Food and water security, globally, is susceptible to climate-related risks due to climate variability and change (Schmidhuber and Tubiello, 2007; Milly et al., 2008; Falkenmark, 2013; Wheeler and von Braun, 2013; Campbell et al., 2016). Annual economic losses due to climate extremes are estimated to be at about $100B (US Dollars) between 2004 and 2014 (FAO, 2016). Mitigation of the adverse socio-economic impacts of climate extremes, as well as long-term climate resilience and adaptation, is critically dependent on timely access to Earth observations (EOs). These EOs include satellite-based precipitation, temperature, and vegetation greenness, as well as modeled data sets including reanalysis products, simulated hydrologic fluxes, and dynamical sub-seasonal to seasonal climate forecasts. During the past three decades, major strides have been taken in the development and operational production of satellite and model-based EOs, often spanning the entire globe. These efforts have improved monitoring of the Earth for scientific and decision-making applications (Wulder and Coops, 2014). For example, availability (now open access) of 40+ years of Landsat data sets has facilitated routine monitoring at a global scale and long-term ecological change assessments (Wulder et al., 2012). Availability of high-resolution (10–60 m) multi-spectral imagery from the Sentinel series (Donlon et al., 2012; Drusch et al., 2012; Torres et al., 2012; Szantoi and Strobl, 2019) has allowed for potential application of EOs in decision-making at a truly local scale. Satellite-based estimates of precipitation (Huffman et al., 2010; Ashouri et al., 2015; Funk et al., 2015a; Skofronick-Jackson et al., 2017), evapotranspiration (Jiang et al., 2020; Senay et al., 2020), surface soil moisture (Brown et al., 2013; Dorigo et al., 2017; Das et al., 2018), and total moisture—derived from monthly gravity field estimates made by the twin Gravity Recovery and Climate Experiment (Tapley et al., 2004; Rodell et al., 2018)—have allowed for improved understanding of change in water availability, and for monitoring extreme events such as drought and floods. Additionally, model-based (and assimilated observational estimates) climate reanalysis (Saha et al., 2010; Hersbach and Dee, 2016; Gelaro et al., 2017) and weather (Hamill et al., 2006) to subseasonal (Pegion et al., 2019) to seasonal scale (Molteni et al., 1996; Kirtman et al., 2013) climate or hydrologic forecasting products (Arsenault et al., 2020) allow for the early warning of climate extreme events, and hence, support decisions that save lives and resources. However, despite the availability of these and several other EOs, application of these data sets for decision-making in some of the most vulnerable regions across the globe, such as sub-Saharan Africa (SSA), has still been limited. The Early Warning eXplorer (EWX) has been developed and implemented to address this important limitation. The EWX provides easy and routine access to critical EOs with the primary goal of enhancing their application for disaster mitigation and supporting long-term resilience. Here, we provide an overview of the EWX, describe its features, and demonstrate examples of how the EWX can be used to empower decision-making, especially in the vulnerable regions of SSA.

The EWX is implemented based on three main components (Figure 1). The first component, the EWX Viewer, is accessible to users through a browser (the EWX overview). The viewer is the interface through which users can visualize EOs in the form of maps (spatial representation of gridded EOs) and time series graphs, as well as extract EO data for desired focus domain and time period for further analysis.

The second component is the GeoEngine processing server, which is the core processor of the EWX System. The primary functions of the GeoEngine include:

The third component, Support Applications, contains applications including a map server, a GeoServer instance, and a PostgreSQL database for storage and retrieval of the statistics and the file system, which is continually monitored for new or deleted EOs. GeoServer is an open-source geospatial data server that implements Open Geospatial Consortium (OGC) standards, serves maps to the EWX Viewer and other GeoEngine clients, and allows for serving of data via different web services such as Web Map Service (WMS), Web Feature Service (WFS), Web Coverage Service (WCS), and Web Mapping Service with Time (WMST). A PostgreSQL database stores derived data sets and statistics, maintains metadata about calculated statistics (for example, when they were last calculated), and maintains an inventory of coverages and their associations, periodic data sets, and vector data sets.

