SOLAR RESOURCE AND PV POTENTIAL OF MALAWI SOLAR RESOURCE ATLAS December 2018 This report was prepared by Solargis, under contract to the World Bank. Solar Power Resource Mapping: Malawi [Project ID: P151289]. 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. The content of this document is the sole responsibility of the consultant authors. 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 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. Rights and Permissions 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, 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. All images remain the sole property of their source and may not be used for any purpose without written permission from the source. Attribution Please cite the work as follows: World Bank. 2018. Solar resource and PV potential of Malawi: Solar Resource Atlas. Washington, DC: World Bank. Solar Resource Atlas Based on regional adaptation of Solargis model Republic of Malawi Reference No. 141-09/2018 Date: 7 December 2018 Customer Consultant World Bank Solargis s.r.o. Energy Sector Management Assistance Program Contact: Mr. Marcel Suri Contact: Mr. Dhruva Sahai Mytna 48, 811 07 Bratislava, Slovakia 1818 H St NW, Washington DC, 20433, USA Phone +421 2 4319 1708 Phone: +1-202-473-3159 E-mail: marcel.suri@solargis.com E-mail: mailto: dsahai@worldbank.org http://solargis.com http://www.esmap.org/RE_Mapping Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Table of contents Table of contents ................................................................................................................................................4 Acronyms ...........................................................................................................................................................5 Glossary .............................................................................................................................................................7 Executive summary.............................................................................................................................................9 1 Introduction................................................................................................................................................10 1.1 Past and on-going solar resource assessment initiatives ......................................................................... 10 1.2 Evaluation of the existing data and studies ................................................................................................ 12 1.3 Structure of this study .................................................................................................................................. 13 2 Methods and data.......................................................................................................................................14 2.1 Solar resource data ...................................................................................................................................... 14 2.2 Meteorological data...................................................................................................................................... 23 2.3 Simulation of solar photovoltaic potential .................................................................................................. 25 2.4 Outline of solar concentrating technologies ............................................................................................... 29 3 Solar resource and PV potential of Malawi .................................................................................................31 3.1 Geography ..................................................................................................................................................... 31 3.2 Air temperature ............................................................................................................................................. 39 3.3 Global Horizontal Irradiation ........................................................................................................................ 43 3.4 Direct Normal Irradiation .............................................................................................................................. 49 3.5 Global Tilted Irradiation ................................................................................................................................ 53 3.6 Photovoltaic power potential ....................................................................................................................... 58 3.7 Evaluation ...................................................................................................................................................... 62 4 Data delivered for Malawi ...........................................................................................................................63 4.1 Spatial data products ................................................................................................................................... 63 4.2 Project in QGIS format.................................................................................................................................. 67 4.3 Map images .................................................................................................................................................. 67 5 List of maps ...............................................................................................................................................69 6 List of figures .............................................................................................................................................70 7 List of tables ..............................................................................................................................................71 8 References .................................................................................................................................................72 Support information ..........................................................................................................................................74 © 2018 Solargis page 4 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Acronyms AC Alternating current AERONET The AERONET (AErosol RObotic NETwork) is a ground-based remote sensing network dedicated to measuring atmospheric aerosol properties. It provides a long-term database of aerosol optical, microphysical and radiative parameters. AOD Aerosol Optical Depth at 670 nm. This is one of atmospheric parameters derived from MACC database and used in Solargis. It has a notabe impact on the accuracy of solar calculations in arid zones. CFSR Climate Forecast System Reanalysis. The meteorological model operated by the US service NOAA. CFSv2 The Climate Forecast System Version 2. CFSv2 meteorological models operated by the US service NOAA (Operational extension of Climate Forecast System Reanalysis, CFSR). CPV Concentrated Photovoltaic systems, which uses optics such as lenses or curved mirrors to concentrate a large amount of sunlight onto a small area of photovoltaic cells to generate electricity. CSP Concentrated solar power systems, which use mirrors or lenses to concentrate a large amount of sunlight onto a small area, where it is converted to heat for a heat engine connected to an electrical power generator. DC Direct current 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. ECMWF European Centre for Medium-Range Weather Forecasts is independent intergovernmental organisation supported by 34 states, which provide operational medium- and extended-range forecasts and a computing facility for scientific research. ESMAP Energy Sector Management Assistance Program, a multi-donor trust fund administered by the World Bank EUMETSAT European Organisation for the Exploitation of Meteorological Satellites, an intergovernmental organisation for establishing, maintaining and exploiting European systems of operational meteorological satellites GFS Global Forecast System. The meteorological model operated by the US service NOAA. GHI Global Horizontal Irradiation, if integrated solar energy is assumed. Global Horizontal Irradiance, if solar power values are discussed. GIS Geographical Information System GTI Global Tilted (in-plane) Irradiation, if integrated solar energy is assumed. Global Tilted Irradiance, if solar power values are discussed. KSI Kolmogorov–Smirnov Index, a statistical index for comparing functions or samples © 2018 Solargis page 5 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 MACC Monitoring Atmospheric Composition and Climate – meteorological model operated by the European service ECMWF (European Centre for Medium-Range Weather Forecasts) Meteosat MFG Meteosat satellite operated by EUMETSAT organization. MSG: Meteosat Second Generation; and MSG MFG: Meteosat First Generation MERRA Modern-Era Retrospective Analysis for Research and Applications, a NASA reanalysis for the satellite era using an Earth observing systems MERRA-2 Modern-Era Retrospective analysis for Research and Applications, Version 2 NASA National Aeronautics and Space Administration organization NOAA NCEP National Oceanic and Atmospheric Administration, National Centre for Environmental Prediction NOAA ISD NOAA Integrated Surface Database with meteorological data measured by ground-based measurement stations NOCT The Nominal Operating Cell Temperature, is defined as the temperature reached by open circuited cells in a module under the defined conditions: Irradiance on cell surface = 800 W/m2, Air Temperature = 20°C, Wind Velocity = 1 m/s and mounted with open back side. PV Photovoltaic PVOUT Photovoltaic electricity output calculated from solar resource and air temperature time series. RSR Rotating Shadowband Radiometer SOLIS Solar Irradiance Scheme, Solar clear-sky model for converting meteorological satellite images into radiation data SRTM Shuttle Radar Topography Mission, a service collecting topographic data of Earth's land surfaces STC Standard Test Conditions, used for module performance rating to ensure the same measurement conditions: irradiance of 1,000 W/m², solar spectrum of AM 1.5 and module temperature at 25°C. TEMP Air Temperature at 2 metres UV Ultraviolet radiation © 2018 Solargis page 6 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Glossary AC power output Power output measured at the distribution grid at a connection point. of a PV power plant 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 considering the impact of cloud cover. Fixed-mounted modules Photovoltaic modules assembled on fixed bearing structure in a defined tilt to the horizontal plane and oriented in fixed azimuth. Frequency of data Period of aggregation of solar data that can be obtained from the Solargis database. (30-minute, hourly, daily, monthly, yearly) Installed DC capacity Total sum of nominal power (label values) of all modules installed on photovoltaic power plant. 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 a 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. PV electricity production AC power output of a PV power plant expressed as percentage part of installed DC capacity. Root Mean Square Represents spread of deviations given by random discrepancies between measured Deviation (RMSD) and modelled data and is calculated according to this formula: 6 * ∑7 (* +,-./0,1 − +31,4,1 5 = & *89 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 © 2018 Solargis page 7 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 discrepancy is natural - a satellite monitors a large area (of approx. 3 x 4 km), while a sensor sees only a 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/m2]. 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/m2 or kWh/m2]. Spatial grid resolution In digital cartography the term applies to the minimum size of the grid cell or in other words, minimum size of the pixels in the digital map. © 2018 Solargis page 8 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Executive summary This report presents results of the solar resource assessment and mapping activity undertaken by The World Bank in Malawi, as a part of a broader technical assistance project covering biomass, solar and wind mapping funded by the Energy Sector Management Assistance Program (ESMAP). The data used in this report is based on the Solargis model. The uncertainty of the solar resource data has been reduced by the regional model adaptation based on the ground measurements collected at three solar meteorological stations across Malawi, commissioned by The World Bank during the years 2016 to 2018 under the same activity. The ground-based solar resource measurements have been supplied by GeoSUN Africa, based in South Africa. The measurement campaign has been technically supported by SGS Malawi Limited, based in Blantyre, Malawi. The report has two objectives: • To explain the methodologies and outcomes of the solar resource and photovoltaic power potential assessment, based on the combined use of models and measured data. The report documents the uncertainty of solar and meteorological data as key inputs in the technical and financial evaluation of solar energy systems. • To improve the awareness and knowledge of resources for solar energy technologies by producing a comprehensive countrywide dataset and maps based on the accuracy-enhanced models. The report evaluates key solar climate features, and the geographic and time variability of solar power potential in the country. The results of this report compare to interim solar resource validation at the beginning of the project, which were based on the Solargis model, validated by the ground measurements available in a wider region (ESMAP Solar Resource Mapping for Malawi, Interim Solar Modelling Report, 141-01/2015, March 2015). The uncertainty estimates in this report have been found as too optimistic. The validation of the model based on 24 months of measurements conducted at three solar meteorological stations revealed higher model uncertainty. The uncertainty of the Solargis model yearly estimates for DNI, has been reduced from the findings for the original model ±12.0% to ±22.0% for yearly values to the range of ±5% and ±7% for the regionally adapted solar irradiation values. For yearly GHI, the uncertainty was reduced from the range ±9.0% to ±13% for original model to the range of ±4% and ±5% for the regionally adapted model. These figures represent a majority of the country’s territory with flat and monotonous terrain. In specific conditions with complex terrain we expect a higher model uncertainty. The key achievement of this project is supplying country-wide data and maps, based on the extensive validation of the solar model by high accuracy solar measurements acquired in Malawi. The data underlying this report are delivered in two formats: • Raster GIS data for the whole territory of the Republic of Malawi, representing long-term monthly and yearly average values. This data layers are accompanied by geographical data layers in raster and vector data formats. • High-resolution digital maps prepared for poster printing, as well as Google Earth maps. The maps show that, throughout most of Malawi, yearly sum of global horizontal irradiation is in the range of 1680 to 2050 kWh/m2. This translates to a specific yearly PV electricity output in the range of 1350 kWh/kWp to more than 1700 kWh/kWp. The seasonal variability is smaller, compared to other countries further away from the equator. This qualifies Malawi as a country with highly feasible potential for PV power generation. The aggregated data for Malawi can be accessed online via an interactive map-based application, or as downloadable files and maps at http://globalsolaratlas.info/. The ground-measured data is accessible through https://energydata.info/. © 2018 Solargis page 9 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 1 Introduction Solar electricity offers a unique opportunity to achieve long-term sustainability goals, such as the development of a modern economy, healthy and educated society, clean environment, and improved geopolitical stability. Solar power plants exploit local solar resources; they do not require heavy support infrastructure, they are scalable, and improve electricity services. A key feature of solar electricity is that it is accessible in remote locations, thus providing development opportunities anywhere. While solar resources are fuel to solar power plants, the local geographical and climate conditions determine the efficiency of their operation. Free fuel makes solar technology attractive; however, effective investment and technical decisions require detailed, accurate and validated solar and meteorological data. Accurate data are also needed for the cost-effective operation of solar power plant. High quality solar resource and meteorological data can be obtained by satellite-based meteorological models and by instruments installed at meteorological stations. 