figure XXIX: Measured vs. satellite-based DNI values in Kadhdhoo .................................................................... 83 Figure XXX: Measured vs. satellite-based DNI values in Gan............................................................................... 83 Figure XXXI: 1-minute and 30-minute GHI ramps (measured and satellite data) at Hanimaadhoo. ................... 84 Figure XXXII: 1-minute and 30-minute GHI ramps (measured and satellite data) at Hulhulé ............................... 84 Figure XXXIII: 1-minute and 30-minute GHI ramps (measured and satellite data) at Kadhdhoo .......................... 85 Figure XXXIV: 1-minute and 30-minute GHI ramps (measured and satellite data) at Gan ................................... 85 Figure XXXV: 1-minute and 30-minute DNI ramps (measured and satellite data) at Hanimaadhoo .................... 85 Figure XXXVI: 1-minute and 30-minute DNI ramps (measured and satellite data) at Hulhulé ............................. 86 Figure XXXVII: 1-minute and 30-minute DNI ramps (measured and satellite data) at Kadhdhoo ........................ 86 Figure XXXVIII: 1-minute and 30-minute DNI ramps (measured and satellite data) at Gan ................................. 86 © 2018 Solargis page 88 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 LIST OF TABLES Table 1.1 Delivered data characterstics .......................................................................................................... 9 Table 1.2 Parameters in the delivered site-adapted time series and TMY data (hourly time step)................ 10 Table 2.1 Overview information on the solar meteorological stations installed in Maldives ......................... 12 Table 3.1 Overview information on measurement stations operated in the region ....................................... 13 Table 3.2 Instruments installed at the solar meteorological stations ........................................................... 13 Table 3.3 Technical parameters and accuracy class of the instruments ...................................................... 13 Table 3.4 Overview information on solar meteorological stations operating in the region ........................... 14 Table 3.5 Data recovery statistics of the two-year measurement campaign period...................................... 14 Table 3.6 Data recovery statistics of the measurement campaign period including prolongation ................ 15 Table 3.7 Period of measurements analysed in this report ........................................................................... 15 Table 3.8 Meteorological stations maintenance and instruments field verification ...................................... 15 Table 3.9 Occurrence of data readings for Hanimaadhoo meteorological station ........................................ 17 Table 3.10 Excluded ground measurements after quality control (Sun above horizon) in Hanimaadhoo ....... 17 Table 3.11 Quality control summary – Hanimaadhoo..................................................................................... 20 Table 3.12 Occurrence of data readings for Hulhulé meteorological station .................................................. 21 Table 3.13 Excluded ground measurements after quality control (Sun above horizon) in Hulhulé ................. 21 Table 3.14 Quality control summary – Hulhulé ............................................................................................... 24 Table 3.15 Occurrence of data readings for Kadhdhoo meteorological station .............................................. 25 Table 3.16 Excluded ground measurements after quality control (Sun above horizon) in Kadhdhoo ............. 25 Table 3.17 Quality control summary – Kadhdhoo ........................................................................................... 28 Table 3.18 Occurrence of data readings for Gan meteorological station ........................................................ 29 Table 3.19 Excluded ground measurements after quality control (Sun above horizon) in Gan ....................... 29 Table 3.20 Quality control summary – Gan..................................................................................................... 32 Table 3.21 Updated re-calibration schedule proposed by Suntrace ................................................................ 34 Table 3.22 Overview of instruments to be re-calibrated and/or exchanged .................................................... 35 Table 4.1 Input data used in the Solargis and related GHI and DNI outputs for Maldives ............................. 36 Table 4.2 Direct Normal Irradiance: bias and KSI before and after model site-adaptation ............................ 