Currently, multiple versions of the EWX exist, and they are listed in Table 1. Each of these versions is based on the above-mentioned framework. The U.S. Geological Survey's (USGS) EWX is the first instance of EWX. The USGS also hosts EWX-lite, which is a limited version of the EWX designed for regions with poor internet connections. The University of California, Santa Barbara's (UCSB) EWX provides access to several EOs that are currently “experimental.” The Regional Centre for Mapping of Resources for Development (RCMRD) EWX is the first instance of EWX in SSA. As described in detail in section concluding remarks, this version of EWX was implemented to support capacity building in the region of eastern and southern Africa. It is expected that, depending on users' EO needs, they can decide to access EOs using any of the versions of the EWX. There are several EOs that are common to all versions of the EWX. If a user wants to access those common EOs, they can choose any EWX version. If a user wants to access “experimental” EOs that are not available on the USGS's EWX version, they can access them on UCSB's version. Similarly, users from eastern and southern Africa may want to access EOs from RCMRD's version of EWX, especially if they want to access EOs at custom zones (e.g., Administrative-3 polygons or grazing area polygons) that are not available on USGS and UCSB's versions. The variety within these different instances of the EWX highlights the flexibility of the platform to respond to the needs of specialized communities.

Additionally, the USGS's instance of EWX is available for different regions of the globe. Currently (2020), the regions for which EWX is available are (shown in Figure 2): (i) Global, (ii) Africa, (iii) East Africa (EA), (iv) Central America, and Caribbean, (v) Central Asia, (vi) Middle East, and (vii) South America. Different regional instances of EWX allow for easy and quick access to the EOs by zooming to the spatial extent of the given region. Additionally, regional instances allow for access to EOs that are most useful for the given region. For example, the regional instance for Central Asia provides access to snow depth and snow water equivalent, as snow is a prominent source of water supply in that region.

In the following sections, we provide a detailed overview of the viewer, describe several features of the viewer that facilitate visualization and access to EOs, describe select, widely used EOs that EWX provides access to, and finally we, provide examples of how EWX can be used to monitor climate extremes and perform analyses to support decision-making.

This section provides an overview of the EWX viewer, including its main components, examples of the EOs available through EWX, and select primary features for spatial and temporal mapping of the EOs. Figure 2 shows the USGS Earth Resources Observation and Science (EROS) Center's version of the EWX viewer. The left panel of this page lists all Datasets (top left panel) that are accessible to the users and Layers (e.g., Boundaries), which are available to overlay the data sets to extract spatially averaged time series. The USGS's EWX currently provides access to EOs such as precipitation estimates, land surface temperature, and vegetative greenness. These EOs are best suited to monitor meteorological and agricultural drought conditions, and examine long-term changes and trends. In addition to EOs, the EWX also provides access to Ancillary data sets, such as topography and population density based on Landscan. Topography is helpful for providing physical context for EOs (for example, high-elevation areas tend to have higher rainfall,) and population is helpful for providing social context (i.e., the number of people in a region affected by climate extreme events). The EWX allows users to access EOs in terms of absolute values, as well as absolute anomaly and standardized anomaly. EOs are also typically available for different temporal aggregation periods, ranging from the pentadal (5-days accumulations) to a 3-monthly period. Availability of EOs at different temporal aggregation periods allows for monitoring of short-term variability of weather conditions (e.g., dry and wet spells, and cool and hot spells) using pentadal and dekadal (10-days accumulations) aggregation periods, and monitoring of the season as a whole, with a 3-months averaging period. The EWX allows access to historical data, which can be located by using a drop-down calendar menu and selecting years, months, or seasons, and day, pentad or dekad during the month.