1.1 Past and on-going solar resource assessment initiatives Several solar resource assessment initiatives are documented below as publications and online data resources. The works show steadily growing interest and different stages of development of solar resource assessment and energy modelling in the region. Below we compare just several satellite-based databases out of more available. NASA Power Project Data Sets, NASA Monthly and yearly averages are available from the NASA Power Project Data Sets [1]. The data and documentation are updated in 2018. Specific parameters are available at higher time resolution (e.g. daily). The data includes numerous atmospheric and solar radiation parameters, the solar data represents a period from 1983 to 2005, resolution of approx. 55 km. Data is not validated for the region and it can be accessed from https://power.larc.nasa.gov/. Photovoltaic Geographical Information System (PVGIS), European Commission JRC Geographical Assessment of Solar Resource and Performance of Photovoltaic Technology. The online tools are accessible from http://re.jrc.ec.europa.eu/pvgis.html. The database is based on Meteosat satellite data calculation and offers solar resource long-term averages as well as hourly data. PVGIS HelioClim-1 is an older version of PVGIS based on HelioClim-1 product by Ecole des Mines, Paris, that makes use of interpolation of clear-sky index derived from low resolution Meteosat MFG satellite images and terrain shading. The PVGIS Helioclim-1 database has been validated at only very limited number of ground stations in Africa and the outputs are of lower accuracy [2]. PVGIS CM-SAF is more modern and more accurate version of satellite database which makes use of Meteosat MSG satellite images. CM-SAF data is primarily offered at hourly resolution, the accuracy is better, compared to HelioClim 1. Yet the data has also limited validation in Africa and no validation in this region. The most recent update of the project has been made in 2017 [3]. HelioClim 1 Project, Ecole des Mines Paris Online solar radiation satellite-derived database is available for free on http://www.soda-pro.com/web- services/radiation/helioclim-1. This database is one of first attempts in Europe to provide satellite-based solar radiation database covering Europe and Africa. HelioClim 1 uses reduced dataset of Meteosat MFG satellite images, © 2018 Solargis page 10 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 with temporal resolution from 3 hours and the cloud index is calculated via Heliosat-2 method. With the coverage period from 1985 to 2005, it represents daily values of solar radiations with a coarse resolution: 20 to 30km [4]. CAMS - JADE, European Union’s Earth observation programme CAMS (Copernicus Atmosphere Monitoring Service) solar radiation services provide detailed assessment of optical variables in the atmosphere. Implemented by ECMWF (European centre for Medium-range Weather Forecasts) as part of the Copernicus programme, https://atmosphere.copernicus.eu/, it covers Europe, Africa and adjacent territories between ±66° latitude and longitude and it processes Meteosat satellites (MFG and MSG) images with Heliosat-4 method to create the dataset of solar radiation components with JADE being the specific CAMS radiation service dataset over Africa [5]. CAMS solar radiations services are validated at BSRN network of ground stations, but without a reference station in the wide region of South-eastern Africa. The service is still in the phase of accuracy improvements. Meteonorm, Meteotest Meteonorm, https://meteonorm.com/en/ is a global meteorological database of ground stations around the world, with a support from satellite-based solar radiation for Europe and Africa available from CM-SAF data service. The measurements are used to interpolate the specific conditions from nearby stations for the site of interest and to calculate synthetic hourly data for one artificial year. This approach provides data with limited accuracy and use, and there are little prospects for meeting the needs of development and operation of commercial PV power plants. The accuracy of calculation database depends the density of good-quality solar meteo stations. In Africa however, the availability of high-quality solar meteorological stations (based on the use of high-accuracy sensors and adequate maintenance) is very limited, with no site available in Southeast Africa. Moreover, the micro-climate conditions of a specific site would be completely overlooked by spatial interpolation, which may result in large errors in the calculation output. Synthetic hourly data cannot be validated by high resolution ground measurements [6]. Renewable Energy Resource Mapping for Malawi, World Bank (ESMAP) This report refers to the outcomes achieved by this project, closed in 2018. A set of data and reports for Malawi has been prepared by Solargis and its subcontractor GeoSUN Africa, working on this project until the present. Three areas were addressed, in consecutive phases: • Preliminary modelling that has been conducted by Solargis • Installation, operation and data acquisition for three ground-based solar meteorological stations by GeoSUN Africa, South Africa supported by SGS Malawi. All the measured data is accessible via the portal https://energydata.info/ • This report refers to final Phase 3 of the project, and it accompanies the delivery of the final outputs based on the combination of the modelled and the measured data. Solargis provides the final mapping outputs for Malawi. All outputs are accessible from https://globalsolaratlas.info. Global Solar Atlas, World Bank Group The World Bank and the International Finance Corporation have provided the Global Solar Atlas to support the scale- up of solar power in their client countries. This work is funded by the Energy Sector Management Assistance Program (ESMAP), a multi-donor trust fund administered by The World Bank. The Atlas has been prepared by Solargis under a contract to The World Bank. The primary aim is to provide quick and easy access to solar resource data and maps globally [7, 8]. The project is ongoing, and regular updates are planned in the following years https://globalsolaratlas.info. © 2018 Solargis page 11 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 1.2 Evaluation of the existing data and studies Malawi has considerable though diverse potential for solar power generation. The previously developed solar and meteorological data sets (See Chapter 1.1) do not fulfil the requirements for accuracy, reliability and features needed for commercial development of solar PV power plants in present times. Table 1.1 compares Solargis results to the previous solar resource assessment initiatives. The main features that differ Solargis database from the above- mentioned data sets, include the following: • The Solargis models are based on new and advanced algorithms, validated at various climate zones • Use of modern and systematically updated input data for the models: satellite, atmospheric and meteorological • Database has global coverage at high resolution • Historical sub-hourly time series data is updated in real time • Data can be used for project development but also for monitoring and forecasting • Data is systematically validated and quality controlled • There is customer support and supporting consultancy services available The new database from Solargis focuses on a systematic supply of data and services for the development and financing of large-scale solar power plants worldwide, including Malawi. The main objective is to systematically supply reliable, validated and high-resolution data to the solar industry with low uncertainty and systematic quality control. The solar industry requires models that offer map-based data covering extensive territories at a high level of a detail using both historical and the most recent data. Modern solar measuring stations are used for accuracy enhancement of such models and to gain a better understanding of the local climate. Thus, a combination of the model data with modern solar and meteorological measurements is used to support solar energy development in all stages of its lifecycle. High accuracy solar resource and meteorological data are needed for the development and operation of commercial solar power plants. Typically, detailed data describing the local climate is needed for a location of interest; however, high accuracy meteorological measurements for sites of interest are rarely available. Therefore, data from solar and meteorological models are initially used to evaluate the energy yield and assess the performance of the power plants. When the location for commercial project development is selected, a solar meteorological station is installed as the second step. The high accuracy meteorological equipment is used to collect local data for an initial period of at least one year. Such measurements are then used for the site adaptation of solar models and for delivering high accuracy solar resource and meteorological time series that covers a long historical period. At larger power plants, solar measurements are collected over the lifetime of the project. The solar and meteorological data is used for the following tasks related to solar power generation: 1. Country-level evaluation and strategical assessment 2. Prospection; selection of candidate sites for future power plants, and prefeasibility analysis 3. Project evaluation; solar and energy yield assessment, technical design and financing 4. Monitoring and performance assessment of solar power plants and forecasting of solar power 5. Quality control of solar measurements. This report addresses the first topic from the list above. © 2018 Solargis page 12 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Table 1.1: Comparison of longterm GHI estimate: Solargis vs. previous studies Kasungu site (Lat: -13.01530, Lon: 33.46840) Source Reference Daily GHI Yearly GHI GHI difference Indicated Year of Data estimate estimate to validated uncertainty publication coverage (kWh/m2) (kWh/m2) Solargis of yearly (access) value NASA POWER [1] 5.51 2012 +1.2% ? 2018 1983 – 2005 PVGIS HelioClim-1 [2] 5.69 2078 +4.3% ? 2017 1998 – 2011 PVGIS CMSAF [3] 5.89 2151 +7.6% ? 2016 1985 – 2004 Helioclim-1 [4] 5.57 2034 +2.3% ? (2018) 1985 – 2005 CAMS- JADE [5] 5.85 2138 +7.0% ? 2018 1991 – 2010 Meteonorm 7.3 [6] 5.76 2103 +5.5% ±4.0% 2018 1991 – 2010 Solargis and [7] 5.67 2069 +3.9% ±6.0% 2017 1994 – 2016 Global Solar Atlas Solargis (regionally This report 5.44 1988 - ±4.0% 2018 1994 – 2017 adapted) 1.3 Structure of this study Following an introduction to the activity (Chapter 1), Chapter 2 of this Solar Resource Atlas provides an outline of solar radiation basics and principles of photovoltaic power potential calculation. Chapters 2.1 and 2.2 describe measuring and modelling approaches for developing reliable solar and meteorological data, including information about solar and meteorological data uncertainty. These chapters document the role of solar measurements in reducing the uncertainty of solar, meteorological and PV power potential data for the country. Chapter 2.3 and 2.4 explain the relevance of solar resource and meteorological information for the deployment of solar power technologies. An emphasis is given to photovoltaic (PV) technology, which has high potential for developing utility- scale projects close to larger load centres, as well as deployment of rooftop PV systems, off-grid, hybrid systems and mini-grids for community electrification. Chapter 3 presents an analysis and evaluation of meteorological and solar resource data in Malawi. Three representative sites are selected to show potential regional differences in Malawi through tables and graphs. Chapter 3.1 introduces ancillary geographical data that influence the performance of solar power plants. Chapter 3.2 to 3.5 summarizes geographical differences and seasonal variability of the solar resource in Malawi, while Chapter 3.6 presents the PV power generation potential of the country. The theoretical specific PV electricity output is calculated from the most commonly used PV technology: a fixed system with crystalline-silicon (c-Si) PV modules, optimally tilted and oriented towards the Equator. Chapter 3.7 summarizes the analytical information and presents conclusions. Chapter 4 summarizes the technical features of the delivered data products. Chapters 5 to 8 provide support information. © 2018 Solargis page 13 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 2 Methods and data 2.1 Solar resource data 2.1.1 Introduction Solar resource (physical term solar radiation) is fuel to solar energy systems. The solar radiation available for solar energy systems at the ground level depends on processes in the atmosphere. This leads to a high spatial and temporal variability at the Earth’s surface. The interactions of extra-terrestrial solar radiation with the Earth’s atmosphere, surface and objects are divided into three groups: 1. Solar geometry, trajectory around the sun and Earth's rotation (declination, latitude, solar angle) 2. Atmospheric attenuation (scattering and absorption) by: 2.1 Atmospheric gases (air molecules, ozone, NO2, CO2 and O2) 2.2 Solid and liquid particles (aerosols) and water vapour 2.3 Clouds (condensed water or ice crystals) 3. Topography (elevation, surface inclination and orientation, horizon) 4. Shadows, reflections from surface or local obstacles (trees, buildings, etc.) and re-diffusion by atmosphere. The atmosphere attenuates solar radiation selectively: some wavelengths are associated with high attenuation (e.g. UV) and others with a