Category Archives: Forecasting

Climate Hazards Center Collaborates with RCMRD and SERVIR to Facilitate Climate-informed Decision Making in Eastern and Southern Africa

Climate Hazards Center Collaborates with RCMRD and SERVIR to Facilitate Climate-informed Decision Making in Eastern and Southern Africa

Shraddhanand Shukla, Greg Husak, Juliet Way-Henthorne, & Denis Macharia

Key Takeaways:

  • The Climate Hazards Center, in collaboration with RCMRD and SERVIR, moves into year-3 of the “Enhancing Eastern and Southern Africa Climate Services by Increasing Access to Remote Sensing and Model Data Sets” project.
  • The CHC and partners utilize three methods of capacity building—introduction to web services, hands-on training, and an “empower-the-trainers” approach to maximize impact.
  • This project benefits both the trainees and the CHC through the reciprocal nature of collaborative training. The trainees are introduced to advanced EOs and web services, and the CHC comprehends the existing needs and challenges in the application of EOs and web services.

Project History: Scope and Goals

Since July 2016, the University of California, Santa Barbara’s Climate Hazards Center (CHC) has worked with the Regional Centre for Mapping of Resources for Development (RCMRD) through support from the SERVIR-Applied Sciences Team (AST) program to empower technical professionals in key regions of southern and eastern Africa by utilizing various methods of capacity building, transfers of technology, and an “empower-the-trainers” approach.  To briefly summarize the project—which will end in June 2019—the primary outcome of employing a variety of techniques to best equip trainees to integrate Earth observation (EO) information and geospatial technologies into their climate services to regional decision makers has been largely successful. Additionally, the project allows the CHC to further its own mission of protecting the lives and livelihoods of at-risk communities through hands-on trainings and demonstrations of CHC tools and techniques that allow for early warning as it relates to food security and climate. Knowledge sharing builds a wider network of capable climate service providers, which, in turn, creates a ripple effect, as these trainees share new resources and skills with decision-making stakeholders in the region.

Obtaining publicly accessible, analysis-ready EO data is challenging for decision makers across the developing world, as the ability to transform data to workable information is often a critically missing link. Data sets and models can be applied to inform food security and water resource-related decisions, helping to improve monitoring and forecasting of droughts, water availability, and climatic conditions. However, without knowledge of how to access or understand this information, these valuable data sets and models go underused.

To combat this challenge within a SERVIR-AST project, the CHC collaborated with SERVIR eastern and southern Africa hub at RCMRD to co-host this series of regional workshops in Kenya, Tanzania, and Zambia. These trainings focused on enabling local and national climate service providers to confidently and effectively utilize such data sets and tools, ultimately allowing these key individuals to make better-informed decisions.

Regional Workshops:

The first workshop, which served primarily as a framing workshop through which to gauge climate service providers ’ needs, was held in Nairobi, Kenya between September 12th and 15th, 2017. The workshop’s location was strategically chosen as a base for local attendees who represented a variety of agencies focused on different sectors of agricultural development, relief effort, and crop insurance, including Kenya Forest Services (KFS Hqs), the World Food Programme (WFP), and IGAD Climate Prediction and Applications Centre (ICPAC). Attendees were drawn from RCMRD’s mandated countries—the parameters of which aided in selecting all workshop locations. By beginning with workshops focused on a local scale before gradually building outward to incorporate more regions, the CHC was able to give an overview of several web services and case studies to further refine training materials. This introductory workshop also demonstrated the needs of a typical user of data sets and models to best assess the needs of participating technical professionals.

The second workshop (held between January 30th and February 2nd, 2018, in Dar Es Salam, Tanzania) of year-2 of the AST project trained partners in web applications to access climate and hydrologic data sets and demonstrated how to apply this information to generate seasonal climate scenarios, agricultural drought monitoring, water-resource management, and index insurance. The next workshop (held between September 4th and 7th, 2018 in Lusaka, Zambia) focused on further enhancing climate services in the region.

“Connecting Space to Village” SERVIR Hydroclimate Training in Lusaka, Zambia. Day 1 comprised of an introduction to Climate Hazards Center data sets.