The EWX allows quick visualization of EOs in the form of maps and time series graphs. Maps show the value or anomalies of a selected EO (Figure 3), averaging period, and region at the native grid-cell resolution [which varies from 5 km X 5 km for Climate Hazards Center InfraRed Precipitation with Station data (CHIRPS) and CHIRPS-Global Ensemble Forecast System to 250 m for Normalized Difference Vegetation Index] of the EO. Time series graphs provide spatially averaged (averaged over all grid cells within a selected boundary) values or anomalies of a selected EO for the selected region and boundary (Figure 4). In the case of a variable like snow water equivalent (SWE), the graphs indicate SWE totaled over the given basin. The mapping option provides maps of absolute value, additive anomaly, and standardized anomaly (or percent anomaly, in the case of NDVI) for a given EO (Figure 3). These map images can be downloaded in PNG format. The underlying spatial data for the selected region can also be downloaded in GeoTIFF format. Thus, a user can extract the EOs in GeoTIFF format and input the data to any widely used Geographical Information System tool for further analysis, or for overlaying with other maps that are not available from the EWX. A user may also want to download data in GeoTIFF format to work with their existing mapping tools to provide maps in the standard format for their agency's bulletin. For example, the Zambia Meteorological Department (ZMD) downloads CHIRPS-GEFS (bias-corrected and downscaled weather forecasts) from the UCSB's EWX (Table 1), maps it using their own tools, and includes it in the ZMD's Crop Weather Bulletin (Shukla et al., 2020b). Maps are most useful for national, regional, and international decision-makers who may be interested in assessing the climate and/or vegetation conditions at a broad spatial scale. Maps allow for identification of “hotspots” that may be experiencing the most severe anomalies (e.g., droughts and floods) during a given event. Maps are also helpful for assessing the spatial extent (e.g., regional vs. local) of a given climate extreme. The “Spatial Locking Tool” option, highlighted in Figure 3, is useful for comparing maps of (i) absolute magnitude, anomaly, and Z-score for a given time period for a given EO, or (ii) a given EO for different time periods, or (iii) different EOs as well as different time periods (Figure 2). Typically, maps are used as a first glimpse of the climate and/or vegetation conditions, followed by time series graphs to get a more in-depth, localized assessment of the conditions and its evolution in time.

Time series graphs provide a localized, spatially averaged view of the climate and vegetation conditions within a historical perspective (Figure 4). The USGS's EWX provides access to EO time series spatially averaged over (i) Administrative Unit-1 or−2 and (ii) Crop zones in several developing regions across the globe, such as Africa, Central America, the Caribbean, Central Asia, the Middle East, and South America. For the Middle East region, the spatial aggregation is also available for a few selected watersheds. Aggregation over these boundaries allows the data to be directly relevant for decision-making, including urgent mitigation and/or long-term resilience. For example, aggregation over administrative unit-1 or−2 may allow for a targeted delivery of relief at a sub-national scale in the case of climate-driven food insecurity. Aggregation over crop zones ignores areas that are not major-producing regions at a national level, and are most useful for assessing agricultural conditions and for crop production outlooks. Figure 4B shows the time series of CHIRPS anomaly over a 3-months averaging period, over an administrative unit-1 (Manicaland, Zimbabwe). The location of this administrative unit can be seen by the “target symbol” in the map of CHIRPS anomaly, Figure 4A. The “target symbol” appears after the user selects the “Time series graph” option, as shown in Figure 4A. After selecting the EO (CHIRPS 3 monthly anomaly, in this case) and desired spatial domain, (Manicaland, Zimbabwe) a time series graph is created (Figure 4B). This graph can be linked with the map using the docking option in the time series graph, as in Figures 4A,B. This option allows quick access to time series for several different domains within the map (Figures 4A,B). The “target symbol” will move as the user selects different domains. The CHIRPS 3-monthly anomaly time series selected for Manicaland, Zimbabwe shows precipitation anomaly for each of the 3-months averaging periods in the latest “water year.” In Southern Hemisphere regions, the “water year” starts from July and ends in June. The time series graph (Figure 4B) shows a deficit of rainfall during the beginning of the rainy season. Both Oct-Dec (OND) and Nov-Jan (NDJ) seasonal anomalies are substantially below normal. It also shows a recovery in rainfall deficit late in the season, resulting in Dec-Feb (DJF) anomalies being near normal. Depending on the region and crop type, early and mid-seasonal precipitation deficit can have severe implications in terms of agricultural production and food insecurity. Figure 4C shows several different mapping options available for time series graphs (top right) that allow for the selection of any combination of years from the historical record (current and past years), as well as the climatological mean, chart types (line and bar plots), and download options (.CSV to get the underlying data and PNG to get the figure). Figure 4C also shows the 3-months anomaly of CHIRPS for several different years in the past, which can be used to quickly assess the severity of the anomalies in the current years relative to the anomalies in the past years. The user can also see values of anomalies as they move the cursor to a different 3-months averaging period. For example, in Figure 4C, the cursor is on the DJF season to display the anomalies for all years from 2013 to 14 through the current year. This option is also helpful when assessing impacts of the climate or vegetation anomalies. For example, a user can compare the anomalies of the current year with past years to identify the year with the most similar level of anomalies. This can help identify analog years for which impacts of anomalies are already known (e.g., shortfall in production or level of food insecurity), and similar impacts can therefore be assumed for the current year. For example, Figure 4C shows that, although this region experienced a drought in December-February 2019/2020 with precipitation deficit being 65.9 mm, this drought was not as severe as the past major drought in 2015-16, during which the precipitation deficit was 255.77 mm.

