How climate forecasts strengthen food security

The Famine Early Warning Systems Network (FEWS NET) is a decision-support activity that produces timely, evidence-based information on food insecurity. It supports humanitarian-response programming while helping to reveal the root causes of food insecurity around the world.

FEWS NET was created by the US Agency for International Development (USAID) in 1985, immediately following famines that claimed over a million lives in Africa. It is a response to the need for better and earlier warnings of the potential risk that such human catastrophes may recur. Since then, FEWS NET has developed a body of practices, data, and information that directly contributes to Sustainable Development Goal 2 and the challenge of ending hunger and achieving food security for all.

Since the livelihoods of food-insecure populations are often climate-sensitive, climate services and applied climate science form an important component of FEWS NET. They complement the close monitoring and analysis of markets and prices, nutrition and disease, livelihoods, government policies and conflict.

Developing consensus agroclimatology assumptions

Every month, FEWS NET updates its eight-month Food Security Outlooks (www.fews.net) through a rigorous, structured scenario-development process. In order to contribute to the agroclimatological component of theoutlooks, food-security analysts and their science partners use a comprehensive procedure for developing working assumptions about agroclimatology that are used to build scenarios for food-security outlooks. This is a three-part process in which (1) food security analysts draft preliminary agroclimatology assumptions; (2) the draft analyses are reviewed by climate scientists in light of available scientific evidence; and (3) the reviewed assumptions are presented to the analysts.

In the initial step of this procedure, the analysts develop draft assumptions about how agroclimatology could potentially impact the food-security scenarios under consideration. These assumptions are based on a preliminary analysis of publicly available climate and forecast information. They are region-specific, focusing on the most vulnerable and high-impact areas, and covering all of the FEWS NET countries.

An eight-month forecast would be desirable in order to provide an extended lead-time, but most long-range forecasting systems have a comparatively short reach. FEWS NET has therefore developed an operational approach utilizing multiple lines of evidence for building agroclimatology assumptions. A core objective of this approach is to provide forward-looking information for a range of timeframes up to several months in advance and to continuously increase the accuracy of this information as the lead time decreases.

It works as follows. When an agroclimatology assumption is required for a time period when no other information is available – say six to eight months ahead – assumptions are based on historical climatology, for example typical variability and trends in rainfall or temperature. For shorter lead times, climate modes such as the El Niño-Southern Oscillation (ENSO) are used to derive assumptions. These are based on the known relationships between these climate modes and the response of seasonal climate in many areas.

As time progresses, lead times up to the food-security outlook period become shorter. At lead times on the order of three to six months, available long-range climate forecasts that are updated monthly are also used for building assumptions. Finally, with the onset of the rainfall season, seasonal monitoring data are progressively incorporated into the analysis. Together with short and long-range forecasts, these data make increasingly accurate assumptions more feasible.

Here are two examples of assumptions drafted in June 2017 that illustrate how this works:

Rainfall during the remainder of the March to June 2017 Gu/Diraac/Sugum rains in Somalia is expected to be minimal. Given seasonal progress to date, cumulative rainfall totals for the season are likely to be below-average to well-below-average in many central and southern areas of the country.
The start of the 2017/2018 rainy season across southern Africa is likely to be normal, with most model forecasts favoring the continuation of ENSO-neutral conditions through 2017. Rainfall during the October to December 2017 and January to March 2018 periods is likely to be average in most of the southern Africa region. This is a more frequent occurrence in areas during ENSO-neutral conditions.

Assumptions like these have key implications for future food security for various reasons. For example, Assumption 1, with the expectation for well-below-average seasonal rainfall totals, can imply negative outcomes for rain-fed crop and livestock production systems. Given that in this specific case the rainfall season is almost over (in June), there is little likelihood for change from this expected outcome. The low uncertainty makes it possible to generate strong early-warning statements that can clearly guide decision-makers towards appropriate responses.

In the case of the second assumption, the start of the season has implications for various food-security components such as labor opportunities (for ploughing and planting), harvest potential based on the timing of planting, and rainfall outcomes based on climatology. All of this information, appropriately packaged, enables decision-makers to begin formulating preliminary, but adjustable, intervention options that can mitigate the severity of various scenarios of food insecurity.

