Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services [ 1 ] Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services Photo: Sandsun. Summary Authors This Technical Note is aimed primarily at professionals in National Meteorological and David P. Rogers, Lead Meteorological Hydrological Services (NMHSs), who are trying to improve the quality and relevance of Consultant, World Bank; their services to match their societies’ growing needs for meteorological and hydrological Vladimir Tsirkunov, Lead Specialist, information. At the same time, the Note serves as a useful resource for the experts and Head of the Hydromet Program, teams involved in the activities and projects that tackle various aspects of transforming World Bank; hydromet services. It provides insight into some of the technical challenges that NMHSs Anna-Maria Bogdanova, Disaster Risk face and suggests approaches to addressing these challenges. Why is it important to Management Specialist, World Bank; continuously improve numerical weather prediction (NWP) and what observations are Makoto Suwa, Senior Disaster Risk needed to do this? What data should be shared with the international community and Management Specialist, World Bank how will this sharing lead to benefits at the national level? These are complex technical issues, which are not necessarily well understood by the staff of NMHSs and perhaps not at all by finance ministries, or by development partners or their advisors. The objective of this Note is to bring attention to these technical issues to better under- stand them and prioritize the potential solutions. The main messages are: Investments in upper-air stations should be prioritized. Investment in the ability to pro- file the atmosphere should be a priority. Many countries do not consider upper-air sta- tions during the network modernization because of the high operational costs. Instead the preference is to add more automatic weather observations with the assertion that such a network is fit-for-purpose; this has little or no justification. Upgrading of observation networks should support the Global Basic Observing Net- work (GBON) minimum requirements. The World Meteorological Organization (WMO) has established the minimum requirements for observations, which is referred to as the [ 2 ] Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services Global Basic Observing Network (GBON). It establishes both temporal and spatial re- Abbreviations quirements for automatic weather stations, radiosondes, and ocean observations. In- ECMWF European Centre for vestments in the GBON-compliant observation networks should be priorities. Medium-Range Weather Forecasts Moving to the ensemble prediction systems (EPS) should be scaled up and supported EFI Extreme Forecast Index by sufficient investments. Numerical models underpin all forecasting. In the past, the fo- EPS ensemble prediction systems EUMETSAT European Organisation cus has been on pursuing a higher and higher resolution of deterministic forecasts—that for the Exploitation of is, for a given starting point, you will always get the same result (a single value forecast). Meteorological Satellites This approach has given way to probabilistic forecasts, which take into account errors in GBON Global Basic Observing Network the initial observations or in the model to provide a set of forecasts, which represent the GWE Global Weather Enterprise complete range of possible outcomes. The approach applies to meteorology, hydrology, IMO International Meteorological and many other predictive fields. The approach has considerable implications for invest- Organization NMHSs National Meteorological and ment in people; for computational requirements; and for working with national, regional, Hydrological Services and global partners contributing to these prediction systems. Investing in a regional ap- NWP numerical weather prediction proach to EPS has significant benefits. SoT Shift of Tails SWFP Severe Weather Forecasting Programme WHO World Health Organization Introduction WMO World Meteorological Organization A major conundrum in the transformation of NMHSs in developing countries is to deter- WRF Weather Research and mine how investment can most effectively support the provision of meteorological and Forecasting Model hydrological services. Extensive analyses have been conducted over the past decades that make an irrefutable case for the value of high-quality meteorological and hydrolog- ical services to society (Hallegatte 2012; Hallegatte et al. 2017; Kull et al., forthcoming; WMO 2015). This issue is not, therefore, discussed further here. Lack of institutional analysis and lack of strategic-level dialogue is characteristic of many investments. In the absence of this analysis and dialogue, there is no quantitative way to evaluate how effective and sustainable proposed technical or infrastructure investments are, even if they support a country’s national priorities. Assessing what is achievable in terms of staff qualifications, capacity, and numbers plus the budget requirements to sustain operations and maintain any future systems is challenging, and therefore these are poorly defined elements of any investments (Rogers et al. 2019). Given the difficulties in retaining qual- ified staff and maintaining high-capital-cost observation equipment, it is essential that investments do not themselves