good transmission. Solar radiation called "short wavelength" (in practice, 300 to 4000 nm) is of primary interest to solar power technology and is used as a reference. The component that is neither reflected nor scattered, and which directly reaches the surface, is called direct radiation; this is the component that produces shadows. The component scattered by the atmosphere that also reaches the ground is called diffuse radiation. A small portion of the radiation reflected by the surface that reaches an inclined plane is called the reflected radiation. These three components together create global radiation. A proportion of individual component at any time is given by Sun position and by the actual state of atmosphere – mainly the occurrence of clouds, air pollution and humidity. According to the generally adopted terminology, in solar radiation two terms are distinguished: • Solar irradiance indicates power (instant energy) per second incident on a surface of 1 m2 (unit: W/ m2). • Solar irradiation, expressed in MJ/ m2 or Wh/m2; it indicates the amount of incident solar energy per unit area during a lapse of time (hour, day, month, etc.). Often, the term irradiance is used by the authors of numerous publications in both cases, which can sometimes cause confusion. In solar energy applications, the following three solar resources are relevant: • Direct Normal Irradiation/Irradiance (DNI): it is the direct solar radiation from the solar disk and the region closest to the sun (circumsolar disk of 5° centred on the sun). DNI is the component that is important to concentrating solar collectors used in Concentrating Solar Power (CSP) and high-performance cells in Concentrating Photovoltaic (CPV) technologies. • Global Horizontal Irradiation/Irradiance (GHI): sum of direct and diffuse radiation received on a horizontal plane. GHI is a reference radiation for the comparison of climatic zones; it is also the essential parameter for calculation of radiation on a flat plate collector. • Global Tilted Irradiation/Irradiance (GTI) or total radiation received on a surface with defined tilt and azimuth, fixed or sun-tracking. This is the sum of the scattered radiation, direct and reflected. A term Plane of Array (POA) irradiation//irradiance is also used. In the case of photovoltaic (PV) applications, GTI can occasionally be affected © 2018 Solargis page 14 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 by shading from the surrounding terrain or objects, and GTI is then composed only from diffuse and reflected components. This typically occurs for sun at low angles over the horizon. Solar radiation data can be acquired by two complementary approaches: 1. Ground-mounted sensors are good in providing high frequency and accurate data (for well-maintained, high accuracy measuring equipment) for a specific location. 2. Satellite-based models provide data with a lower frequency of measurement, but cover a long history over lager areas. Satellite-models are not capable of producing instantaneous values at the same accuracy as ground sensors, but can provide robust aggregated values. Chapter 2 summarizes approaches applied for measuring and computing solar resource parameters, for Malawi, and the main sources of uncertainty. It also discusses methods for combining data acquired by these two complementary approaches with the aim of maximizing strengths of both approaches. 2.1.2 ESMAP Solar resource measurements in Malawi Data from the three ESMAP measuring stations in Malawi was collected and harmonized with the objective of acquiring reference solar radiation data for reducing the uncertainty of the model. The quality data from these meteorological stations are available for this assessment (Tables 2.1 and 2.2, Figure 2.1, Map 2.1). Position and detailed information about the measurement sites is also available in Global Solar Atlas website, http://globalsolaratlas.info/. More detailed information related to the measurement campaign in Malawi can be found in the report “Annual solar resource report for solar meteorological stations after completion of 24 months of measurements”, Ref. Nr. 141- 07/2018 (December 2018) [12]. The report presents quality control of ground measured data and results of site adaptation of the Solargis model for three solar meteorological sites, with estimate of relevant data uncertainties. Table 2.1: Overview information on measurement stations operated in the region No. Site name Nearest town Latitude Longitude Altitude Measurement station host [º] [º] [m a.s.l.] 1 Chileka airport Blantyre -15.67984 34.97229 767 Malawi Meteorological Services 2 Kasungu airport Kasungu -13.01530 33.46840 1065 Malawi Meteorological Services 3 Uni Mzuzu Mzuzu -11.41990 33.99530 1285 Mzuzu University Year, month 2016 2017 2018 Station 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 Chileka Kasungu Mzuzu Figure 2.1: Solar resource data availability (GHI, DNI and DIF). © 2018 Solargis page 15 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Table 2.2: Overview information on solar meteorological stations operating in the region No. Site name Type Parameters Time step Period of data used in study 1 Chileka TIER1 GHI, DNI, DIF 1 min 19 March 2016 – 31 March 2018 2 Kasungu TIER2 GHI, GHI2, DNI2, DIF2 1 min 18 March 2016 – 31 March 2018 3 Mzuzu TIER2 GHI, GHI2, DNI2, DIF2 1 min 18 March 2016 – 31 March 2018 Map 2.1: Position of the solar meteorological stations used for the model validation © 2018 Solargis page 16 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 2.1.3 Solargis satellite-based model Models using satellite and atmospheric data have become a standard for calculating solar resource time series and maps. The same models are also used for real-time data delivery for system monitoring and solar resource forecasting. Data from reliable operational solar models are routinely used by the solar industry. In this study, we applied a model developed and operated by the company Solargis. This model operationally calculates high-resolution solar resource data and other meteorological parameters. Its geographical extent covers most of the land surface between 60º North and 45º South latitudes. A comprehensive overview of the Solargis model was made available in several publications [9, 10, 11]. The related uncertainty and requirements for bankability are discussed in [12, 13, 14]. In the Solargis approach, solar irradiance is calculated in 5 steps: 1. Calculation of clear-sky irradiance, assuming all atmospheric effects except clouds, 2. Calculation of cloud properties from satellite data, 3. Integration of clear-sky irradiance and cloud effects and calculation of global horizontal irradiance (GHI) 4. Calculation of direct normal irradiance (DNI) from GHI and clear-sky irradiance. 5. Calculation of global tilted irradiance (GTI) from GHI and DNI. Models used in individual calculation steps are parameterized by a set of inputs characterizing the cloud properties, state of the atmosphere and terrain conditions. The clear-sky irradiance is calculated by the simplified SOLIS model [15]. This model allows the fast calculation of clear-sky irradiance from the set of input parameters. Sun position is a 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 core data set, representing a period from 2003 to the present, is from the MACC-II/CAMS project (ECMWF) [16, 17]. An important feature of this data set is that it captures daily variability of aerosols and allows simulating more precisely the events with extreme atmospheric load of aerosol particles. Thus it reduces uncertainty of instantaneous estimates of GHI and especially DNI, and it allows for improved statistical distribution of irradiance values [18, 19]. For years 1994 to 2002, data from the MERRA-2 model (NASA) [20] is used and it is homogenized with MACC- II/CAMS model are used. The Solargis calculation accuracy of the clear-sky irradiance is especially sensitive to information on aerosols. • 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 daily GFS and CFSR values (NOAA NCEP) are used in Solargis, thus representing the daily variability from 1994 to the present [21, 22, 23]. • 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. The effect of clouds is calculated from satellite data in the form of the cloud index (cloud transmittance). The cloud index is derived by relating irradiance recorded by the satellite in several spectral channels and surface albedo to the cloud optical properties. In this study, a data from the Meteosat MFG and MSG satellites is used. Data is available for a period from 1994 to the present (24-hour delay) in a time step of 30 and 15 minutes. In Solargis, the modified calculation scheme by Cano has been adopted to retrieve cloud optical properties from the satellite data [25]. A number of improvements have been introduced to better cope with specific situations such as snow, ice, or high albedo areas (arid zones and deserts), and complex terrain. © 2018 Solargis page 17 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 To calculate Global Horizontal Irradiance (GHI) for all atmospheric and cloud conditions, the clear-sky global horizontal irradiance is coupled with the cloud index. From GHI, other solar irradiance components (direct, diffuse and reflected) are calculated. Direct Normal Irradiance (DNI) is calculated by the modified Dirindex model [26]. Diffuse horizontal irradiance is derived from GHI and DNI according to the following equation: DIF = GHI - DNI * Cos Z (1) Where Z is the zenith angle between the solar position and the Earth’s surface. Calculation of Global Tilted Irradiance (GTI) from GHI deals with direct and diffuse components separately. While calculation of the direct component is straightforward, estimation of diffuse irradiance for a tilted surface is more complex, and is affected by limited information regarding shading effects and albedo of nearby objects. For converting diffuse horizontal irradiance for a tilted surface, the Perez diffuse transposition model is used [27]. The reflected component is also approximated considering that knowledge of local conditions is limited. A model for the simulation of terrain effects (elevation and shading) based on high-resolution elevation and horizon data is used in the standard Solargis methodology [28]. The model by Ruiz Arias is used to achieve enhanced spatial representation – from the resolution of satellite (several km) to the resolution of the digital terrain model. Solargis model version 2.1 has been used for computing the data. Table 2.3 summarize technical parameters of the model inputs and of the primary outputs. Table 2.3: Input data for Solargis solar radiation model and related GHI and DNI outputs for Malawi Inputs into the Solargis Source Time representation Original Approx. grid model of input data time step resolution Cloud index Meteosat MFG and 1994 to 2004 30 minutes 2.8 x 3.3 km MSG satellites 2005 to date 15 minutes 3.3 x 4.0 km (EUMETSAT) Atmospheric optical depth MACC/CAMS* 2003 to date 3 hours 75 km and 125 km (aerosols)* (ECMWF) MERRA-2 (NASA) 1994 to 2002 1 hour 50 km Water vapour CFSR/GFS 1994 to date 1 hour 35 and 55 km (NOAA) Elevation and horizon SRTM-3 - - 250 m (SRTM) Solargis primary data - 1994 to date 15 minutes 250 m outputs (GHI and DNI) 2.1.4 Measured vs. satellite data – adaptation of solar model For a qualified solar resource assessment, it is important to understand the characteristics of ground measurements and satellite-modelled data (Table 2.4). The ground measurements and satellite data complement each other, and it is beneficial to correlate them and adapt the satellite model for the specific site or region. Within this project, regional model adaptation has been performed using the data from three measuring stations (Table 2.1, Map 2.1). The model adapted for regional conditions provides long history solar resource time series as well as recent data with lower uncertainty. The model adaptation procedures has two steps: 1. Identification of systematic differences between hourly satellite data and local measurements for the period when both data sets overlap; © 2018 Solargis page 18 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 2. Development of a correction method that is applied for the whole period represented by the satellite time series over the whole region. In the case of regional adaptation, the method aims to identify and reduce regional systematic deviations of a model compared to the measured data, typically driven by the insufficient characterization of aerosols or specific cloud patterns. The result of regional adaptation is an improved solar resource database in the regional context with overall reduction of systematic errors. The regional-adaptation of satellite-based model data was performed for the whole territory of Malawi and the methodology and results are described in the report “Solar Model Validation Report; Regional adaptation of Solargis model based on data acquired in 24-months solar measurement campaign; Republic of Malawi”, Ref. Nr. 141-08/2018 [29]. The regional-adaptation improves knowledge about uncertainty of the model in specific conditions of Malawi, and more generally in tropical regions, where the Solargis model shows higher uncertainty. The new knowledge developed from the analysis of ground measurements collected during the project creates an important base for further model developments and improvements. Table 2.4: Comparing solar data from solar measuring stations and from satellite models Data from solar measuring stations Data from satellite-based models (Solargis) Availability/ Available only for limited number of sites. Mostly, Data are available for any location within latitudes accessibility data covers only recent years. 60º N and 45º S. Data covers long period, in Malawi, historical data for more than 24 years. Original spatial Data represent the microclimate of a site. Satellite models represent area with complex spatial resolution resolution: clouds are mapped at approx. 