39 Table 4.3 Global Horizontal Irradiance: bias and KSI before and after model site-adaptation ...................... 40 Table 4.4 Direct Normal Irradiance: RMSD before and after model site-adaptation ...................................... 40 Table 4.5 Global Horizontal Irradiance: RMSD before and after model site-adaptation ................................ 40 Table 4.6 Comparison of long term average of yearly summaries of original and site-adapted values ........ 45 Table 5.1 Original source of Solargis meteorological data: models CFSR and CFSv2. .................................. 46 Table 5.2 Solargis meteorological parameters delivered within this project ................................................. 46 Table 5.3 Air temperature at 2 m: accuracy indicators of the model outputs [ºC]. ........................................ 47 Table 5.4 Relative humidity: accuracy indicators of the model outputs [%]. .................................................. 49 Table 5.5 Wind speed: accuracy indicators of the model outputs [m/s]. ...................................................... 52 Table 5.6 Expected uncertainty of modelled meteorological parameters at the project site. ........................ 55 Table 6.1 Uncertainty of the model estimates for original and site-adapted annual long-term values .......... 56 Table 6.2 Annual GHI that should be exceeded with 90% probability in the period of 1 to 10 (25) years ...... 57 Table 6.3 Annual DNI that should be exceeded with 90% probability in the period of 1 to 10 (25) years. ..... 58 © 2018 Solargis page 89 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 Table 6.4 Combined probability of exceedance of annual GHI for uncertainty of the estimate ±3.0%. ......... 59 Table 6.5 Combined probability of exceedance of annual DNI for uncertainty of the estimate ±6.0%. ......... 60 Table 7.1 Delivered data characteristics ....................................................................................................... 62 Table 7.2 Parameters in the delivered site-adapted time series and TMY data (hourly time step)................ 62 Table 7.3 Monthly and yearly long-term GHI averages as calculated from time series and from TMY ......... 63 Table 7.4 Monthly and yearly long-term DNI averages as calculated from time series and from TMY ......... 64 Table 7.5 Monthly and yearly long-term DIF averages as calculated from time series and from TMY .......... 64 Table 7.6 Monthly and yearly long-term TEMP averages as calculated from time series and from TMY ...... 64 © 2018 Solargis page 90 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 REFERENCES [1] NREL, 1993. User’s Manual for SERI_QC Software-Assessing the Quality of Solar Radiation Data. NREL/TP- 463-5608. Golden, CO: National Renewable Energy Laboratory. [2] Younes S., Claywell R. and Muneer T, 2005. Quality control of solar radiation data: Present status and proposed new approaches. Solar Energy 30, 1533-1549. [3] Perez R., Cebecauer T., Šúri M., 2013. Semi-Empirical Satellite Models. In Kleissl J. (ed.) Solar Energy Forecasting and Resource Assessment. Academic press. http://www.sciencedirect.com/science/article/pii/B9780123971777000024 [4] Cebecauer T., Šúri M., Perez R., High performance MSG satellite model for operational solar energy applications. ASES National Solar Conference, Phoenix, USA, 2010. http://solargis.com/assets/publication/2010/Cebecauer-Suri-Perez--ASES2010--High-performance-MSG- satellite-model-for-operational-solar-energy-applications.pdf [5] Šúri M., Cebecauer T., Perez P., 2010. Quality Procedures of Solargis for Provision Site -Specific Solar Resource Information. Conference SolarPACES 2010, September 2010, Perpignan, France. http://solargis.com/assets/publication/2010//Suri-Cebecaue--Perez-SolarPACES2010--Quality-procedures- of-solargis-for-provision-site-specific-solar-resource-information.pdf [6] Cebecauer T., Suri M., Gueymard C., Uncertainty sources in satellite-derived Direct Normal Irradiance: How can prediction accuracy be improved globally? Proceedings of the SolarPACES Conference, Granada, Spain, 20-23 Sept 2011. http://solargis.com/assets/publication/2011/Cebecauer-Suri-Gueymard--SolarPACES2011--Uncertainty- sources-in-satellite-derived-direct-normal-irradiance--How-can-prediction-accuracy-be-improved-globally.pdf [7] Suri M., Cebecauer T., 2014. Satellite-based solar resource data: Model validation statistics versus user’s uncertainty. ASES SOLAR 2014 Conference, San Francisco, 7-9 July 2014. http://solargis.com/assets/publication/2014//Suri-Cebecauer--ASES-Solar2014--Satellite-Based-Solar- Resource-Data--Model-Validation-Statistics-Versus-User-Uncertainty.pdf [8] Ineichen P., A broadband simplified version of the Solis clear sky model, 2008. Solar Energy, 82, 8, 758-762. [9] Benedictow