The USGS's EWX provides access to several different Earth observations (EOs) to support the timely analysis of climate, vegetation, and land surface conditions in order to monitor the status of natural hazards such as droughts and floods. In addition to EOs facilitating routine monitoring, EWX also provides access to forecasts (such as precipitation forecasts) to facilitate early warning (Table 1). The list of EOs available through the EWX varies based on the focus regions, as well as the version of the EWX (Table 1); however, below is a list of the prominent EOs that are common in different versions of EWX (Table 1), which cover at least quasi-global domain and are also widely used for monitoring and early warning purposes. These EOs are available for all regions through the USGS's EWX and across all versions of EWX (Table 1). It is important to mention that the EWX framework allows for the inclusion of additional gridded EOs as needed or when applicable for a given region. For example, SWE and snow depth data are available for central Asia, where snow plays a crucial role in food and water security.

The strength of EWX lies in its ready access to EOs, including maps, time series graphs, and analysis-ready data to support informed decision-making. The EWX is best suited to support decisions to mitigate impacts of climate-driven disasters, and to support long-term climate adaptation. In this section, we provide a few key examples of the application of the EWX in providing access to EOs for improved assessment of climate-driven extreme events and understanding of long-term climate variability and change.

The EWX is an effective tool for monitoring climate-driven extreme events, particularly drought events, which tend to develop more slowly than pluvial events. Drought conditions during a rainy season can lead to an increase in food insecurity, and drought monitoring during the rainy season can therefore support early warning of food insecurity. The EWX is also particularly attractive for monitoring, as it provides access to low-latency, gridded rainfall data sets, allowing for timely drought monitoring as opposed to relying on local in situ rainfall reports, which are often available with a lag of a few weeks.

In this section, using the example of the SA drought in 2015–16 (Archer et al., 2017; Funk et al., 2018; Shukla et al., 2020a), we demonstrate the value of the EWX as a drought monitoring tool. This drought event led to massive increases in food insecurity in the region. As per the Southern Africa Development Community (SADC) estimate (SADC, 2016), 23 million people in the region needed emergency food assistance.

Early warning of drought to support food insecurity starts several months before a season (Funk et al., 2019). In advance of a rainy season, the main focus is given to large-scale climate and sea surface temperature (e.g., El Niño-Southern Oscillation) conditions. Based on the large-scale climate anomalies, assumptions are made for how the rainy season will evolve. Prior to the start of 2015/16 rainy season in SA, as per the International Research Institute for Climate and Society (IRI) and Climate Prediction Center (CPC), who provided official El Niño-Southern Oscillation forecasts on July 9, 2015, the chances of an El Niño persisting through February were >90%. In addition to the high probability of El Niño development, the dynamical forecasts from the North American Multi-model System (NMME) (Kirtman et al., 2013) were forecasting the El Niño to be in the “strong” category (3-months average Niño 3.4 SST anomaly > 2°C). Historically, El Niño has been linked to enhanced probability of below-normal rainfall in the SA region (Reason and Jagadheesha, 2005; Meque and Abiodun, 2015; Manatsa et al., 2017; Gore et al., 2020), and the past “strong” El Niño events, such as 1982–83 and 1991–92, resulted in severe droughts (Pomposi et al., 2018). The Southern Africa Regional Climate Outlook Forum (SARCOF-19), held in August 2015, indicated enhanced chances of below-normal rainfall in the central and southern part of the SA region. As the rainy season approaches, this context of a higher chance of below-normal rainfall is critical, as it triggers close monitoring of rainfall and drought conditions (Funk et al., 2019).