Participants in the 2018 SERVIR-AST training in Lusaka, Zambia access NASA and FEWS NET’s Land Data Assimilation System (FLDAS) simulations’ spatial and time series maps & data through NASA’s GIOVANNI.

Training events showcased how to access and use analysis-ready EOs, such as the CHC’s Infrared Precipitation with Station Data (CHIRPS), as well as tools like the Early Warning Explorer and ClimateSERV and modeled data from the Famine Early Warning Systems Network Land Data Assimilation System. Exposure to these workshop training materials resulted in use cases across countries and disciplines, with users applying these resources to their own decision-making needs.

A workshop participant noted that they had “been able to apply the skills gained. In January, together with colleagues, we were able to generate scenarios of seasonal rainfall for Malawi for the remainder of the season.”

These trainings then culminated in the enhanced capacity of over twenty agrometeorological technical professionals to better use EO data to address water resources and agricultural decisions in Africa. In addition, several participants discussed the future training of colleagues, which exemplifies the “empower-the-trainers” approach, as well as the increased impact and reach of these workshops.

One participant stated, “I am using some of the data sets introduced [during the training] in my day-to-day climate analysis, including forecasting. In addition, I have shared the tools with my colleague, who works in the agro-climate section, so that they can use the information from these tools to advise farmers and the ministry.”

CHC Project Expectations and Predicted Outcomes:

Through the continued collaboration of the CHC, RCMRD, and SERVIR-eastern and southern Africa, we are able to provide critical guidance to those with the direct, immediate ability to provide much-needed climate information to decision makers, often at crucial and timely moments. By offering such trainings, data becomes relevant, readily applicable information that better informs food security and water resource-related decisions. As one trainee wrote, “The training has equipped participants with new skills to process hydro-climate information for decision making, it is useful, relevant, and important.”

Working directly with partners in Africa has highlighted the need and potential for EOs in eastern and southern Africa. Throughout these trainings, we have seen technical professionals from national and regional met and hydrologic agencies using data and techniques acquired through this project, and we understand that the information is being used to inform decisions about water resources and agriculture. Helping those professionals acquire these tools and seeing monitoring techniques improve as they help to better assess the conditions on the ground is both invigorating and instructive, as such growth is the direct result of this project. As the project nears its final stage, we aim to incorporate and translate the feedback from all previous trainings into instructions to ensure that the framework best targets the needs of its participants. With this objective in mind, we believe that trainees will be better equipped, both in skills and resources, to access and apply these EOs with confidence.  

The value of these trainings cannot be overstated, as they benefit not only decision makers and key stakeholders, but also the CHC itself. Constructive feedback allows the CHC to evolve its training methods and content, creating a cycle of betterment that will ultimately reach a vast audience of eager and engaged climate service providers and decision makers. As such, the capabilities and tools of the CHC will have a greater impact, equipping people in the region with the skills necessary to access critical data sets and monitoring techniques. Additionally, the CHC gains invaluable face-to-face time with key stakeholders at regional and national levels. While currently in Year-3 of the project, continued feedback means that through the remainder of the project, participating technical professionals will receive training that has been tested and then altered to meet the trainees’ primary needs. Such activities that connect technical professionals to the CHC’s tools and techniques speaks to the lasting legacy that the Center intends to create.

CHC members Greg Husak and Shraddhanand Shukla, along with FEWS NET’s Tamuka Magadzire and SERVIR-E&SA’s Denis Macharia, with project participants in Lusaka, Zambia.

The benefits of such trainings span the breadth of the Climate Hazards Center’s widespread network of affiliates and partners. Denis Macharia, Weather and Climate Lead of SERVIR E&SA and CHC’s primary collaborator at RCMRD, states that “Data and skills are the two most critical needs of a hydrometeorologist in order to successfully accomplish his or her duties. In our region, accessing data can be a challenge, and when it is possible, still requires expertise.  Our partnership with UCSB and hydromet services agencies in the E&SA regions addresses these challenges quite remarkably. In the last 5 years, the UCSB team has produced a consistent rainfall data set and recently, an analogous temperature data set. Both data sets support analyses that are proving useful in addressing developmental challenges in the E&SA countries. The partnership is also providing key skills that are necessary to realizing maximum potential in the use of the data sets, training technical users in different applied skills like climate seasonal scenarios development and hydrological modeling. SERVIR will continue to ensure that data and skills gained so far are sustained over the long term and that agencies in the region continue to use these to inform decisions made by various stakeholders to address developmental challenges experienced in different sectors.”