overwhelm existing fragile NMHS systems rather than enhance them. What is essential to provide high-quality national and subnational forecasts is often lost in the NMHSs’ desire to run models locally, operate lots of automatic weather stations, and, in effect, try to emulate the infrastructure of the most-advanced centers without having the human or financial resources of those centers. This “independent” attitude, whereby the NMHS attempts to do everything for itself without considering the costs and benefits of alternative approaches, is often reinforced by development partners wishing to equate production of early warning or climate outlooks with investment in capital equipment—most frequently the automation and expansion of weather stations and capital investment in computers for NWP. The focus on equipment is capital-inten- sive and sometimes pursued at the expense of appropriate education and development of staff, which merit more attention from many governments. Yet, without knowledge- able institutional champions for organizational change directed to improve service deliv- ery and protect lives and property, it is highly unlikely that the transformation or, in the best case, innovation that many NMHSs expect to reach can be achieved. Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services [ 3 ] Although observations are the foundation of all forecasts, how they are collected and used are evolving rapidly as new technologies emerge and improved understanding of the physics and chemistry of the land-ocean-atmosphere system lead to better and bet- ter numerical prediction and hence better forecasts of the system. Despite this, many developing countries’ ability to observe, share data internationally, and use the pre- diction tools developed by leading forecasting centers remains very limited. Overcom- ing these obstacles would improve developing countries’ weather services in the most cost-effective and relatively rapid way. Here we explore three of these obstacles, which often come across many transformation processes. The first and, perhaps, the most egregious obstacle is the notion that auto- matic weather stations alone can be equated with better forecasts and warnings. The second is that sharing data internationally is primarily for the benefit of others, including the private sector, rather than the benefit of the country supplying the data; often the value of additional services to the economy of the country are not appreciated. The third is that running a high-resolution limited-area deterministic model over several days pro- vides forecast guidance superior to the forecast that can be obtained from a lower-reso- lution EPS run at a global center. We consider each of these issues in turn. What Observations Do We Need and Why? Most countries have the capacity to make in-situ, manual, or automatic weather and hydrological observations of basic parameters—pressure, temperature, wind, humidity, rainfall, stream flow, and so on. Some take upper-air measurements of winds, tempera- ture, and humidity. These data have numerous applications. Long-term measurements, if they are of sufficiently high quality, are fundamental to understanding climate; surface measurements, in general, are essential for the verification of numerical predictions and forecasts, and for downscaling numerical predictions to the desired resolution. Warnings Warnings based on based on observations—so-called warn-on-detection observations—have, by definition, observations—so-called warn- short lead times. Today, flash floods and tornados remain in this category, although ef- forts are underway to lengthen warning lead times through forecasting techniques. Most on-detection observations— other hazards can be characterized as warn-on-forecast—that is, with long lead times have, by definition, short lead usually obtained through various numerical prediction schemes capable of providing times. Today, flash floods and timely information to those at risk so they can take effective action (see Box 1 for a brief history of the origin of forecasting). tornados remain in this category, although efforts are underway to lengthen warning lead times Box 1. The Origin of Forecasting through forecasting techniques. The first storm warnings, issued by Vice-Admiral Robert FitzRoy, the founder of the Met Office, were based on real-time observations and the newly invented telegraph in “a race to warn the outpost before the gale reached them” (BBC 2015). Incidentally, producing storm warnings led FitzRoy to introduce general weather forecasts, which indicated the probable weather for two days ahead—a tall order with mixed success. These forecasts were limited by a lack of observational data of the large-scale struc- ture of the atmosphere, which hampered forecasts beyond a few hours until the cre- ation of the International Meteorological Organization (IMO) in 1873. The IMO began the job of creating a global framework for ongoing atmospheric observations and data collection, which continues today as the Global Observing System (Zillman 2017). [ 4 ] Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services Today, very short-, short-, medium-, and long-range forecasts and even nowcasting time and space scales all depend on a global network of in-situ and remotely sensed observ- ing systems and the assimilation of these data into global NWP models. Absent