Using the EWX, rainfall anomalies can be monitored closely given the context of the rainfall outlook. For example, Figure 8 shows the monthly rainfall anomalies from October 2015 through February 2016. It is clear from these maps that close monitoring of rainfall anomalies confirmed the high chances of rainfall drought development in the region. Starting from October 2015, several of the regions in central and southern SA experienced well-below-average rainfall. Rainfall deficits particularly intensified in parts of Zimbabwe and Mozambique. Displaying the crop zone boundaries (in red) on the maps highlights the fact that crop zones experienced the most severe deficit in rainfall, which has a direct effect on crop production in a rainfed region. The December to February (DJF) period is when this region typically receives most of its rainfall, so deficits during those months of 2015/16 were particularly serious.

Post February 2016, the EWX allows for a seasonal (DJF) assessment of the climate and vegetation conditions in the region. Figure 9 shows the December-February anomalies of precipitation (Figure 9A) and LST (Figure 9B). These maps highlight (i) the regions that experienced the most severe seasonal rainfall deficit and (ii) that this rainfall deficit was also accompanied by large positive anomalies in the LST (which, in part, may be due to the rainfall deficit, but also partly due to long-term climate change). The impact of rainfall deficit and high LST is reflected in the NDVI anomaly as of the first dekad of March 2016 (Figure 9C), as several regions of central and southern countries in SA were experiencing negative NDVI anomalies. A closer look at the time series graphs shows NDVI anomalies over the selected crop zones in Mozambique (Figure 9D) and Zimbabwe (Figure 9E) highlights two key points with direct implications to food insecurity: (i) prior to the 2015/16 droughts, these regions had already experienced multiple drought events (negative NDVI anomalies during November to March), and (ii) the NDVI anomalies in 2015/16 were generally more severe than in past years, indicating a more severe drought in this region following multiple droughts. As a result of these multiple drought events, the food insecurity in these regions reached the “Crisis” phase (per Integrated Food Security Phase Classification), as indicated by FEWS NET's food insecurity maps of “current situations” released in June 2016. The “Crisis” phase of acute food insecurity is when emergency food assistance is needed. Note that, by the end of February 2016, maps and graphs showed the severe drought conditions in these regions and demonstrated that this drought followed multiple drought events in the last few years. This highlights the value of drought monitoring using the EWX as an early warning tool for food insecurity.

As another example of monitoring climate extreme events using EWX, Figure 10 shows how near-real-time precipitation maps and time series graphs can be used to monitor flooding conditions. Parts of EA experienced floods and landslides driven by extreme precipitation events. Several news outlets reported loss of lives and structures due to these extreme events, particularly during April and May. Figure 10A shows the March-May standardized precipitation anomaly map for Kenya. Comparing this map with that of the number of people displaced due to floods estimated by the United Nations Office for the Coordination of Humanitarian Affairs (UNOCHA, 2020) (Figure 10B) indicates that several of the regions with the highest number of displaced people also experienced March-May rainfall >2 standard deviations above the mean. A closer examination of the precipitation totals at pentadal scale for the administrative Unit-1 Turkana in northwest Kenya (Figure 10C) and Kisumu in southwest Kenya (Figure 10D) shows several pentads with much above-normal rainfall (blue line) than the climatological mean for these regions (red lines). This is another example of how the EWX can be used to monitor climate extremes, and, in this case, extreme precipitation totals that can lead to flooding and landslides.

In addition to monitoring climate and vegetation conditions and extreme events, the EWX can also be used for visualizing and accessing analysis-ready data for examining long-term climate variability and change. The time series option can be used to extract spatially averaged EOs for a given time period (e.g., a given season) for all years, which can allow for examination of the climate variability and change in the given EO variable, season, and region. For example, Figure 11 shows time series of precipitation for 1981-present for an administrative unit-1 in Mali (Figure 11A) for the July-September season, and in Ethiopia for March-May and October-December seasons (Figure 11B) for the 3 months of their primary rainy seasons. The time series of July-September precipitation in the administrative unit-1 in Mali shows an apparent increase in precipitation. The time series of seasonal precipitation in the administrative unit-1 in Ethiopia shows the past extreme October-December precipitation events (e.g., 1997) and how the most recent October-December season compares with past extreme events. The underlying data of such time series can be downloaded for further analysis of climate variability and change.