As the agroclimatology assumptions are initially developed by non-climatologists for specific food-security applications, their review by climate scientists using the best available information helps to ensure the scientific accuracy of the assumptions. A key objective of this approach is to build better and more accurate linkages between science and practice. To achieve this, the draft assumptions are forwarded to science partners at the US National Oceanic and Atmospheric Administration (NOAA), the University of California, Santa Barbara (UCSB), the US Geological Survey (USGS) and FEWS NET field scientists. These scientists undertake a rigorous review of the assumptions, initially with each scientist providing a detailed analysis of a specific region, followed by presentations and discussions via teleconference, leading to a consensus set of reviewed assumptions.

The assumptions are reviewed in light of climatological analyses of potential outcomes based on historical data; climate modes such as ENSO, Indian Ocean Dipole (IOD) and the Subtropical Indian Ocean Dipole (SIOD), among others; publicly available long-range seasonal forecasts; and analyses of current conditions based on a diverse suite of remotely-sensed products and models and of field information. Novel user-driven scientific research and products have often been inspired by and pursued as a consequence of the FEWS NET development process for working assumptions. Findings from such research then feed back into FEWS NET practice, strengthening the process.

To facilitate robust climatological analysis, FEWS NET developed the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) data set. This operational, high-resolution, quasi-global precipitation dataset spans the years back to 1981 (Funk et al. 2015). It has enabled a deeper understanding of rainfall characteristics such as variability, expected amounts, and trends in areas of interest. Long-term datasets such as CHIRPS have also facilitated the study and characterization of the regional impacts of climate modes such as ENSO, IOD and SIOD. FEWS NET scientists have published several journal articles detailing these impacts (Hoell et al. (2015), Funk et al. (2016), Hoell et al. (2017)), ultimately helping to improve the process for producing agroclimatology assumptions.

FEWS NET, in collaboration with the Southern African Development Community (SADC) Climate Services Centre, has also developed the GeoCOF, a statistical seasonal-forecast development software tool that streamlines the process for generating forecasts. It is used by regional climate centres in their WMO regional climate outlook forums (RCOFs). This initiative has helped to improve the generation of regional forecasts used for the assumptions. Finally, a suite of web-based and off-line tools has been developed and published to facilitate user-friendly access to the data. This includes the USGS/FEWS NET Portal, which hosts several analytical tools and satellite-based datasets.

A key principle for FEWS NET is to rely on the convergence of evidence. Higher agreement between different, independent sources of evidence-based information provides greater confidence in analytical outcomes. To this end, seasonal forecasts from various national, regional and international centres are compared and evaluated as part of the process for reviewing assumptions. Information relaying the uncertainty and the skill of the various forecasts helps considerably in this evaluation. Forecasts with detailed information regarding their uncertainty or underlying principles are generally given greater consideration.

Contextual interpretation is important for evaluating the different data streams. To facilitate this, FEWS NET has developed, and continuously updates, a knowledge base of local vulnerabilities with respect to agroclimatology, livelihoods, markets and trade, and nutrition. This knowledge base contributes to the identification of high-probability events and high impacts in different geographical areas and specific sectors.

After the consensus-reviewed assumptions have been adopted, a follow-up meeting is held with the food-security analysts. At this meeting the reviewed assumptions are presented, together with the scientific evidence supporting the revisions. Analysts are given the opportunity to request clarifications and added inputs. This increases their understanding of the climate system and ultimately enhances the usage of climate science in food-security applications. The finalized, regional-level assumptions are then disseminated to field-based food-security analysts as guidance and for incorporation into country-level agroclimatology assumptions.

The 2010/2011 and 2016/2017 East Africa droughts

Several recent examples illustrate the benefits of the FEWS NET process. During the past decade, FEWS NET, working with its collaborating network partners, has been able to successfully provide food-security early-warning advisories with a six-to-eight months month lead-time. These advisories have supported appropriate responses and contingency planning in East Africa during the recent devastating droughts of 2010/2011, 2015/2016 and 2016/2017.

Figure 2. Integrated Food Insecurity Phase Classification (IPC) Maps showing the levels of food insecurity for (a) July-September 2011 and (b) June-September 2017

The 2010/2011 drought was the worst on record over the eastern Horn of Africa in the past 60 years. It had severe impacts in Somalia, Ethiopia and Kenya. Using an analysis of the ENSO climate mode, available long-range climate forecasts, and an interpretation drawing on local knowledge and prevailing food security conditions, FEWS NET was able to issue a food-security early-warning report by August 2010.