3 km, aerosols at 50-125 km and water vapour at 34 km. Terrain can be modelled at spatial resolution of up to 250 m. Methods for enhancement of spatial resolution are often used. Original time Seconds to minutes 15 and 30 minutes in Africa resolution Quality Data need to go through rigorous quality control, gap Quality control of the input data is necessary. filling and cross-comparison. Outputs are regularly validated. Under normal operation, the data have only few gaps, which are filled by intelligent algorithms. Stability Instruments need regular cleaning and control. If data are geometrically and radiometrically pre- Instruments, measuring practices, maintenance and processed, a complete history of data can be calibration may change over time. Thus, regular calculated with one single set of algorithms. Data calibration is needed. Long-term stability is typically computed by an operational satellite model may a challenge. change slightly over time, as the model and its input data evolve. Thus, regular reanalysis and temporal harmonization of inputs is used in state-of-the-art models. Uncertainty Uncertainty is related to the accuracy of the Uncertainty is given by the characteristics of the instruments, maintenance and operation of the model, resolution and accuracy of the input data. equipment, measurement practices, and quality Uncertainty of models is higher than high quality control. local measurements. The data may not exactly represent the local microclimate, but are usually stable and may show systematic deviation, which can be reduced by good quality local measurements (regional adaptation or site adaptation of the model). © 2018 Solargis page 19 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 2.1.5 Validation and regional adaptation of Solargis model Regional model adaptation has been performed in order to reduce overall model uncertainty in the region. Tables 2.5 and 2.6 show the Solargis model quality indicators for solar primary parameters, DNI and GHI, after the regional model adaptation. The uncertainty is evaluated for the version that has been regionally adapted. All information shown in this report is derived from the regionally adapted Solargis model. Comparison of the validation statistics, computed for the solar meteorological sites in Malawi, shows overall stability of the Solargis model and of the underlying input data. Locally, an increase of bias was identified, which is the effect of local geographical conditions, limited accuracy of the model and its input data, as well as the properties of ground measurements. The statistics show that the model uncertainty has been reduced after the regional adaptation. The results of the regional model adaptation are comparable to those achieved in other regions [30, 31]. Table 2.5: Direct Normal Irradiance: bias before and after regional model adaptation DNI after regional Meteo station Original DNI data adaptation Bias Bias Bias Bias 2 2 [kWh/m ] [%] [kWh/m ] [%] Chileka 42 10.5 3 0.7 Kasungu 41 9.6 -2 -0.4 Mzuzu 78 20.5 7 1.8 Mean 54 13.5 3 0.7 Standard deviation 21 6.1 4 1.1 Table 2.6: Global Horizontal Irradiance: bias before and after regional model adaptation GHI after regional Meteo station Original GHI data adaptation Bias Bias Bias Bias [kWh/m2] [%] [kWh/m2] [%] Chileka 38 8.3 4 0.9 Kasungu 27 5.4 5 0.9 Mzuzu 59 12.7 16 3.5 Mean 41 8.8 8 1.8 Standard deviation 16 3.7 7 1.5 © 2018 Solargis page 20 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 2.1.6 Uncertainty of solar resource estimates The uncertainty of regionally adapted satellite-based DNI and GHI is determined by uncertainty of the model, ground measurements, and the model adaptation method. More specifically it depends on [15]: 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 Meteosat satellite data, MACC-II/CAMS and MERRA-2 aerosols and CFSR/GFS water vapour • Solis clear-sky model and its capability to properly characterize various states of the atmosphere • Simulation accuracy of the Solargis cloud transmittance algorithms, being able to properly distinguish different states of various surface types, albedo, clouds and fog • Diffuse and direct decomposition by Perez model • Transposition from global horizontal to in-plane irradiance (for GTI) by Perez model • Terrain shading and disaggregation by Ruiz-Arias model 2. Uncertainty of the ground-measurements, which is determined by: • Accuracy of the instruments • Maintenance practices, including sensor cleaning, service and calibration • Data post-processing and quality control procedures. 3. Uncertainty of the model adaptation at regional scale and residual uncertainty after adaptation The uncertainty from the interannual variability of solar resource is not considered in this study. Accuracy statistics, such as bias (Chapter 2.1.5) characterize the accuracy of the Solargis model in the given validation points, relative to the ground measurements. The validation statistics are affected by local geography and by the quality and reliability of ground-measured data. Therefore, the validation statistics only indicate performance of the model in this region. The majority of Malawi territory has uncertainty of the regionally-adapted model yearly values in the range of ±4% to ±5% for GHI and ±5% to ±7% for DNI. We expect higher uncertainty in regions with more complex geography (Map 2.2), up to ±7% for GHI and up to ±10% for DNI. This is partly a result of uncertainty of ground measurements, limited number of solar meteorological stations and higher model uncertainty in regions with specific micro-climatic conditions (e.g. convective clouds close to steep slopes surrounding the Lake Malawi). Table 2.7: Uncertainty of the model estimate for original and regionally-adapted annual GHI, DNI and GTI and how does it compare to the best-achievable uncertainty case. Direct Normal Irradiation Global Horizontal Irradiation Global Tilted Irradiation Low Medium Low Medium Low Medium Original data < ±12% < ±21% < ±9% < ±13% < ±9% < ±13% After ±5% to ±7% < ±10% ±4% to ±5% < ±6% ±4.5% to ±5% < ±7% adaptation Best- ±3.5% - ±2.5% - ±3.0% achievable* The lowest (best achievable) uncertainty in Table 2.7 can only be achieved by model site-adaptation so that it represents only the very local microclimate of the site recorded in the ground measurements. In the case of the regional adaptation, used in this study, the uncertainty is usually higher because it describes uncertainty of any location in a wider region. © 2018 Solargis page 21 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Moreover, a residual discrepancy between ground measurements, and the model data can be found after regional adaptation (Tables 2.5 and 2.6). This model adaptation approach is designed to correct only regional discrepancy patterns, not to resolve site-specific issues. Map 2.2: Geographic distribution of the model uncertainty in Malawi L: low; M: medium © 2018 Solargis page 22 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 2.2 Meteorological data 2.2.1 Measured vs. modelled data Meteorological parameters are an important part of a solar energy project assessment as they determine the operating conditions and the effectiveness of solar power plant operations. The most important meteorological parameter for the operation of photovoltaic power plants is air temperature, which directly impacts power production (energy yield is decreasing when temperature is increasing). Meteorological data can be collected by two approaches: (1) by measuring at meteorological sites and (2) computing by meteorological models. The requirements for the meteorological data for solar energy projects are: • Long and continuous record of data, preferably covering the same time period as satellite-based solar resource data, • Data should reliably represent the local climate, • Data should be accurate, quality-controlled and without gaps. Table 2.8: Comparing data from meteorological stations and weather models Meteorological station data Data from meteorological models Availability/ Available only for selected sites. Data are available for any location accessibility Data may cover various periods of time Data cover long period of time (decades) Original spatial Local measurement representing Regional simulation, representing regional weather patterns resolution microclimate with all local weather with relatively coarse grid resolution. Therefore the local occurrences values may be smoothed, especially extreme values. Original time From 1 minute to 1 hour 1 hour resolution Quality Data needs to go through rigorous quality No need of special quality control. No gaps, control, gap filling and cross-comparison. relatively stable outputs if data processing systematically controlled. Stability Sensors, measuring practices, maintenance In case of reanalysis, long history of data is calculated with and calibration may change over time. Thus, one single stable model. long-term stability is often a challenge. Data for operational forecast model may slightly change over time, as model development evolves Uncertainty Uncertainty is related to the quality and Uncertainty is given by the resolution and accuracy of the maintenance of sensors and measurement model. Uncertainty of meteorological models is higher than practices, usually sufficient for solar energy high quality local measurements. The data may not exactly applications. represent the local microclimate, but are usually sufficient for solar energy applications. Several models are available: a good option is to use Modern-Era Retrospective analysis for Research and Applications (MERRA-2) model (source NASA, USA) [23] and the Climate Forecast System Version 2 (CFSv2) model (source NOAA, NCEP, USA), which cover a long period of time with continuous data [24]. The results of these models are implemented in Solargis. The role of meteorological ground measurements in solar energy development has two aspects: • Measurements are used for the validation and accuracy enhancement of historical data derived from the solar and meteorological models • The high frequency measurements (typically one-minute data) are used for improved understanding of the microclimate of the site, especially for capturing the extremes. Data from the two sources described above have their advantages and disadvantages (Table 2.8). Air temperature derived from the meteorological models has lower spatial and temporal resolution compared to ground © 2018 Solargis page 23 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 measurements, and lower accuracy. Thus, the modelled parameters characterize regional climate patterns rather than the local microclimate. Extreme values, in particular, may not be well represented. 2.2.2 Method and validation In this delivery, the air temperature data is derived from the meteorological models: MERRA-2 and CFSv2 (Table 2.9). It is important to note that the numerical weather models have lower spatial and temporal resolution compared to the solar resource data. The original spatial resolution of the models is enhanced to 1 km for air temperature by spatial disaggregation and the use of the Digital Elevation Model SRTM-3. Table 2.9: Original source of Solargis air temperature at 2 m for Malawi: MERRA-2 and CFSv2. Modern-Era Retrospective analysis for Climate Forecast System Research and Applications (MERRA-2) (CFSv2) Time period 1994 to 2010 2011 to the present time Original spatial resolution 45 x 50 km 19 x 22 km Original time resolution 1 hour 1 hour For the purpose of validating the meteorological models in Malawi, we have used the data collected at three meteorological stations (Table 2.1, Map. 2.1). The summary of basic statistical parameters is presented in Table 2.10. The main issue identified is underestimation of night-time temperature by the model, yet the day-time temperature is represented with higher accuracy. This issue was partially mitigated by site adaptation of the temperature model, see the report “Annual Solar Resource Report for solar meteorological stations after completion of 24 months of measurements, Republic of Malawi, Report number: 141-07/2018” [8]. This site-specific correction was not applied to the map data discussed in this report. Table 2.10: Air temperature at 2 m: accuracy indicators of the model outputs [ºC]. Meteorological Validation period Bias mean RMSD RMSD RMSD station hourly daily monthly Chileka 03/2016 – 03/2018 -1.8 3.1 2.3 2.0 Kasungu 03/2016 – 03/2018 -1.7 2.5 2.0 1.7 Mzuzu 03/2016 – 03/2018 -1.7 2.3 1.9 1.7 2.2.3 Uncertainty of air temperature In general, the data from the meteorological models represent larger area, and it is not capable to represent accurately the microclimate. The main issue identified is underestimation of night-time temperature by the model, yet the day-time temperature is represented with higher accuracy than nigh-time. The uncertainty of the model estimate for air temperature is summarised in Table 2.11. Table 2.11: Expected uncertainty of air temperature in Malawi. © 2018 Solargis page 24 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Unit Annual Monthly Hourly Air temperature at 2 m °C ±2.5 ±2.5 ±3.5 2.3 Simulation of solar photovoltaic potential Solar radiation is the most important parameter for PV power simulation, as it is fuel for solar power plants. The intensity of global irradiance received by the tilted surface of PV modules (GTI) is calculated from two primary parameters stored in the Solargis database and delivered in this project: • Global Horizontal Irradiance (GHI) • Direct Normal Irradiance (DNI) There are two main types of solar energy technologies: photovoltaic (PV) and concentrating solar power (CSP). Photovoltaics have high potential in Malawi, and this technology is discussed in this Chapter. CSP technology is not expected to be implemented in Malawi. Photovoltaic technology exploits global horizontal or tilted irradiation, which is the sum of direct and diffuse components (see equation (1) in Chapter 2.1.3). To simulate power production from a PV system, global irradiance received by a flat surface of PV modules must be calculated. Due to clouds, PV power generation reacts to changes in solar radiation in a matter of seconds or minutes (depending on the size of a module field), thus intermittency (short-term variability) of the PV power production is to be considered. Similarly, the effect of seasonal variability is to be considered as well. For possible PV installations, several technical options are available. They are briefly described below. Two types of mounting of PV modules can be considered: • PV modules mounted on the ground in a fixed position or on sun-trackers • PV modules mounted on rooftops or façades of buildings Three types of PV systems can be considered for Malawi: • Grid-connected PV power plants • Mini-grid PV systems • Off-grid PV systems The majority of larger PV power plants are built in an open space and often these have PV modules mounted at a fixed position. Fixed mounting structures offer a simple and efficient choice for implementing PV power plants. A well-designed structure is robust and ensures long-life performance, even during harsh weather conditions, at low maintenance costs. Sun-tracking systems are another alternative for the design of a PV module field. Solar trackers adjust the orientation of the PV modules during the day towards the sun, so the PV modules collect more solar radiation. Roof or façade mounted PV systems are typically small to medium size, i.e. ranging from hundreds of watts to hundreds of kilowatts. Modules can be mounted on rooftops, façades or can be directly integrated as part of a building structure. PV modules in this type of system are often installed in a suboptimal position (deviating from the optimum angle), and this results in a lower performance efficiency. Some reduction of PV power output can be expected due to nearby shading structures. Trees, masts, neighbouring buildings, roof structures or self-shading of PV modules determine the reduced PV system performance. Mini-grid PV systems include power generation facility and distribution grids connecting the local consumers. The typical size of installed PV systems is in the range of 10s to 100s of kWp. Mini-grids may be adapted to meet the © 2018 Solargis page 25 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 requirements of local needs, they can be combined with diesel generators and battery storage. This option appears to be most economic for supply of electricity for small rural communities. Off-grid PV systems are small systems that are not connected into a distribution grid. They are usually equipped with energy storage (classic lead acid or modern-type