data assimilation, NWP has very little practical operational application; conversely, it might be argued that, absent the NWP scheme, observations have limited application in fore- casting. A typical global NWP data assimilation system processes more than 40 million observa- tions daily. The vast majority of these data are satellite measurements, but non-satellite sources remain important, including surface and aircraft observations. Radiosondes are the most influential terrestrial-based observations in reducing forecast error (Kull et al., forthcoming; Singh et al. 2014) because of the importance of the measurements of the surface pressure field and the vertical distribution of winds throughout the entire depth of the atmosphere. Unfortunately, the importance of radiosondes is often overlooked or underrepresented in development projects given the high cost of expendables. Investment in the ability to profile the atmosphere should be a priority. Upper-air sta- tions are not operational in many countries, however, and in many cases restoring this capability is a lower priority than adding more and more automatic weather observations with the assertion that such a network is fit-for-purpose with little or no justification. Avoiding the ongoing operational cost of expendables (balloons and radiosondes) for upper-air stations seems to outweigh the importance of these observations in many investment projects. Unfortunately, this approach is prolific and is the primary feature of many transformation efforts, where it seems the desire to procure high-value capital equipment precedes the intellectual guidance needed to use this information optimally. Historically, development projects have contributed significantly to the proliferation of surface-based observations, particularly automatic weather stations, in response to requests from NMHSs and often from other ministries and agencies involved in some Ultimately, the decision of aspects of hydromet services. Understanding the cost-benefits of the entire hydromete- orological value chain needs to be applied rigorously and driven by specific-user require- whether to site automatic ments. Ultimately, the decision of whether to site automatic weather stations 1 to 50 weather stations 1 to 50 kilometers apart should be guided by user needs, the technology available to meet these kilometers apart should be needs, and the sustainability of systems. guided by user needs, the Acknowledging the importance of in-situ observations, the decision to expand the obser- vation network should be guided by the needs of NWP and forecasting. Priority should technology available to be given to investments in upper-air stations, while expansion of automatic weather meet these needs, and the observations should be fit-for-purpose since the cost of maintenance of unnecessary sustainability of systems. installed equipment may impede investments in other critical infrastructure and services. What You Give Is What You Get, or the Importance of Data Sharing The importance of in-situ observations is highlighted in the WMO GBON, which is re- sponding to the need to reduce the inhomogeneity across the globe in the volume of observations internationally exchanged. The GBON requirements are seemingly not ex- ceptionally onerous. At a basic level, the requirements are hourly measurements of at- mospheric pressure, air temperature, humidity, horizontal wind, precipitation, and snow depth at a maximum spacing of 250 kilometers. Additional measurements of the same parameters at 100 kilometers or less apart are desirable. Upper-air measurements of Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services [ 5 ] temperature, humidity, and winds should be made twice a day or more frequently with a vertical resolution of 100 meters or higher reaching an altitude of 30 hectopascals or higher with a horizontal spacing of 500 kilometers or less. A subset of these measure- ments should reach 10 hectopascals at least once a day with a horizontal spacing of 1,000 kilometers or less (WMO 2019). All of these data should be shared internationally. At the time of writing, however, only 20–25 percent of WMO Members are compliant with GBON; 25–30 percent are not in full compliance because, although they have data, these are not internationally exchanged; and the rest are noncompliant primarily be- cause of a lack of resources. Addressing this third category, which represents nearly half of WMO Members, is the Rarely are systematic archiving target of many development partners, especially because of the connection that exists of model outputs, forecast between local observations and the local quality of NWP produced by the global produc- tion centers. The more local data are shared internationally, the better the output of the verification, and model post- global production centers that can be applied locally. The benefit of sharing national data processing and calibration with global NWP centers is exemplified by Ukraine, which recently increased the number considered and operationalized. of observations available to the European Centre for Medium-Range Weather Forecasts (ECMWF) from 30 to 130 weather stations (ECMWF 2018). These extra data help im- prove global forecasts, which, in turn, helps improve regional and national forecasts. In particular, near-surface temperature and humidity observations improve estimates