Sadly, conflict obstructed an effective response, resulting in over 250 000 human deaths in Somalia. Over 12 million people in East Africa required urgent humanitarian assistance. The severe and prolonged drought had adverse socio-economic, environmental and political impacts in the worst affected regions of the eastern Horn, more specifically in Somalia, Kenya and Ethiopia. Overall, US$ 1.7 billion was provided to mitigate this human catastrophe, representing 71% of the United Nations appeal for US$ 2.4 billion in early 2011.

Similarly, FEWS NET was also able to predict the devastating 2015/2016 drought in Ethiopia with adequate lead-time. More recently, it predicted the ongoing and unprecedentedly severe 2016/2017 drought, which resulted in about 27 million people (June 2017) requiring urgent food assistance. It also led to a UN appeal for US$ 4.4 billion in funding, twice the 2010/2011 appeal. The 2016/2017 drought was relatively more widespread, extending from the eastern Horn into the western sector of the region. It adversely affected main staple food-production zones of Kenya and Uganda, and into Tanzania. The improved early-warning systems and multi-agency responses employed during this drought, as well as improved humanitarian access thanks to less-adverse patterns of conflict, meant that famine was averted, unlike the case in 2010/2011 in Somalia.

USGS/FEWS NET and UCSB scientists and FEWS NET field staff in East Africa, were able to predict and continuously monitor key climatic drivers (modes). As a result, they were able to provide early warning of the impending severe droughts with adequate lead-times of six-to-eight months. This was possible owing to the use of innovative agroclimatic monitoring indicators and decision-support tools, backed by a strong field-science and applied-research teams. The use of remotely-sensed agroclimatic data facilitated the continuation of this work under very challenging and volatile conflict situations. There was limited access to critical information on field assessments, which, where available, facilitated the timely and comprehensive analysis of agroclimatic conditions.

The 2015/2016 Southern Africa drought

The 2015/2016 rainfall season in southern Africa was one of the driest in over 35 years. This had significant adverse impacts on agriculture and food security. The previous season (2014/2015) had been characterized by the erratic onset of rains, extended dry spells and abnormally high temperatures. This led to reduced yields in eight of the 13 mainland countries of the Southern African Development Community. Poor rainfall over the past few seasons had also caused poor grazing conditions and low water supplies for livestock, resulting in deteriorating body conditions of the livestock. The regional cereal deficit stood at 7.9 million metric tonnes below what was required, maize prices across the region were above the US yellow maize prices, and 27.4 million people in the region were determined to be food insecure, 13% more than in the previous year (2013/2014). Food security conditions were therefore already compromised going into the 2015/2016 season.

Figure 3. Map showing areas where the 2015/16 seasonal rainfall totals were the driest or wettest since 1981. The map was effectively used to communicate the severity of the drought to decision makers.

As early as June 2015, FEWS NET’s consensus agroclimatological assumptions were for below-average rainfall in the southern parts of the SADC region. This was based on the then-current and the forecasted ENSO state, with predictions for a strong El Niño in 2015/2016. FEWS NET’s July 2015 Food Security Outlook stated that late onset of rains associated with El Nino conditions could limit labor opportunities, thus worsening food insecurity. By August, most major global models were forecasting below-average rainfall in southern Africa, thus strengthening confidence in FEWS NET’s agroclimatological assumptions. The Southern African Regional Climate Outlook Forum (SARCOF) forecast for normal to below-normal rainfall, released in late August 2015, further cemented confidence in a poor rainfall outcome.

As assumed, the 2015/2016 season turned out to be dry. Analysis using the CHIRPS precipitation dataset showed that many areas had received their lowest amount of rainfall since 1981. The consensus seasonal forecast review, monthly FEWS NET Food Security outlook updates, and monthly seasonal monitor reports co-produced with the SADC Food Agriculture and Natural Resources Directorate as the SADC Agromet Update, were used to monitor and report on the evolution of the drought.