batteries, such as Li-on) and/or connected to diesel generators. Batteries are maintained through charge controllers for protection against overcharging or deep discharge. Depending on size and functionality of the off-grid PV system, it can work with AC (together with inverter) or DC voltage source. In this study, the PV power potential is studied for a system with fixed-mounted PV modules, considered here as the mainstream technology. Installed capacity of a PV power plant is usually determined by the available space and options to maintain the stability of the local grid. 2.3.1 Principles of PV electricity simulation Results of PV electricity simulation, presented in Chapter 3.6, are based on software developed by Solargis. This Chapter summarizes key elements of the simulation chain. Table 2.12: Specification of Solargis database used in the PV calculation in this study Data inputs for PV simulation Global tilted irradiation (GTI) derived from GHI and DNI Air temperature at 2 m (TEMP) Spatial grid resolution (approximate) 250 m (9 arc-sec) Time resolution 15-minute Geographical extent (this study) Republic of Malawi Period covered by data (this study) 01/1994 to 12/2017 The PV software implemented by Solargis has scientifically proven methods [32 to 37] and uses full historical time series of solar radiation and air temperature data on the input (Table 2.12). Data and model quality are checked using field tests and ground measurements. In PV energy simulation procedure, there are several energy losses that occur in the individual steps of energy conversion (Figure 2.2): 1. Losses due to terrain shading caused by far horizon. On the other hand, shading of local features such as nearby building, structures or vegetation is not considered in the calculation, 2. Energy conversion in PV modules is reduced by losses due to angular reflectivity, which depends on the relative position of the sun and plane of the module and temperature losses, caused by the performance of PV modules working outside of STC conditions defined in datasheets, 3. DC output of PV array is further reduced by losses due to dirt or soiling depending mainly on the environmental factors and module cleaning, losses by inter-row shading caused by preceding rows of modules, mismatch and DC cabling losses, which are caused by slight differences between the nominal power of each module and small losses on cable connections, 4. DC to AC energy conversion is performed by an inverter. The efficiency of this conversion step is reduced by inverter losses, given by the inverter efficiency function. Further factors reducing AC energy output are losses in AC cabling and transformer losses (applies only to large-scale open space systems), © 2018 Solargis page 26 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 5. Availability. This empirical parameter quantifies electricity losses incurred by the shutdown of a PV power plant due to maintenance or failures, including issues in the power grid. Availability of well operated PV systems is approximately 99%. According to experience in many countries, the crystalline silicon PV modules show a relatively low performance reduction over time. The rate of the performance degradation is higher at the beginning of exposure, and then stabilizes at a lower level. Initial degradation may be close to the value of 0.8% for the first year and 0.5% or less for subsequent years [37]. The performance ageing of PV modules is not considered in this study. The calculation results of PV power potential for Malawi are shown in Chapter 3.6. Figure 2.2: Simplified Solargis PV simulation chain 2.3.2 Technical configuration of a reference PV system Theoretical photovoltaic power production in Malawi has been calculated using numerical models developed and implemented in-house by Solargis. As introduced in Chapter 2.1, 15-minute time series of solar radiation and air temperature, representing last 24 years, are used as an input to the simulation. The models are developed based on the advanced algorithms, expert knowledge and recommendations given in [38] and tested using monitoring results from existing PV power plants. Table 2.14 summarizes losses and related uncertainty throughout the PV computing chain. In this study, the reference configuration for the PV potential calculation is a PV system with crystalline-silicon (c-Si) modules mounted in a fixed position on a table facing North and inclined at an angle close to optimum, i.e. at the angle at which the yearly sum of global tilted irradiation received by PV modules is maximized (a range between 9º and 19º depending on latitude and geographical features). The fixed-mounting of PV modules is very common and provides a robust solution with minimal maintenance effort. Geographic differences in potential PV production are shown for three selected sites (Chapter 3.6). Map 3.16 shows theoretical potential power production of a PV system installed with a typical technology configuration for open space PV power plants. The technical parameters are described in Table 2.13. © 2018 Solargis page 27 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Table 2.13: Reference configuration - photovoltaic power plant with fixed-mounted PV modules Feature Description Nominal capacity Configuration represents a typical PV power plant of 1 MWp or higher. All calculations are scaled to 1 kWp, so that they can be easily multiplied for any installed capacity. Modules Crystalline silicon modules with positive power tolerance. Nominal Operating Cell Temperature (NOCT) 46ºC and temperature coefficient of the Pmax -0.438 %/K Inverters Central inverter with declared datasheet efficiency (Euro efficiency) 97.5% Mounting of PV modules Fixed mounting structures facing North with optimum tilt (the range from 9º to 19º). Relative row spacing 2.5 (ratio of absolute spacing and table width) Transformer Medium voltage power transformer Table 2.14: Yearly energy losses and related uncertainty in PV power simulation Simulation step Losses Uncertainty Notes [%] [± %] 1 Global Tilted Irradiation N/A 5.0 Annual Global Irradiation falling on the surface of PV (model estimate with terrain modules shading) 2 Module surface angular -2.5 to -2.9 1.0 Slightly polluted surface is assumed in the calculation of the reflectivity (numerical model) module surface reflectivity Conversion in modules relative -7.1 to -13.1 3.5 Depends on the temperature and irradiance. NOCT of 46ºC is to STC (numerical model) considered 3 Polluted surface of modules -4.0 1.5 Losses due to dirt, dust, soiling, and bird droppings (empirical estimate) Power tolerance (value from 0.0 0.0 Value given in the module technical data sheet (modules with the data sheet) positive power tolerance) Module inter-row shading -0.1 to -0.5 0.5 Partial shading of strings by modules from adjacent rows (model estimate) Mismatch between modules -0.5 0.5 Well-sorted modules and lower mismatch are considered. (empirical estimate) DC cable losses -2.0 1.5 This value can be calculated from the electrical design (empirical estimate) 4 Conversion in the inverter -2.5 0.5 Given by the Euro efficiency of the inverter, which is (value from the technical data considered at 97.5% sheet) AC cable losses -0.5 0.5 Standard AC connection is assumed (empirical estimate) Transformer losses -1.0 0.5 Standard transformer is assumed (empirical estimate) 5 Availability 0.0 1.5 100% technical availability is considered; the uncertainty considered here assumes a well-integrated system; the real value strongly depends on the efficiency of PV integration into the existing grid. Range of cumulative losses -18.6 to -24.5 6.8 These values are indicative and do not consider project and indicative uncertainty specific features and performance degradation of a PV system over its lifetime © 2018 Solargis page 28 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 The results presented in Chapter 3.6 do not consider the performance degradation of PV modules due to aging; they also lack the required level of detail. Thus, these results cannot be used for financial assumptions of any specific project. Detailed assessment of energy yield for a specific power plant is within the scope of a site-specific bankable expert study. PV electricity potential is calculated based on a set of assumptions shown in Tables 2.13 and 2.14. These assumptions are approximate values, and they will differ in the site-specific projects. As can be seen, the uncertainty of the solar resource is the key element of energy simulation. 2.4 Outline of solar concentrating technologies Concentrating technologies can only utilize DNI (as diffuse irradiance cannot be concentrated). Instant (short-term) variability of DNI is very high and this is especially relevant for Concentrating PV (CPV) systems. On the contrary, solar thermal power plants, often denoted as Concentrating Solar Power (CSP) technology, have several methods to control short term, as well as daily, variability. This is given by the inertia of the whole system (solar field, heat transfer and storage), which can additionally be supported by storage or fossil fuels. DNI solar resource availability in Malawi does not give prospects for installation of solar concentrating technologies. We present this chapter only to provide overview information. 2.4.1 Concentrating Solar Power (CSP) A distinctive characteristic of Concentrated Solar Power technology (CSP) is that, when deployed with thermal energy storage, it can produce electricity on demand, providing a dispatchable source of renewable energy. Therefore, it can provide electricity whenever needed to meet demand, performing like a traditional base-load power plant. There are several groups of solar thermal power plants: • Parabolic troughs: solar fields using trough systems capture solar energy using large mirrors that track the sun’s movement throughout the day. The curved shape reflects most of that heat onto a receiver pipe that is filled with a heat transfer fluid. The thermal energy from the heated fluid generates steam, which in turn generates electricity in a conventional steam turbine. Heated fluid in the trough systems can also provide heat to thermal storage systems, which can be used to generate electricity at times when the sun is not shining; • Power towers: they use flat mirrors (heliostats) to reflect sunlight onto a solar receiver at the top of a central tower. Water is pumped up the tower to the receiver, where concentrated thermal energy heats it up. The hot steam then powers a conventional steam turbine. Some power towers use molten salt in place of water and steam. That hot molten salt can be used immediately to generate steam and electricity, or it can be stored and used at a later time. • Fresnel reflectors: they are made of many thin, flat mirror strips to concentrate sunlight onto tubes through which working fluid is pumped. The rest of the energy cycle works similarly as in the above-mentioned systems. • Stirling dish: consists of a stand-alone parabolic reflector that concentrates light onto a receiver positioned at the reflector's focal point. The reflector tracks the Sun along two axes. The working fluid in the receiver is heated and then used by a Stirling engine to generate power. One of the advantages of technology is thermal storage, often in the form of molten salt. CSP can also be integrated with fossil-based generation sources in a hybrid configuration. © 2018 Solargis page 29 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 2.4.2 Concentrating photovoltaics (CPV) A different conversion method of DNI into electricity is Concentrated Photovoltaic (CPV). This technology is based on the use of lenses or curved mirrors to concentrate sunlight onto a small area of high-efficiency PV cells. High concentration CPV requires very precise solar trackers. The advantage of CPV over flat plate PV is a potential for cost reduction due to the smaller area of photovoltaic material required. The necessity of sun tracking partially balances out the smaller price of the semiconductor material used. CPV technology also requires more maintenance during the lifetime of the power plant. Power production from CPV may be more sensitive to changing weather conditions. The advantage of CPV over CSP is full scalability, similar to flat plate PV modules. © 2018 Solargis page 30 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 3 Solar resource and PV potential of Malawi 3.1 Geography This report analyses solar and meteorological data for Malawi, which determine solar power production and influence its performance efficiency. We also analyse other geographical factors that influence the development and operation of solar power plants. Malawi is located in Southeast Africa, approximately between latitudes 9° and 17° South and longitudes 32° and 36° East. We demonstrate the variability of the solar resource and photovoltaic power potential in two forms: • At the country level in the form of maps • Graphs and tables for three selected sites that, to a large extent, represent the variability of the climate and solar power (ESMAP solar meteorological stations). The position of these sites is summarised in Table 2.1 and Map 2.1. The data in the tables and graphs shown in Chapter 3 relate to these three sites. Geographic information and maps bring additional value to the solar data. Geographical characteristics of the country from a regional to local scale may represent technical and environmental prerequisites, as well as constraints, for solar energy development. In this report, we collected the following data that have some relevance to solar energy: • Map of the administrative division and important cities/towns informs about the country spatial organization and population distribution (Map 3.1) • Terrain, where elevation above sea level and slope inclination may pose physical limitations for development (Map 3.2 and 3.3) • Rainfall (precipitation) has impact on efficiency (performance ratio) and operation (modules cleaning effect) of the PV installations (Map 3.4) • Land cover defines primary areas used for human economic activities and settlements and possible land availability for solar PV installations (Map 3.5) • Transport network (roads and railways), defining accessibility of sites for location of PV power plants (Map 3.6) • Population density is a good indicator of electricity consumption (Map 3.7). © 2018 Solargis page 31 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.1: Administrative division, towns and cities in Malawi. Source: Administrative boundaries by Cartography Unit, GSDPM (World Bank Group), GeoNames, adapted by Solargis © 2018 Solargis page 32 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.2: Terrain elevation above sea level. Source: SRTM v2 © 2018 Solargis page 33 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.3: Terrain slope. Based on: SRTM v2 data, calculated by Solargis. © 2018 Solargis page 34 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.4: Long-term yearly average of rainfall (sum of precipitation). Source: Global Precipitation Climatology Centre (DWD) © 2018 Solargis page 35 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.5: Land cover. Source: ESA Climate Change Initiative - Land Cover led by UCLouvain (2017) © 2018 Solargis page 36 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.6: Transport corridors. Source: OpenStreetMap.org contributors. © 2018 Solargis page 37 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.7: Population density. Source: Gridded Population of the World (GPW) v.4 © 2018 Solargis page 38 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 From the geographical viewpoint, Malawi is a diverse country, with Lake Malawi, Great Rift Valley and mountains in the North influencing the variability of geographical conditions. The central and eastern part is taken by plateaus while lowland areas are located in the South. Majority of the country is covered by agricultural cropland and forests with Lake Malawi taking sizeable portion of the border. The map of the land cover shows the most appropriate conditions for human activities, including settlements and economic activities (industry, agriculture) that require substantial amount of electrical power. These regions are developed mainly on the southern plateaus and lowlands. Smaller settlements are dispersed in the mountains and alongside the Lake Malawi coast. More complex orographic conditions (terrain) are generally less populated and are typically unsuitable for large- scale solar energy development; however, they are suitable for smaller, off-grid or hybrid installations. Urbanisation centres constitute the main energy demand centres. At present, about 46% of urban inhabitants in Malawi are connected to electricity grid (in rural areas it is only 1%) [39]. 