Finally, RCMRD'S EWX provides another important example of the value of this tool, highlighting the application of EWX in the region of eastern and southern Africa. An instance of EWX was installed at RCMRD through a recently completed SERVIR Applied Science Team (AST) project (Shukla et al., 2020b) (Table 1). A regional implementation of the EWX allowed RCMRD to add EOs that are most needed for addressing environmental challenges in the region. This implementation also allowed RCMRD to provide spatially averaged EOs for the boundaries that are most relevant for regional decision-making purposes. For example, RCMRD's EWX provides spatially averaged EOs for administrative three polygons for Kenya and Tanzania, which can help facilitate application of EOs for decisions at the ward level. Additionally, RCMRD's EWX allows access to EOs averaged over Northern Rangelands Trust (NRT) conservancies (Figure 12A) and Rapid grazing blocks (Figure 12B), and NRT grazing blocks (Figure 12C) which makes the EOs directly applicable for decision-makers such as NRT, which is a Kenya-based conservancy organization that aims to support the development of resilient communities in the region. Ever since its implementation, this version of the EWX has regularly provided access to EOs to regional users—mainly in Kenya, Tanzania, and Ethiopia (as per the usage statistics). Figure 13A shows the key usage statistics, including the number of users and number of page views for the cloud instance of the RCMRD EWX page between October 2019 through June 2020. During this time period, this page was viewed a total of 1,222 times, out of which 634 were unique views. On average, users spent about 5 min on this page. For comparison, Figure 13B shows the number of users and page views of USGS's EWX and EWX-Lite.

Climate-related risks to food and water security are on the rise globally. Regions of the globe that are prone to food and water insecurity, most of which are in sub-Saharan Africa, are not only sensitive to climate-related risks, but they also face several challenges in terms of monitoring climate extreme events, mainly due to lack of in situ observations. Major strides have been made in terms of development and real-time availability of remote sensing-based EOs; however, access to those data sets and their applications for informed decision-making is limited. The EWX is a web service that aims both to fill that important existing gap and to facilitate application of EOs for decision-making to support mitigating the most adverse impacts of climate extreme events and climate adaptation.

Here we provide a comprehensive overview of the EWX, which includes:

Additionally, it is important to emphasize that, in addition to tools such as EWX, capacity-building training is needed to further enhance the application of EOs for decision-making. For example, in addition to the EWX implementation at RCMRD, through the SERVIR AST project, four training courses on the EWX and its applications were provided throughout the region. These four training courses involved 70+ technical professionals from regional, national, climate, meteorological, hydrological, and environmental agencies, as well as representatives from basin-scale water management agencies. These training courses facilitated several applications of the EWX for supporting informed decision-making in the region; for example, RCMRD, Tanzania Ministry of Agriculture's Food Security division, SADC's Climate Service Center, and Zambia Meteorological Department have utilized the EWX for regional support (Shukla et al., 2020b).

Finally, although the focus of this manuscript has been on the SSA region, several of the EOs available through the EWX are at least quasi-global (between 50 deg north to south), and hence, the EWX is applicable to supporting decision-making in other climatically vulnerable regions of the globe, such as the regions that SERVIR—the National Aeronautics and Space Administration (NASA) and United States Agency for International Development (USAID)-funded initiative—focuses on. Implementation of the EWX in these regions would further increase access to EOs and expand their application in empowering decision-making in order to mitigate the impacts of climate extremes and support climate adaptation. Furthermore, in order to facilitate EWX's application for EO access, in future we also plan to provide an user manual and hope to describe access and analysis of forecasts (weather to seasonal scale) using EWX in a future manuscript.

Publicly available datasets were analyzed in this study. This data can be found at: https://earlywarning.usgs.gov/fews/ewx/index.html.

SS wrote the first draft of the manuscript and prepare all the figures. ML, MA, MB, GH, JR, and CF reviewed and edited the manuscript. ML and MA contributed to development of Figure 1. ML, MA, MB, and JR contributed to development of EWX. JR, GH, and CF serve as investigators on miscellaneous funds that support development of EWX. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.