By January 2016, nearly five months before the end-of season crop harvests, analyses of seasonal rainfall to date and historical El Niño rainfall performance led to a consensus FEWS NET agroclimatological assumption that crop performance in most southern Africa countries was likely to be below average. National reports, which provide critical field information, corroborated many of the observations made using satellite- and model-based analyses. The February 2016 SADC Agromet Update reported on the largely decreased planted area, long dry spells, and high temperatures that were impacting the region’s harvest prospects.

Based on their internal national assessments, individual countries started issuing Drought Emergency Declarations as early as December 2015. Some countries, such as South Africa and Zimbabwe, began arranging for grain imports by January. In early February 2016, the World Food Programme (WFP), the Food and Agriculture Organization of the UN (FAO), FEWS NET, and the European Commission Joint Research Centre released a joint statement entitled “El Niño Set To Have A Devastating Impact On Southern Africa’s Harvests And Food Security”, garnering high international visibility.

In late February, SADC held a Consultative Meeting on Preparedness and Response to the Impact of the 2015/2016 El Niño on Agriculture and Food and Nutrition Security, with the support of FAO and WFP. In March, the SADC Council of Ministers recommended that a regional drought disaster be declared. A SADC El Niño Logistics and Coordination Team was established at the SADC Secretariat. In June 2016, SADC launched a Regional Humanitarian Appeal, with an opening statement by the SADC Chairperson, the President of Botswana. The Appeal document stated that the 2015/2016 drought was the worst in 35 years and, together with other factors, had resulted in approximately 40 million people in the region requiring humanitarian assistance, at a cost of approximately US$ 2.4 billion.

This history provides clear evidence that the streamlined assumption building processes and agroclimatological monitoring and analyses that were undertaken by early warning agencies during this drought event led to effective early warning, sound decision-making, and timely action that mitigated the overall impact of the drought.

Communication, collaboration, capacity

The successes of FEWS NET’s early warning information systems during recent severe droughts have made an important contribution to the implementation of the Sustainable Development Goals, and in particular to SDG 2. Several factors contributed to this. Key among these has been a close working relationship between climate scientists and food security analysts, enabling scientists to respond to the analytical needs of food-security experts. These effective communication systems have facilitated the production of well-targeted briefs, reports and web-based interventions delivered to high-level decision-makers in both donor countries and organizations and in the affected countries.

Strategic network partnerships with lead national agencies and regional and international stakeholders have been instrumental to developing food-security outlooks. Continuous capacity building for institutions has strengthened these networks and supported sustainable technology transfer. Innovative climate datasets such as CHIRPS together with their accompanying decision-tools have enabled the development of a solid local knowledge base covering prevailing agro-climatic trends, food-security risks, and vulnerabilities to climate variability and change.

These datasets and tools have also supported reliable seasonal climate forecasting and its contextual interpretation. They have strengthened the ability to generate area-specific, agro-climatic assumptions, high-resolution monitoring and the tracking of climatic extreme events based on independent data-streams. This capacity has been backed by regular field assessments. The result has been inter-agency consensus updates on food security status and outlooks with six-to-eight month lead-times for early-action by response and contingency-planning agencies.

FEWS NET’s linkages with international and local universities and research and development agencies also allow it to leverage applied research, promoting the continual development of relevant products. Strong agro-climatic early warning and monitoring systems are supported by strategic science partners at the international and national levels.

FEWS NET’s suite of remotely sensed products and model outputs is available at:

References

Funk C, Harrison L, Shukla S, Korecha D, Magadzire T, Husak G, Galu G, Hoell A. 2016. Assessing the contributions of local and east Pacific warming to the 2015 droughts in Ethiopia and Southern Africa. Bulletin of the American Meteorological Society 97:S75-S80.

Funk C, Peterson P, Landsfeld M, Pedreros D, Verdin J, Shukla S, Husak G, Rowland J, Harrison L, Hoell A. 2015. The climate hazards infrared precipitation with stations--a new environmental record for monitoring extremes. Scientific data 2:150066.

Hoell A, Funk C, Magadzire T, Zinke J, Husak G. 2015. El Niño–Southern Oscillation diversity and southern Africa teleconnections during austral summer. Climate Dynamics 45:1583-1599.

Hoell A, Funk C, Zinke J, Harrison L. 2017. Modulation of the Southern Africa precipitation response to the El Niño Southern Oscillation by the subtropical Indian Ocean Dipole. Climate Dynamics 48:2529-2540.