3.2 Air temperature Air temperature determines the operating environment and performance efficiency of the solar power systems. Air temperature is used as one of the inputs in the energy simulation models. In this report, the yearly and monthly average maps are shown. Map 3.8 and Map 3.9 show the yearly and monthly averages. The long-term averages of air temperature are derived from the MERRA-2 and CFSv2 models (see Chapter 2.2) by Solargis post-processing. In the case of PV power plants, higher air temperature reduces the power conversion efficiency of the PV modules, as well as on other components (inverters, transformers, etc.). Monthly averages of daily values show the seasonal variation of air temperature at three selected sites in Malawi (Figure 3.1). See Chapter 2.2 discussing the uncertainty of the air temperature model estimates. 30.0 25.0 Monthly air temperature [°C] 20.0 15.0 10.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chileka Kasungu Mzuzu Figure 3.1: Monthly averages of air-temperature at 2 m for selected sites. © 2018 Solargis page 39 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Table 3.1: Monthly averages of air-temperature at 2 m at 3 sites Temperature [°C] Month Chileka Kasungu Mzuzu January 23.2 21.8 19.9 February 22.5 21.5 19.6 March 22.1 21.0 19.3 April 20.1 19.6 18.1 May 18.1 18.0 16.1 June 16.5 16.3 14.0 July 16.1 16.1 13.3 August 18.4 18.1 14.6 September 22.1 20.5 16.9 October 24.3 22.5 19.0 November 24.8 23.0 20.0 December 23.5 22.0 19.8 YEAR 21.0 20.0 17.6 Table 3.1 shows monthly characteristics of air temperature at three selected sites; they represent statistics calculated over a 24-hour diurnal cycle. © 2018 Solargis page 40 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.8: Long-term yearly average of air temperature at 2 metres, period 1994-2017. Source: Models CFSv2, MERRA-2, post-processed by Solargis © 2018 Solargis page 41 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.9: Long-term monthly average of air temperature at 2 metres, period 1994-2017. Source: Models CFSv2, MERRA-2, post-processed by Solargis © 2018 Solargis page 42 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 3.3 Global Horizontal Irradiation Global Horizontal Irradiation (GHI) is used as a reference value for comparing geographical conditions related to PV electricity systems, as it eliminates possible variations influenced by the choice of components and the PV system design. Table 3.2 shows long-term average, and average minima and maxima of daily totals of Global Horizontal Irradiation (GHI) for a period 1994 to 2017 for three selected sites. Figure 3.2 compares daily values of GHI at selected sites. When comparing GHI for these sites, they demonstrate a very similar pattern. The weather with highest GHI values is observed in October and November. Some variability of GHI between sites is observed in June and July. Very small variability of values is determined by similar geographical characteristics, and Figure 3.2 indicates that all sites will experience similar PV power performance. Table 3.2: Daily averages of Global Horizontal Irradiation at 3 sites 2 Variability Global Horizontal Irradiation [kWh/m ] Month between Chileka Kasungu Mzuzu sites [%] January 5.06 4.97 4.67 4.2 February 5.34 5.11 4.63 7.2 March 5.19 5.24 4.81 4.6 April 4.90 5.13 4.47 7.0 May 4.58 5.06 4.47 6.7 June 4.01 4.73 4.29 8.4 July 4.00 4.80 4.38 9.1 August 4.84 5.43 5.23 5.8 September 5.75 6.34 6.18 5.0 October 6.01 6.67 6.52 5.4 November 5.91 6.37 6.27 3.9 December 5.50 5.46 5.21 2.9 YEAR 5.09 5.44 5.09 3.9 © 2018 Solargis page 43 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 7.0 6.0 5.0 Daily sums of GHI [kWh/m2] 4.0 3.0 2.0 1.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chileka Kasungu Mzuzu Figure 3.2: Long-term monthly averages of Global Horizontal Irradiation. 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 shows the magnitude of this change. 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 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Chileka 4.8% Kasungu 4.0% Mzuzu 4.2% Figure 3.3: Interannual variability of Global Horizontal Irradiation for selected sites. The interannual variability of GHI for the representative sites is calculated from the unbiased standard deviation of GHI over 24 years taking into consideration the long-term, normal distribution of the annual sums. All sites show similar patterns of GHI changes over the recorded period (Figure 3.3). More stable GHI (the lowest interannual variability) is observed at Kasungu site. Higher variability is observed at Chileka site (4.8%). © 2018 Solargis page 44 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.10: Global Horizontal Irradiation – long-term average of daily and yearly totals. Source: Solargis © 2018 Solargis page 45 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.11: Global Horizontal Irradiation – long-term monthly average of daily totals. Source: Solargis © 2018 Solargis page 46 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 The highest GHI is identified in the Southern part of the Malawi lake, where average daily totals exceed 5.6 kWh/m2 (yearly sum about 2050 kWh/m2) and more (Map 3.10). The season of highest irradiation with daily totals up to 7.0 kWh/km2 lasts three months (from September to November, Map 3.11). The lowest values of GHI is documented can be expected in June and July. Map 3.12 shows the ratio of diffuse to global horizontal irradiation. This ratio is important for the performance of PV systems and may have impact during the consideration process of PV modules technology. A higher ratio of diffuse to global horizontal irradiation (DIF/GHI) indicates less stable weather, higher occurrence of clouds, higher atmospheric pollution or water vapour. In general, higher values occurs along the Lake Malawi and mountains in the North and South-east of the country (up to 55%). In the lowlands the values fall to 35%. Temporary, lower DIF/GHI values are identified from May to October, with highest being in January and February. This indicates that the potential for concentrated solar technologies (CSP, CPV) in Malawi is low due to high DIF/GHI ration and seasonality of solar radiation. 70 60 Monthly averages of DIF/GHI [%] 50 40 30 20 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Chileka Kasungu Mzuzu Figure 3.4: Monthly averages of DIF/GHI. © 2018 Solargis page 47 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.12: Long-term average for ratio of diffuse and global irradiation (DIF/GHI). Source: Solargis © 2018 Solargis page 48 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 3.4 Direct Normal Irradiation Direct Normal Irradiation (DNI) is important solar resource parameters needed for the computation of Global Tilted Irradiation (GTI) (Chapter 3.5). Table 3.3 and Figure 3.5 show long-term average daily totals of DNI for the three selected sites, during the period from 1994 to 2017. The highest DNI is reached in Kasungu, and the lowest in Mzuzu. Table 3.3: Daily averages of Direct Normal Irradiation at three sites 2 Variability Direct Normal Irradiation [kWh/m ] Month between Chileka Kasungu Mzuzu sites [%] January 3.14 2.68 2.44 12.9 February 3.75 3.04 2.43 21.5 March 4.14 3.61 2.89 17.7 April 4.83 4.47 3.14 21.5 May 5.35 5.49 3.98 16.8 June 4.71 5.37 4.16 12.8 July 4.25 5.11 4.12 12.0 August 4.75 5.09 4.71 4.4 September 4.90 5.32 5.36 4.9 October 4.76 5.41 5.56 8.2 November 4.47 4.98 5.27 8.2 December 3.75 3.49 3.48 4.2 YEAR 4.40 4.51 3.97 6.7 © 2018 Solargis page 49 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 7.0 6.0 5.0 Daily sums of DNI [kWh/m2] 4.0 3.0 2.0 1.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chileka Kasungu Mzuzu Figure 3.5: Daily averages of Direct Normal Irradiation at selected sites. Interannual variability of DNI for selected sites (Figure 3.6) is calculated from the unbiased standard deviation of yearly DNI over 24 years and it is based on a simplified assumption of normal distribution of the yearly sums. Three sites show similar patterns of DNI variability over recorded period. The most stable DNI (the lowest interannual variability) is observed in Kasungu. 6.0 2192 Average annual sum of Direct Normal Irradiation [kWh/m2] Average daily sum of Direct Normal Irradiation [kWh/m2] 5.5 2009 5.0 1826 4.5 1644 4.0 1461 3.5 1278 3.0 1096 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Chileka 10.7% Kasungu 9.5% Mzuzu 9.6% Figure 3.6: Interannual variability of Direct Normal Irradiation at representative sites The highest DNI in the very North region and South-west lake Malawi area represents average daily totals over 5.0 kWh/m2 (equal to yearly sum of about 1825 kWh/m2, Map 3.13). High DNI occurs during the months from May to October, often exceeding the daily totals over 5.5 kWh/m2 (Map 3.14). However, in January and February DNI daily totals drop significantly. © 2018 Solargis page 50 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.13: Direct Normal Irradiation – long-term average of daily and yearly totals. Source: Solargis © 2018 Solargis page 51 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.14: Direct Normal Irradiation – long-term monthly average of daily totals. Source: Solargis © 2018 Solargis page 52 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 3.5 Global Tilted Irradiation Global Tilted Irradiation (GTI) is the key source of energy for flat-plate photovoltaic (PV) technologies (Chapter 3.6). Optimally tilted PV module produces more energy output annually compared to non-tilted module. The magnitude of the tilt also determines the ability of self-cleaning effect of the modules during the rainfall events (by washing dust and dirt). Table 3.4 shows the long-term averages of average daily total Global Tilted Irradiation (GTI) for selected sites. It is assumed that solar radiation is received by PV modules with surface at optimum tilt. Table 3.4: Daily averages of Global Tilted Irradiation at 3 sites 2 Variability Global Tilted Irradiation [kWh/m ] Month between Chileka Kasungu Mzuzu sites [%] January 4.71 4.65 4.41 3.5 February 5.16 4.93 4.50 6.9 March 5.33 5.32 4.85 5.2 April 5.41 5.54 4.72 8.5 May 5.41 5.81 4.95 8.0 June 4.85 5.58 4.86 8.1 July 4.74 5.57 4.92 8.6 August 5.47 6.01 5.68 4.8 September 6.07 6.61 6.41 4.3 October 5.93 6.55 6.42 5.1 November 5.53 5.95 5.91 4.0 December 5.04 5.03 4.84 2.2 YEAR 5.30 5.63 5.21 4.1 In Malawi, the optimum tilt of PV modules (for maximized yearly production) is between 9° and 19° (increasing from North to South) with North orientation (Map 3.15). Figure 3.7 compares long-term daily averages at selected sites. Stable weather with high GTI values is seen from August to November. Variability of GTI in all selected sites is relatively small. Lower daily averages in period from December to March are very similar for all sites, which are related to the rainy season. © 2018 Solargis page 53 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.15: Optimum tilt of PV modules to maximize yearly PV power production. Source: Solargis © 2018 Solargis page 54 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 7.0 6.0 5.0 Daily sums of GTI [kWh/m2] 4.0 3.0 2.0 1.0 0.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chileka Kasungu Mzuzu Figure 3.7: Global Tilted Irradiation – long-term daily averages. A surface inclined at an optimum angle (tilt) gains more yearly irradiation than a horizontal surface (depending on the latitude of a site). In Malawi optimum tilt ranges between 9° and 19°. While seasonal gains of GTI in comparison to GHI are high (between 13% to 21%), the yearly gains of GTI are relatively small. Compared to GHI, GTI gain in the North of the country reaches about 1-2%, in the South it can be above 4% (Map 3.16 and 3.17). This is documented in Figure 3.8, where a positive gain of GTI is about 2.2% (Mzuzu) to 4.2% (Chileka). Despite relatively small yearly gain of GTI compared to GHI, the installation of modules in inclined position has additional positive effect of natural cleaning of the modules by rain. 30.0 20.0 10.0 Relative gain of GTI to GHI [% ] 0.0 -10.0 -20.0 -30.