For further reading on this project, please view the following articles courtesy of SERVIR GLOBAL, RCMRD, and NASA Applied Sciences Program:


Mid-season assessment of maize growing conditions in Southern Africa (2018-2019) reveals reason for concern

Mid-season assessment of maize growing conditions in Southern Africa (2018-2019) reveals reason for concern

Authors: Will Turner, Laura Harrison, Greg Husak

Editor: Juliet Way-Henthorne


  • Season Precipitation Performance Probability shows a high likelihood of below-normal rainfall throughout Angola, western Mozambique, Namibia, Botswana, Zimbabwe, and central South Africa with little chance of recovery.
  • Northeastern South Africa’s maize triangle does show a likelihood of normal season precipitation totals due to DJF recovery.
  • Current vegetation conditions (based on NDVI) are quite poor in the aforementioned countries.
  • End-of-season vegetation conditions, based on WRSI, are projected to be below average in the same countries.


This blog communicates season-to-date rainfall along with modeled and observed crop impacts for the 2018-2019 maize growing season in southern Africa. It also employs some of the new operational monitoring products available from the Climate Hazards Center.  

The southern Africa agricultural season broadly extends through the October-April season. While all regions are more dependent on subsets of that interval, the main producer of the region, South Africa, is shown to be strongly responsive to the December-February (DJF) rainfall (Shukla, in review). To capture these regional conditions and, specifically, conditions for South Africa, we begin with estimated probabilities of rainfall being below normal, normal, and above normal for each of the two monitoring windows (October to April and DJF).

The Season Precipitation Performance Probability (SPPP) quantitatively evaluates the probability of a current season’s total precipitation to finish in a given tercile, corresponding to below-normal (< 33rd percentile), normal (33rd–66th percentile), and above-normal (> 66th percentile) conditions, with respect to the historical record. Below normal to-date rainfall has significantly reduced the probability of normal total season (October to February) rainfall for much of the Southern Africa (Figure 1, above). Figure 2 (below) includes the 2-pentad CHIRPS-GEFS forecast in order to provide additional certainty of season outcomes.

The DJF monitoring window (Figures 3 & 4) is largely in agreement with the full-season window for our areas of concern. In both the October to April and the DJF monitoring windows, we see a significant uptick in the likelihood of given tercile outcomes (darker hues). This supports Shukla’s findings regarding the strong response of total seasonal rainfall to the DJF subset. Notably,  the two outlooks differ in the positive change in potential DJF precipitation performance in northeastern South Africa (northern maize triangle area). This region shows an increased likelihood for above-normal DJF rainfall (recovery of rains; potential for groundwater recharge).

The DJF subset shows a similar ability to capture the impact of to-date rainfall on likely season total outcomes. Figure 3 (above) is based only on to-date rainfall. Figure 4 (below) includes the 2-pentad CHIRPS-GEFS forecast.

NDVI products, which capture the greenness of the vegetation, are useful for indicating how this to-date rainfall has impacted vegetation so far (Figure 5). Percent of average NDVI during mid-to-late January (Source: NDVI-MODIS GLAM) is more than 30% below normal for approximately 20% of South Africa’s major cropping area for this season. Some areas of northeastern South Africa and Eswatini show closer to normal or above normal NDVI; these correspond to areas that the CHIRPS monitoring products (above) showed recent recovery in rains.

Figure 5: Percent of normal NDVI during mid to late January (Source: NDVI-MODIS GLAM). Shades of green indicate areas of above normal vegetation greenness, while shades of brown indicate areas of below normal.