A. et al. 2012. Validation report of the MACC reanalysis of global atmospheric composition: Period 2003-2010, MACC-II Deliverable D83.1. [10] Molod, A., Takacs, L., Suarez, M., and Bacmeister, J., 2015: Development of the GEOS-5 atmospheric general circulation model: evolution from MERRA to MERRA2, Geosci. Model Dev., 8, 1339-1356, doi:10.5194/gmd-8- 1339-2015 [11] GFS model. http://www.nco.ncep.noaa.gov/pmb/products/gfs/ [12] CFSR model. https://climatedataguide.ucar.edu/climate-data/climate-forecast-system-reanalysis-cfsr/ [13] CFSv2 model http://www.cpc.ncep.noaa.gov/products/CFSv2/CFSv2seasonal.shtml [14] Cano D., Monget J.M., Albuisson M., Guillard H., Regas N., Wald L., 1986. A method for the determination of the global solar radiation from meteorological satellite data. Solar Energy, 37, 1, 31 –39. [15] Perez R., Ineichen P., Maxwell E., Seals R. and Zelenka A., 1992. Dynamic global-to-direct irradiance conversion models. ASHRAE Transactions-Research Series, pp. 354-369. [16] Perez, R., Seals R., Ineichen P., Stewart R., Menicucci D., 1987. A new simplified version of the Perez diffuse irradiance model for tilted surfaces. Solar Energy, 39, 221-232. [17] Ruiz-Arias J.A., Cebecauer T., Tovar-Pescador J., Šúri M., 2010. Spatial disaggregation of satellite-derived irradiance using a high-resolution digital elevation model. Solar Energy, 84,1644-57. http://www.sciencedirect.com/science/article/pii/S0038092X10002136 [18] Zelenka A., Perez R., Seals R., Renne D., 1997. Effective Accuracy of Satellite-Derived Hourly Irradiances. Theoretical Appl. Climatology, 62, 199-207. [19] Espinar B., Ramírez L., Drews A., Beyer H. G., Zarzalejo L. F., Polo J., Martín L., Analysis of different comparison parameters applied to solar radiation data from satellite and German radiometric stations. Solar Energy, 83, 1, 118-125, 2009. © 2018 Solargis page 91 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 [20] Lohmann S., Schillings C., Mayer B., Meyer R., 2006. Long-term variability of solar direct and global radiation derived from ISCCP data and comparison with reanalysis data, Solar Energy, 80, 11, 1390-1401. [21] Gueymard C., Solar resource assessment for CSP and CPV. Leonardo Energy webinar, 2010. http://www.leonardo-energy.org/webfm_send/4601 [22] Cebecauer T., Šúri M., 2015. Typical Meteorological Year Data: Solargis Approach. Energy Procedia, Volume 69, 1958-1969. http://dx.doi.org/10.1016/j.egypro.2015.03.195 [23] Meyer, R., Butron, J. T., Marquard, G., Schwandt, M., Geuder, N., Hoyer-Klick, C., et al. (2008). Combining solar irradiance measurements and various satellite-derived products to a site-specific best estimate. In 14th biennial SolarPACES Symposium. Retrieved from http://elib.dlr.de/53849/ [24] Jessen, W., Wilbert, S., Nouri, B., Geuder, N., & Fritz, H. (2016). Calibration methods for rotating shadowband irradiometers and optimizing the calibration duration. Atmospheric Measurement Techniques, 9(4), 1601– 1612. https://doi.org/10.5194/amt-9-1601-2016 [25] Wilbert, S., Röger, M., Csambor, J., Breitbach, M., Klinger, F., Nouri, B., et al. (2017). Sunshape Measurements with Conven-tional Rotating Shadowband Irradiometers. In Proc. of the SolarPACES Conference 2017 (p. 8). Santiago, Chile: SolarPACES [26] Lezaca, J., Meyer, R., & Heinemann, D. (2018). Towards an in-situ calibration of Rotating Shadowband Irradiometers (RSI) using simultaneous pyranometer measurements. In Proc. of the SolarPACES Conference 2017 (p. 10). Santiago, Chile: SolarPACES, reviewed and accepted for publication © 2018 Solargis page 92 of 94 Annual Solar Resource Report for Solar Meteorological Stations in Maldives after completion of 24 months of measurements Solargis reference No. 129-07/2018 SUPPORT INFORMATION Background on Solargis Solargis is a technology company offering energy-related meteorological data, software and consultancy services to solar energy. We support industry in the site qualification, planning, financing and operation of solar energy systems for more than 18 years. We develop and operate a new generation high-resolution global database and applications integrated within Solargis® information system. Accurate, standardised and validated data help to reduce the weather-related risks and costs in system planning, performance assessment, forecasting and management of distributed solar power. Solargis is ISO 9001:2015 certified company for quality management. This report has been prepared by Tomas Cebecauer, Marcel Suri, Branislav Schnierer, Nada Suriova, Juraj Betak, and Artur Skoczek from Solargis All maps in this report are prepared by Solargis Solargis s.r.o., Mytna 48, 811 07 Bratislava, Slovakia Reference No. (Solargis): 129-08/2018 http://solargis.com © 2018 Solargis page 93 of 94