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chileka Kasungu Mzuzu Figure 3.8: Monthly relative gain of GTI relative to GHI at selected sites. © 2018 Solargis page 55 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.16: Global Tilted Irradiation at optimum tilt – long-term average of daily and yearly totals. Source: Solargis © 2018 Solargis page 56 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.17: Global Tilted Irradiation at optimum tilt – long-term monthly average of daily totals. Source: Solargis © 2018 Solargis page 57 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 3.6 Photovoltaic power potential The PV potential from a reference system for three representative sites is shown in Table 3.5. Despite the geographic distribution of selected sites, electricity production from a PV power system is similar for all sites and follows a combined pattern of global solar irradiation and air temperature. Considering three selected sites, the difference between production from the “best” site (Kasungu, 4.44 kWh/kWp) and “the least productive” site (Chileka, 4.18 kWh/kWp) is low, about 6%. Also, monthly power production profiles are very similar for all sites. The highest seasonal production occurs from August to October (Table 3.6). Table 3.5: Annual performance parameters of a PV system with modules fixed at the optimum angle Chileka Kasungu Mzuzu PVOUT 4.18 4.44 4.20 Average daily total [kWh/kWp] PVOUT 1525 1624 1534 Yearly total [kWh/kWp] Annual ratio of DIF/GHI 39.9% 42.6% 45.1% System PR 78.7% 78.9% 80.7% PVOUT - PV electricity yield for fixed-mounted modules at optimum angle; DIF/GHI – Ratio of Diffuse/Global horizontal irradiation; PR - Performance ratio for fixed-mounted PV Table 3.6: Average daily sums of PV electricity output from an open-space fixed PV system with a nominal peak power of 1 kW [kWh/kWp] Average daily sum of electricity production [kWh/kWp] Site Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chileka 3.68 4.04 4.19 4.30 4.34 3.93 3.84 4.36 4.72 4.57 4.25 3.92 4.18 Kasungu 3.66 3.88 4.20 4.41 4.65 4.49 4.49 4.78 5.17 5.07 4.61 3.93 4.44 Mzuzu 3.53 3.60 3.90 3.82 4.05 4.02 4.07 4.64 5.14 5.09 4.68 3.85 4.20 6.00 5.00 Electricity production [kWh/kWp] 4.00 3.00 2.00 1.00 0.00 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Chileka Kasungu Mzuzu Figure 3.9: Monthly averages of daily totals of power production from the fixed tilted PV systems with a nominal peak power of 1 kW at three sites [kWh/kWp] © 2018 Solargis page 58 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Maps 3.18 and 3.19 shows yearly and monthly production from a PV power system, and Figure 3.9 breaks down the values for the three sites. The season of relatively high PV yield is long enough for the effective operation of a PV system. As shown in Chapter 3.5, it is recommended to install modules at optimum tilt towards equator rather than on a horizontal surface. Besides higher yield, a benefit of tilted modules is improved self-cleaning of the surface pollution by rain. The monthly and yearly performance ratios (PR) of a reference installation for the selected sites are shown in Table 3.7 and Figure 3.10. The range of yearly PR for the selected sites is between 78.7% and 80.7%. Mzuzu site has the highest PR due to lower temperature (Chapter 3.2). Table 3.7: Monthly and annual Performance Ratio of a free-standing PV system with fixed modules Monthly Performance Ratio [%] Site Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Chileka 78.0 78.3 78.7 79.5 80.1 81.0 80.9 79.6 77.8 77.0 76.9 77.8 78.7 Kasungu 78.5 78.6 78.9 79.5 80.0 80.5 80.6 79.4 78.2 77.4 77.4 78.2 78.9 Mzuzu 79.9 80.1 80.3 81.0 81.8 82.5 82.7 81.7 80.3 79.4 79.2 79.6 80.7 85.0 83.0 Performance ratio [%] 81.0 79.0 77.0 75.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Chileka Kasungu Mzuzu Figure 3.10: Monthly performance ratio of a PV system at selected sites. Fixed mounted modules at optimum tilt towards equator are considered Map 3.18 shows the average daily total of specific PV electricity output from a typical open-space PV system with optimally tilted c-Si modules and a nominal peak power of 1 kW (thus the values are in kWh/kWp). Calculating PV output for 1 kWp of installed power makes it simple to scale the PV power production relative to the size of a power plant. Besides the technology choice, the electricity production depends on the geographical position of the power plant. In most regions of Malawi, the average daily sums of specific PV power production from a reference system vary between 4.3 kWh/kWp (equals to yearly sum of about 1570 kWh/kWp) and 4.5 kWh/kWp (about 1640 kWh/kWp yearly). Average daily totals for the year are very uniform throughout all of Malawi. The best season for PV power production is from August to October, with highest values in September, when they can exceed 5.2 kWh/kWp. © 2018 Solargis page 59 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.18: PV electricity output from an open space fixed-mounted PV system with PV modules mounted at optimum tilt towards equator and a nominal peak power of 1 kWp. Long-term averages of daily and yearly totals. © 2018 Solargis page 60 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Map 3.19: PV power generation potential for an open-space fixed-mounted PV system. Long-term monthly averages of daily totals. Source: Solargis © 2018 Solargis page 61 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 3.7 Evaluation The chapters above describe various aspects of PV power generation potential in Malawi, and its relevance for the development and operation of photovoltaic systems. A large extent of the country has an average PV electricity daily output within a range from around 3.9 to 4.6 kWh/kWp (equals to average yearly totals between 1420 and 1680 kWh/kWp). This fact positions Malawi into the category of countries with very feasible potential for PV power generation. Additionally, the seasonal variability in the country is low, when compared to other regions further away from the equator. The ratio between months with maximum and minimum GHI is about 1.50 in Chileka, which is better than the ratio for Upington, South Africa (2.29) or Sevilla, Spain (3.54) (Figure 3.11). 10 GHI 8 6 kWh/m2 4 2 Sevilla, 1838 kWh/m2 year Chileka, 1859 kWh/m2 year Upington, 2272 kWh/m2 year 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Figure 3.11: Comparing seasonal variability in three locations for GHI © 2018 Solargis page 62 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 4 Data delivered for Malawi The following data and maps is delivered for Malawi: Site data available for the locations of three solar meteorological stations. The data can be accessed through an online application https://energydata.info/: • High accuracy 1-minute measurements (time series) acquired over a period of 24 months (2016-2018) • High accuracy site-adapted Solargis model 15-minute historical time series and hourly TMY data. The data represent history of years 1994 to 2017 Country-wide spatial data (GIS files) and maps. These outputs can be accessed as downloadable GIS files and maps through an online map-based application https://globalsolaratlas.info/: • Harmonized solar and meteorological GIS-based data. Regionally adapted solar resource and temperature data for Malawi. The long-term averages represent history of years 1994 to 20178 at 1-km grid resolution. • High resolution poster maps and medium resolution maps The delivered data and maps offer a good basis for knowledge-based decision making and project development. Solargis database is updated in real time and this data can be further used in solar monitoring, performance assessment and forecasting. More information about site specific data products is available in [29]. More information about spatial data products is available in chapters below. 4.1 Spatial data products High-resolution Solargis data have been delivered in the format suitable for common GIS software. The Primary data represent solar radiation, meteorological data and PV power potential. The Supporting data include various vector data, such as administrative borders, cities, etc. Tables 4.1 and 4.2 show information about the data layers and the technical specification is summarized in Tables 4.3 and 4.4. File name convention, used for the individual data sets, is described in Table 4.5. Metadata is delivered with the data files in two formats, according to ISO 19115:2003/19139 standards: • PDF - human readable • XML - for machine-to-machine communication The snapshots of most of the data can be viewed on the maps in Chapter 3. Table 4.1: General information about GIS data layers Geographical extent Republic of Malawi, including 10 km buffer zone along the country border between 18°S and 9°S, 32°E and 37°E Map projection Geographic (Latitude/Longitude), datum WGS84 (also known as GCS_WGS84; EPSG: 4326) Data formats ESRI ASCII raster data format (asc) GeoTIFF raster data format (tif) Notes: • Data layers of both formats (asc and tif) contain the same information; the operator is free to choose the preferential data format. Data layers can be also converted to other standard raster formats. • More information about ESRI ASCII grid format can be found at http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#/ESRI_ASCII_raster_format/009t0000000z000000/ © 2018 Solargis page 63 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 • More information about GeoTIFF format can be found at https://trac.osgeo.org/geotiff/ Table 4.2: Description of primary GIS data layers Acronym Full name Unit Type of use Type of data layers GHI Global Horizontal kWh/m2 Reference information for the Long-term yearly and monthly Irradiation assessment of flat-plate PV average of daily totals (photovoltaic) and solar heating technologies (e.g. hot water) DNI Direct Normal kWh/m2 Assessment of Concentrated PV (CPV) Long-term yearly and monthly Irradiation and Concentrated Solar Power (CSP) average of daily totals technologies, but also calculation of GTI for fixed mounting and sun-tracking flat plate PV DIF Diffuse Horizontal kWh/m2 Complementary parameter to GHI and Long-term yearly and monthly Irradiation DNI average of daily totals GTI Global Irradiation at kWh/m2 Assessment of solar resource for PV Long-term yearly and monthly optimum tilt towards technologies average of daily totals equator OPTA Optimum tilt ° Optimum tilt of PV modules to maximise Long-term average the yearly yield PVOUT Photovoltaic power kWh/kWp Assessment of power production Long-term yearly and monthly potential potential for a PV power plant with free- average of daily totals standing fixed-mounted c-Si modules, optimally tilted towards equator to maximize yearly PV production TEMP Air Temperature at 2 °C Defines operating environment of solar Long-term (diurnal) annual and m above ground power plants monthly averages level Table 4.3: Characteristics of the raster output data files Characteristics Range of values West − East 32:00:00E − 37:00:00E North − South 9:00:00S − 19:00:00S Resolution GHI, DNI, GTI, DIF, PVOUT 00:00:09 (2000 columns x 3600 rows) Resolution TEMP 00:00:30 (1200 columns x 1800 rows) Resolution OPTA 00:02:00 (150 columns x 270 rows) Data type Float No data value -9999, NaN © 2018 Solargis page 64 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Table 4.4: Technical specification of primary GIS data layers Acronym Full name Data format Spatial resolution Time No. of data layers (pixel size) representation GHI Global Horizontal Irradiation Raster 9 arc-sec. 1994 – 2017 12+1 (approx. 275 x 275 m) DNI Direct Normal Raster 9 arc-sec. 1994 – 2017 12+1 Irradiation (approx. 275 x 275 m) DIF Diffuse Horizontal Irradiation Raster 9 arc-sec. 1994 – 2017 12+1 (approx. 275 x 275 m) GTI Global Irradiation at optimum Raster 9 arc-sec. 1994 – 2017 12+1 tilt towards equator (approx. 275 x 275 m) OPTA Optimum tilt Raster 2 arcmin - 1 (approx. 3650 x 3650 m) PVOUT Photovoltaic power potential Raster 9 arc-sec. 1994 – 2017 12+1 (approx. 275 x 275 m) TEMP Air Temperature at 2 m Raster 30 arc-sec. 1994 – 2017 12+1 above ground level (approx. 930x930 m) Explanation: • MM: month of data – from 01 to 12 • ext: file extension (asc or tif) Data layers are provided as separate files in a tree structure, organized according to • File format (ASCII or GEOTIF) • Time summarization (yearly and monthly) Complementary files: • Project files (*.prj) complement ESRI ASCII grid files (*.asc) The support GIS data are provided in a vector format (ESRI shapefile, Table 4.6). © 2018 Solargis page 65 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Table 4.5: File name convention for GIS data Acrony Full name Filename pattern Number m of files GHI Global Horizontal Irradiation, long-term yearly average of daily totals GHI.ext 1+1 GHI Global Horizontal Irradiation, long-term monthly averages of daily totals 12+12 GHI_MM.ext DNI Direct Normal Irradiation, long-term yearly average of daily totals DNI.ext 1+1 DNI Direct Normal Irradiation, long-term monthly averages of daily totals DNI_MM.ext 12+12 DIF Diffuse Horizontal Irradiation, long-term yearly average of daily totals DIF.ext 1+1 DIF Diffuse Horizontal Irradiation, long-term monthly averages of daily totals DIF_MM.ext 12+12 GTI Global Irradiation at 7° tilt towards equator, long-term yearly average of daily GTI.ext 1+1 totals GTI Global Irradiation at 7° tilt towards equator, long-term monthly averages of GTI_MM.ext 12+12 daily totals PVOUT Photovoltaic power potential , long-term yearly average of daily totals PVOUT.ext 1+1 PVOUT Photovoltaic power potential , long-term monthly averages of daily totals PVOUT_MM.ext 12+12 TEMP Air Temperature at 2 m above ground, long-term yearly average TEMP.ext 1+1 TEMP Air Temperature at 2 m above ground, long-term monthly averages TEMP_MM.ext 12+12 Total size of unpacked data layers is 4200 MB, packed (with ZIP compression) 468 MB respectively. Table 4.6: Support GIS data Data type Source Data format City location OpenStreetMap.org contributors, GeoNames.org, Point shapefile adapted by Solargis Administrative borders Cartography Unit, GSDPM, World Bank Group Polyline shapefile Large water bodies SWBD, USGS Polygon shapefile Solar meteorological stations Solargis Point shapefile © 2018 Solargis page 66 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 4.2 Project in QGIS format For easy manipulation with GIS data files, selected vector and raster data files are integrated into ready-to-open Quantum GIS (QGIS) project file with colour styles and annotations (see Figure 4.1). QGIS is state-of-art open-source GIS software allowing visualization, query and analysis on the provided data. QGIS includes a rich toolbox to manipulate data. More information about the software and download packages can be found at http://qgis.org. Figure 4.1: Screenshot of the map and data in the QGIS environment 4.3 Map images Besides GIS data layers, digital maps are also delivered for selected data layers for presentation purposes. Digital images (maps) are prepared in two types; each suitable for different purpose: • High-resolution poster maps, printing size 120 x 80 cm, prepared as the colour-coded maps in a TIFF format at 300 dpi density and lossless compression • Mid-size maps suitable for A4 printing or on-screen presentation, prepared in PNG format at 300 dpi density and lossless compression The following three parameters are processed in the form of maps: • Global Horizontal Irradiation – Yearly average of the daily totals • Direct Normal Irradiation − Yearly average of the daily totals • Photovoltaic electricity production from a free-standing power plant with optimally tilted c-Si modules − Yearly average of the daily totals The maps will be released to be downloadable from the Download section of Global Solar Atlas (see Figure 4.2): http://globalsolaratlas.info/downloads/Malawi © 2018 Solargis page 67 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 Figure 4.2: Screenshot of the Download section at Global Solar Atlas in Dec 2018. © 2018 Solargis page 68 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 5 List of maps Map 2.1: Position of the solar meteorological stations used for the model validation ............................................... 16 Map 2.2: Geographic distribution of the model uncertainty in Malawi ......................................................................... 22 Map 3.1: Administrative division, towns and cities in Malawi. ...................................................................................... 32 Map 3.2: Terrain elevation above sea level..................................................................................................................... 33 Map 3.3: Terrain slope. .................................................................................................................................................... 34 Map 3.4: Long-term yearly average of rainfall (sum of precipitation). .......................................................................... 