of soil moisture, which influence forecasts of near-surface temperature and rainfall. Extra snow depth data from Denmark, Hungary, the Netherlands, Romania, Sweden, and Swit- zerland have improved two-meter temperature forecasts. Lack of data sharing can also be traced to missing information technology (IT) capabili- ties within NMHSs. Comparing salaries of public and private sector employees explains why so many IT experts get trained, often through development projects, and leave for better remuneration in the private sector. Contracting this type of work to the private sector might in many cases be a more sustainable approach. Putting more data into the international exchange and doing it more frequently leads to direct multiple benefits for the country, including an overall improvement in mod- el performance, better local forecasts, and improved forecast verification, which helps monitor, improve, and compare forecast quality and forecasting systems. Local Deterministic Modeling Versus Global Ensemble Prediction Frequently, NMHSs prioritize the development of their own national local limited-area deterministic NWP modeling systems that use in-house high-performance computing. Today it is difficult to justify this approach. NMHSs tend to run outdated versions of de- terministic NWP models on workstations at spatial resolutions much lower than those currently available from the global centers, and with no ability to assimilate local obser- vational data into the models. A global or regional modeling system will provide better results because such a system is fed new data every six hours, whereas local area models either run for 12 hours without new data from outside their boundaries or without data assimilation at all. Rarely are systematic archiving of model outputs, forecast verification, and model post-processing and calibration considered and operationalized. This makes it difficult to assess the costs and benefits of model skill improvement over time. [ 6 ] Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services An important and continuing development for operational weather forecasting is the use of EPS, which are capable of providing uncertainty information associated with NWP results. EPS are a powerful tool in the prediction and early detection of severe weather events. The value of these systems is often lost on forecasters in developing countries, who focus on model resolution rather than the underlying reliability of the forecasting systems. While higher resolution is important, it should not come at the loss of reliability. As Palmer (2019) points out, since deterministic forecasts are by their nature unreliable because of the intermittent butterfly effect—whereby small errors in the initial state can lead to explosive error growth—a small number of such deterministic forecasts will always be badly wrong. This compromises the trustworthiness of the forecasting system for decision making. Increasing the resolution of a deterministic model may exacerbate this problem because small initial errors will be better resolved and may grow faster and larger in the simulation. One example is a comparison of the ECMWF Extreme Forecast Index (EFI), which mea- sures the level of extremity of a given ensemble forecast to the model climate, and an operational high-resolution deterministic forecast using the WRF model (Weather Re- search and Forecasting Model) for Sri Lanka. In this example, the EFI indicates a high risk of an extreme rainfall event, which is not evident in the deterministic forecast (Figure 1). The resulting impact was a severe flood event caused by over 300 millimeters of rain, which fell over the northwest of the country in 24 hours. Figure 1. A forecaster’s view of the weather situation using products from the 3-kilometer mesoscale model of 24-hour rainfall from the Indian Met Department (left) and ECMWF Extreme Forecast Index (EFI) and Shift of Tails (SoT)—which is a measure of the difference between the tails of the ensemble forecast and the model climate—for total precipitation (right). The EFI and SoT indicate that an extreme rainfall event is highly likely based on the climatology of extremes and the ensemble distribution. This is not apparent in the 3-kilometer run, which, although it shows high rain, would not be considered extreme. This highlights the point that high-resolution deterministic forecasts are not “better” than lower-resolution ensemble prediction systems (EPS). The individual members of the ensemble (not shown) would also not have indicated extreme rainfall, but by looking at the behavior of the distribution and comparing it with past extreme rainfall events, the forecaster should have been alerted to the possibility of a high-impact rainfall event. Given the severity of the flood impact, much more effort is needed to transform forecast and warning systems to more reliable, impact- and probability- (risk) based services. Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services [ 7 ] This is not to say that higher resolution is bad, but rather it should be focused on the EPS. The route to higher-resolution forecasts is through limited-area model ensembles embedded into the global ensemble (Palmer 2019). A similar approach would apply to hydrological models. Palmer also points out that the added detail provided by a single high-resolution deterministic model run, which cannot by definition be supported by the corresponding ensemble, must be unreliable and therefore not useful for users. Absent a limited-area model ensemble, aspects of the high resolution that depend on the low- er-resolution elements of the flow could be produced by statistical downscaling, includ- ing, increasingly, post-processing methods based on machine learning. It is important to shift the focus from “traditional” business practices in order to em- brace modern approaches to numerical prediction. The transition to ensemble predic- tions should be supported since they provide valuable information about the likelihood of an event, and—coupled with information on the potential impacts—provide users with warnings of the severity risks. Overcoming Inertia The problem we are trying to address is not new or unique to the fields of meteorology and hydrology. The development objective is to help countries protect the lives and livelihoods of their citizens no matter what their level of economic development. How we try to transfer or distribute intellectual, technical, and financial capital is at the heart of problem. As we said above, self-determination is a strong motivator for many developing coun- tries, and this is appropriate in many fields. As evinced by the International Meteorolog- There is a justifiable concern ical Organization (IMO) in 1873, meteorology and weather forecasting, in particular, has always been a participatory science where every country depends on the internation- about depending on international al sharing of data and knowledge. Given the enormous resources required to develop, sources for basic information, but launch, and operate satellites, it is understandable that only a few countries have the as we have said, this is the nature capacity to do so. But this does not impede the use of their data, at least the data from the publicly operated satellites. Similarly, only a few centers have the computational of meteorology and fundamental capacity and deep pockets needed to develop and maintain high-quality global NWP to its success. systems; these data are also available, albeit not necessarily for free or at the resolution available at the source. There is a justifiable concern about depending on international sources for basic infor- mation, but as we have said, this is the nature of meteorology and fundamental to its success. It is also a field that is not dominated by a single country, and international pub- lic sector consortia are common. Europe has perhaps the most-advanced arrangements through entities such as its satellite (European Organisation for the Exploitation of Me- teorological Satellites, or EUMETSAT), observing systems (EUMETNET), and modeling (ECMWF) groups. Similarly, where limited-area models are used, they are often done by regional consortia because of the cost and difficulty of operating and maintaining these computer codes by individual countries. [ 8 ] Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services Conclusions Developing countries must have the opportunity to provide a level of service that pro- tects the lives, livelihoods, and economic opportunities of citizens at least as well as the most advanced societies. By focusing on protecting people rather than trying to close technological gaps through direct technology transfer, we may achieve this ambitious goal, which would align and support the UN Sustainable Development Goals (United Nations 2015) and the Sendai Framework for Disaster Risk Reduction (UNDRR 2015). This all requires a much greater level of understanding and cooperation between public, private, and academic sectors across the whole of the so-called Global Weather Enter- prise (GWE) (Thorpe and Rogers 2018) to exploit fully the strengths of each. It requires a close inspection and analysis of the entire hydrometeorological value chain (GFDRR 2020). The simplified value chain enables each of the elements to be analyzed in the context of how the entire system operates rather than as distinct and unique activities (Figure 2). Institutional capacity Research and development Education and training Numerical Issue Business Generate Tailored Observations weather official data forecasts services prediction warnings aggregation Data aggregation; data and information dissemination Infrastructure; specifically power and ICT Source: GFDRR 2020. Figure 2: The simplified hydromet value chain, which is used to assess the relative importance of the contributions of the public, private, and academic sectors (GFDRR 2020). This may also guide an overall analysis of the integrity and development of the entire system. We need to start rethinking how the entire meteorological system and the services it provides works and convincing development partners and countries to address the is- sues raised in this Note. Fundamentally it will require leadership from developing coun- tries that is informed, technically competent, and confident enough to make the appro- priate decisions. It will require a greater role for individual countries in the governance of regional and global production centers and, despite heroic efforts led by WMO,1 a far greater level of technical and intellectual cooperation among countries than has oc- curred hitherto—or at least since the beginning of the modern era of meteorological forecasting and warning services. The WMO Severe