35 Map 3.5: Land cover. ........................................................................................................................................................ 36 Map 3.6: Transport corridors. .......................................................................................................................................... 37 Map 3.7: Population density. ........................................................................................................................................... 38 Map 3.8: Long-term yearly average of air temperature at 2 metres, period 1994-2017. ............................................. 41 Map 3.9: Long-term monthly average of air temperature at 2 metres, period 1994-2017. ......................................... 42 Map 3.10: Global Horizontal Irradiation – long-term average of daily and yearly totals. ............................................ 45 Map 3.11: Global Horizontal Irradiation – long-term monthly average of daily totals. ................................................ 46 Map 3.12: Long-term average for ratio of diffuse and global irradiation (DIF/GHI). ................................................... 48 Map 3.13: Direct Normal Irradiation – long-term average of daily and yearly totals. .................................................. 51 Map 3.14: Direct Normal Irradiation – long-term monthly average of daily totals....................................................... 52 Map 3.15: Optimum tilt of PV modules to maximize yearly PV power production. ..................................................... 54 Map 3.16: Global Tilted Irradiation at optimum tilt – long-term average of daily and yearly totals. .......................... 56 Map 3.17: Global Tilted Irradiation at optimum tilt – long-term monthly average of daily totals. .............................. 57 Map 3.18: PV electricity output from an open space fixed-mounted PV system ........................................................ 60 Map 3.19: PV power generation potential for an open-space fixed-mounted PV system. .......................................... 61 © 2018 Solargis page 69 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 6 List of figures Figure 2.1: Solar resource data availability (GHI, DNI and DIF). .................................................................................... 15 Figure 2.2: Simplified Solargis PV simulation chain ...................................................................................................... 27 Figure 3.1: Monthly averages of air-temperature at 2 m for selected sites. ................................................................ 39 Figure 3.2: Long-term monthly averages of Global Horizontal Irradiation. .................................................................. 44 Figure 3.3: Interannual variability of Global Horizontal Irradiation for selected sites. ................................................ 44 Figure 3.4: Monthly averages of DIF/GHI........................................................................................................................ 47 Figure 3.5: Daily averages of Direct Normal Irradiation at selected sites. ................................................................... 50 Figure 3.6: Interannual variability of Direct Normal Irradiation at representative sites ............................................... 50 Figure 3.7: Global Tilted Irradiation – long-term daily averages. .................................................................................. 55 Figure 3.8: Monthly relative gain of GTI relative to GHI at selected sites..................................................................... 55 Figure 3.9: Monthly averages of daily totals of power production from the fixed tilted PV systems......................... 58 Figure 3.10: Monthly performance ratio of a PV system at selected sites. ................................................................. 59 Figure 3.11: Comparing seasonal variability in three locations for GHI ....................................................................... 62 Figure 4.1: Screenshot of the map and data in the QGIS environment ........................................................................ 67 Figure 4.2: Screenshot of the Download section at Global Solar Atlas in Dec 2018. .................................................. 68 © 2018 Solargis page 70 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 7 List of tables Table 1.1: Comparison of longterm GHI estimate: Solargis vs. previous studies ................................................. 13 Table 2.1: Overview information on measurement stations operated in the region .............................................. 15 Table 2.2: Overview information on solar meteorological stations operating in the region ................................. 16 Table 2.3: Input data for Solargis solar radiation model and related GHI and DNI outputs for Malawi ............... 18 Table 2.4: Comparing solar data from solar measuring stations and from satellite models ............................... 19 Table 2.5: Direct Normal Irradiance: bias before and after regional model adaptation ........................................ 20 Table 2.6: Global Horizontal Irradiance: bias before and after regional model adaptation .................................. 20 Table 2.7: Uncertainty of the model estimate for original and regionally-adapted annual GHI, DNI and GTI ...... 21 Table 2.8: Comparing data from meteorological stations and weather models ................................................... 23 Table 2.9: Original source of Solargis air temperature at 2 m for Malawi: MERRA-2 and CFSv2. ........................ 24 Table 2.10: Air temperature at 2 m: accuracy indicators of the model outputs [ºC].......................................... 24 Table 2.11: Expected uncertainty of air temperature in Malawi. ......................................................................... 24 Table 2.12: Specification of Solargis database used in the PV calculation in this study .................................. 26 Table 2.13: Reference configuration - photovoltaic power plant with fixed-mounted PV modules .................. 28 Table 2.14: Yearly energy losses and related uncertainty in PV power simulation............................................ 28 Table 3.1: Monthly averages of air-temperature at 2 m at 3 sites .......................................................................... 40 Table 3.2: Daily averages of Global Horizontal Irradiation at 3 sites ...................................................................... 43 Table 3.3: Daily averages of Direct Normal Irradiation at three sites ..................................................................... 49 Table 3.4: Daily averages of Global Tilted Irradiation at 3 sites .............................................................................. 53 Table 3.5: Annual performance parameters of a PV system with modules fixed at the optimum angle ............. 58 Table 3.6: Average daily sums of PV electricity output from an open-space fixed PV system ............................ 58 Table 3.7: Monthly and annual Performance Ratio of a free-standing PV system with fixed modules ............... 59 Table 4.1: General information about GIS data layers ............................................................................................. 63 Table 4.2: Description of primary GIS data layers .................................................................................................... 64 Table 4.3: Characteristics of the raster output data files ........................................................................................ 64 Table 4.4: Technical specification of primary GIS data layers ................................................................................ 65 Table 4.5: File name convention for GIS data........................................................................................................... 66 Table 4.6: Support GIS data ....................................................................................................................................... 66 © 2018 Solargis page 71 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 8 References [1] Paul W. Stackhouse, Jr., Zhang T., Westberg D., Barnett A.J., Bristow T., Macpherson B., Hoell J. M., 2018. POWER Release 8 (with GIS Applications) Methodology (Data Parameters, Sources, & Validation), Data Version 8.0.1. https://power.larc.nasa.gov/documents/POWER_Data_v8_methodology.pdf. Web visited on 3 Oct 2018 [2] Šúri M., Wald L., Huld T., Dunlop E., Albuisson M, 2005. Integration of HelioClim-1 database into PVGIS to estimate solar electricity potential in Africa. In: Palz W., Ossenbrink H., Helm P., (eds.) 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, 6-10 June 2005, Barcelona, pp. 2989-2992” https://www.researchgate.net/publication/43406677 [3] Huld T., Müller R.W., Gambardella A., 2012. A new solar radiation database for estimating PV performance in Europe and Africa. Solar Energy 86 (6), 1803–1815. DOI: 10.1016/j.solener.2012.03.006 [4] Lefèvre M., Wald L., Diabaté L., 2007. "Using reduced data sets ISCCP-B2 from the Meteosat satellites to assess surface solar irradiance". Solar Energy, 81, 240-253. doi:10.1016/j.solener.2006.03.008 [5] ARMINES/MINES ParisTech/TRANSVALOR/ ECMWF − COPERNICUS Atmosphere Monitoring Service. Monthly and yearly average Global Horizontal Irradiation maps from CAMS Radiation Service over Africa. "JADE" volume, April 2018 [6] Meteonorm handbook, Version 7.3, Part II: Theory. Meteotest, 2018 [7] Global Solar Atlas, 2016. The World Bank Group. https://globalsolaratlas.info. [8] Marcel Suri, Tomas Cebecauer, Branislav Schnierer, Artur Skoczek, Daniel Chrkavy, Nada Suriova, 2018. Annual Solar Resource Report for solar meteorological stations after completion of 24 months of measurements. https://www.esmap.org/node/55906. [9] Perez R., Cebecauer T., Suri M., 2014. Semi-Empirical Satellite Models. In Kleissl J. (ed.) Solar Energy Forecasting and Resource Assessment. Academic press. [10] Cebecauer T., Šúri M., Perez R., High performance MSG satellite model for operational solar energy applications. ASES National Solar Conference, Phoenix, USA, 2010. [11] Šúri M., Cebecauer T., Perez P., Quality procedures of SolarGIS for provision site-specific solar resource information. Conference SolarPACES 2010, September 2010, Perpignan, France. [12] Šúri 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. [13] 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. [14] Šúri M., Cebecauer T., Requirements and standards for bankable DNI data products in CSP projects, Proceedings of the SolarPACES 2011 Conference, September 2011, Granada, Spain. [15] Ineichen P., A broadband simplified version of the Solis clear sky model, 2008. Solar Energy, 82, 8, 758-762. [16] Morcrette J., Boucher O., Jones L., Salmond D., Bechtold P., Beljaars A., Benedetti A., Bonet A., Kaiser J.W., Razinger M., Schulz M., Serrar S., Simmons A.J., Sofiev M., Suttie M., Tompkins A., Uncht A., GEMS-AER team, 2009. Aerosol analysis and forecast in the ECMWF Integrated Forecast System. Part I: Forward modelling. Journal of Geophysical Research, 114. [17] Benedictow A. et al. 2012. Validation report of the MACC reanalysis of global atmospheric composition: Period 2003-2010, MACC-II Deliverable D_83.1. [18] Cebecauer T., Perez R., Suri M., Comparing performance of SolarGIS and SUNY satellite models using monthly and daily aerosol data. Proceedings of the ISES Solar World Congress 2011, 28 August – 2 September 2011, Kassel, Germany. © 2018 Solargis page 72 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 [19] Cebecauer T., Šúri M., Accuracy improvements of satellite-derived solar resource based on GEMS re-analysis aerosols. Conference SolarPACES 2010, September 2010, Perpignan, France. [20] 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 [21] GFS model. http://www.nco.ncep.noaa.gov/pmb/products/gfs/ [22] MERRA-2 Overview: The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA- 2), Ronald Gelaro, et al., 2017, J. Clim., doi: 10.1175/JCLI-D-16-0758.1 [23] CFSR model. https://climatedataguide.ucar.edu/climate-data/climate-forecast-system-reanalysis-cfsr/ [24] CFSv2 model http://www.cpc.ncep.noaa.gov/products/CFSv2/CFSv2seasonal.shtml [25] 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. [26] 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. [27] 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. [28] Ruiz-Arias J. A., Cebecauer T., Tovar-Pescador J., Šúri M., Spatial disaggregation of satellite-derived irradiance using a high-resolution digital elevation model. Solar Energy, 84, 1644-1657, 2010. [29] Marcel Suri, Tomas Cebecauer, Branislav Schnierer, Artur Skoczek, Daniel Chrkavy, Nada Suriova, 2018. Solar Model Validation Report; Regional adaptation of Solargis model based on data acquired in 24-months solar measurement campaign; Republic of Malawi, Ref. Nr. 141-08/2018. https://www.esmap.org/node/55906. [30] Ineichen P. Long term satellite hourly, daily and monthly global, beam and diffuse irradiance validation. Interannual variability analysis. University of Geneva/IEA SHC Task 46, 2013. http://www.unige.ch/energie/forel/energie/equipe/ineichen/annexes-iae.html [31] Cebecauer T., Suri M., 2015. Site-adaptation of satellite-based DNI and GHI time series: overview and Solargis approach. SolarPACES conference, September 2015, Cape Town, North Africa. [32] King D.L., Boyson W.E. and Kratochvil J.A., Photovoltaic array performance model, SAND2004-3535, Sandia National Laboratories, 2004. [33] Huld T., Šúri M., Dunlop E.D., Geographical variation of the conversion efficiency of crystalline silicon photovoltaic modules in Europe. Progress in Photovoltaics: Research and Applications, 16, 595-607, 2008. [34] Huld T., Gottschalg R., Beyer H. G., Topic M., Mapping the performance of PV modules, effects of module type and data averaging, Solar Energy, 84, 2, 324-338, 2010. [35] Huld T., Friesen G., Skoczek A., Kenny R.P., Sample T., Field M., Dunlop E.D., 2011. A power-rating model for crystalline silicon PV modules. Solar Energy Materials and Solar Cells, 95, 12, 3359-3369. [36] Skoczek A., Sample T., Dunlop E. D., The results of performance measurements of field-aged crystalline silicon photovoltaic modules, Progress in Photovoltaics: Research and Applications, 17, 227-240, 2009. [37] Martin N., Ruiz J.M. Calculation of the PV modules angular losses under field conditions by means of an analytical model. Solar Energy Material and Solar Cells, 70, 25–38, 2001. [38] The German Energy Society, 2008: Planning and Installing Photovoltaic Systems. A guide for installers, architects and engineers. Second edition. Earthscan, London, Sterling VA. [39] Malawi, Power Africa fact sheet, USAID, https://www.usaid.gov/powerafrica/malawi © 2018 Solargis page 73 of 75 Solar Resource Atlas Based on regional adaptation of Solargis model Solargis reference No. 141-09/2018 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 Marcel Suri, Branislav Schnierer, Nada Suriova, Juraj Betak, Daniel Chrkavy, Artur Skoczek and Tomas Cebecauer from Solargis All maps in this report are prepared by Solargis Solargis s.r.o., Mytna 48, 811 07 Bratislava, Slovakia Reference No. (Solargis): 141-09/2018 http://solargis.com © 2018 Solargis page 74 of 75