Weather Forecasting Programme (SWFP) is an important example that needs to be 1 developed to scale; see https://community.wmo.int/activity-areas/severe-weather-forecasting-pro- gramme-swfp Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services [ 9 ] One goal that needs to be reached, if developing countries’ NMHSs are to develop re- gional ensembles from the global EPS, is perpetual access to the full set of high-quality ensemble forecast data such as those from ECMWF at no cost. If a high-quality regional ensemble can be created from the global ensemble, regional scientists would have a focus for their research and development work. Comparing the skill of such a regional ensemble with the global ensemble would be valuable for global centers such as ECM- WF, too. So, it would benefit both. It would be much easier to embed the regional impact models in the regional ensemble than in the global ensemble. It must also be recognized that there is a cost to the global centers for them to meet their full potential, requiring a creative financing mechanism through investment from the global community of users or development partners or both. Bibliography BBC. 2015. “The Birth of the Weather Forecast.” BBC News. April 30, 2015. https:// www.bbc.com/news/magazine-32483678 ECMWF (European Centre for Medium-Range Weather Forecasts). 2018. “Extra Weath- er Station Data Improve ECMWF’s Forecasts.” ECMWF, June 18, 2018. https://www. ecmwf.int/en/about/media-centre/news/2018/extra-weather-station-data-im - prove-ecmwfs-forecasts 7 GFDRR (Global Facility for Disaster Reduction and Recovery). 2020. The Power of Partnership: Public and Private Engagement in Hydromet Services. Washington, DC: World Bank. https://www.gfdrr.org/sites/default/files/publication/The%20 Power%20of%20Partnership_WEB.pdf Hallegatte, S. 2012. “A Cost Effective Solution to Reduce Disaster Losses in Developing Countries: Hydro-Meteorological Services, Early Warning, and Evacuation.” Policy Re- search Working Paper 6058, World Bank, Washington, DC. Hallegatte, S., A. Vogt-Schilb, M. Bangalore, and J. Rozenberg. 2017. Unbreakable: Build- ing the Resilience of the Poor in the Face of Natural Disasters. Climate Change and Development Series. Washington, DC: World Bank. Kull, D., L. P. Riishojgaard, J. Eyre, and R. A. Varley. Forthcoming. “The Value of Sur- face-Based Meteorological Observation Data.” World Bank, in review. Palmer, T. 2019. “The ECMWF Ensemble Prediction System: Looking Back (more than) 5 Years and Projecting Forward 25 Years.” Quarterly Journal of the Royal Meteorological Society 145 (S1): 12–24. Rogers, D. P., H. Kootval, and V. V. Tsirkunov. 2018. “Early Warning, Resilient Infrastruc- ture and Risk Transfer.” In Bridging Science and Policy Implication for Managing Cli- mate Extremes, edited by Hong-Sang Jung and Bin Wang, 65–79. World Scientific Series on Asia-Pacific Weather and Climate, Volume 10. Singapore, Hackensack, NJ, and London: World Scientific Publishing Co. and Busan, Korea: APEC Climate Center. Rogers, D. P., V. V. Tsirkunov, H. Kootval, A. Soares, D. Kull, A.-M. Bogdanova, and M. Suwa. 2019. Weathering the Change: How to Improve Hydromet Services in Develop- ing Countries. Washington, DC: World Bank. [ 10 ] Mind the Gap: Addressing Critical Technical Issues in Strengthening National Hydrometeorological Services Singh, R., S. P. Ojha, C. M. Kishtawal, and P. K. Pal. 2014. “Impact of Various Observing Systems on Weather Analysis and Forecast over the Indian Ocean.” Journal of Geo- physical Research: Atmospheres 119 (17): 10,232–46. Thorpe A. and D. Rogers. 2018. “The Future of the Global Weather Enterprise: Oppor- tunities and Risks.” Bulletin of the American Meteorological Society. https://journals. ametsoc.org/doi/full/10.1175/BAMS-D-17-0194.1?mobileUi=0. United Nations. 2015. Transforming Our World: The 2030 Agenda for Sustainable De- velopment (A/Res/70/1). United Nations. https:/ /sustainabledevelopment.un.org/ content/documents/21252030%20Agenda%20for%20Sustainable%20Develop- ment%20web.pdf. UNDRR (UN Disaster Risk Reduction). 2015. Sendai Framework for Disaster Risk Reduc- tion 2015-2030. https://www.unisdr.org/files/43291_sendaiframeworkfordrren.pdf. WMO (World Meteorological Organization). 2015. WMO Guidelines on Multi-Hazard Impact-Based Forecast and Warning Services. WMO TD no. 1150. Geneva: WMO. https://www.wmo.int/pages/prog/www/DPFS/Meetings/ET-OWFPS_Montre - al2016/documents/WMOGuidelinesonMulti-hazardImpact-basedForecastandWarn- ingServices.pdf. WMO (World Meteorological Organization). 2019. GBON: The WMO Global Basic Ob- serving Network. https://community.wmo.int/gbon. Zillman, J. W. 2017. “International Cooperation in Meteorology, Part 1: Origin and Early Years.” Weather 73 (2). Credit: aapsky. Elements of this image furnished by NASA. The Global Facility for Disaster Reduction and Recovery (GFDRR) is a global partnership that helps developing countries better understand and reduce their vulnerabilities to natural hazards and adapt to climate change. Working with over 400 local, national, regional, and international partners, GFDRR provides grant financing, technical assistance, training and knowledge sharing activities to mainstream disaster and climate risk management in policies and strategies. Managed by the World Bank, GFDRR is supported by 34 countries and 9 international organizations. www.gfdrr.org