While useful for categorizing projected seasonal rainfall, the SPPP product’s unidimensional nature does limit its ability to wholly provide context to current/projected agricultural performance. Specifically, plant-water availability is not strictly defined by total precipitation; the other half of the equation, which reduces this availability, is evapotranspiration. With this in mind, it is prudent for us to use additional monitoring products that take this evaporative demand into account.

For this, the Famine Early Warning Systems Network (FEWS NET) commonly uses the Water Requirement Satisfaction Index (WRSI) to monitor growing seasons throughout Africa and Central America. The WRSI uses gridded precipitation (PPT) and reference evapotranspiration (RefET) inputs, along with crop phenological parameters, to calculate the ratio of plant-available water to crop-specific water demand at each stage of crop development. In so doing, the index puts specific interest on the timing of the rainfall (as opposed to just the total). Below, we show the results of WRSI model simulations that provide more details about current maize impacts.

Overall, lackluster rainfall early in the 2018-2019 season likely delayed the start of the crop growing season (Figure 7; shown in shades of red). Parts of central South Africa still have not received sufficient rainfall for germination (Figure 7; shades of brown). This late onset of rains could also be influencing the negative NDVI anomalies seen before, as vegetation may not be fully mature this year, while in previous years it is nearly mature by this time.

Figure 6: The dekad identified as the Start of Season (SOS) for a location (pixel) is defined as the first dekad with at least 25 mm of rainfall followed by a total of at least 20 mm in the next two consecutive dekads. Areas colored gray have yet to receive sufficient rainfall for a season start.

Figure 7: 2018-19 start of season anomaly with respect to the 1981-2017 mode SOS. Areas colored in shades of blue indicate regions where the current season started earlier than the historical mode, while areas colored in shades of red started late. 
The lightest beige color indicates areas that have not yet started but are not beyond the mode start date. Darker hues of brown indicate areas that are past the mode SOS for that location and still have yet to receive sufficient rainfall for a season start.

In similar fashion to the SPPP calculation, we can use precipitation and RefET historical records to run out the remainder of the season and provide a collection of scenarios for the end of season WRSI. The median of these scenario outcomes (WRSI Outlook) is then compared to the median historical WRSI. We find that while starts did eventually occur for most of the region, the delay was followed by continued below average rainfall, and thus the outlook WRSI anomaly (Figure 8) and percent of normal (Figure 9) are well-below average throughout the aforementioned dry areas in the SPPP plots.

Figure 8. 2018-19 simulated end of season WRSI anomaly with respect to the 1981-2017 median WRSI. Hues of red indicate below average WRSI, while shades of blue indicate above average. Gray indicates average WRSI.

Figure 9: 2018-19 WRSI as a percent of normal, with respect to the 1981-2017 median WRSI. Green hues indicate above average WRSI, whereas orange/purple hues indicate below average. Pink denotes areas that have yet to receive sufficient rainfall for a start, and are beyond the mode SOS.

Additionally, the poor rainfall outlook for the 3rd dekad of January results in significant worsening of the WRSI Outlook (Figures 10 & 11).

2018-19 simulated WRSI anomaly (Figure 10, above) and percent of normal (Figure 11, below) including CHIRPS-GEFS forecast data for the 3rd dekad of January.

In agreement with our SPPP product, we see similar areas of concern in Angola, western Mozambique, Namibia, Botswana, Zimbabwe, and South Africa. As with the DJF monitoring window, there remains a potential for recovery to average crop conditions in northeastern South Africa.

As previously stated, most of these areas of concern are predominantly driven by DJF rainfall, so confidence in these observations should continue to increase over the next few weeks. More to this point, as season progress continues (Figure 12) and crop phenological cycle advances, crop-water demand will increase, and rainfall performance will have an increasing impact on the overall crop-water satisfaction. This is evident in the dramatic increase in severity of the negative WRSI anomaly and percent of normal when including CHIRPS-GEFS forecast for the third dekad of January.

Figure 12: The 2018-19 Season Progress combines the current SOS and the spatially varying LGP map (from GeoWRSI) to calculate how far along the season is at the current time as a percentage of the LGP.

This collection of products allows for in-depth monitoring of (1) the current state of rainfall-dependent growing seasons and (2) the impact to-date conditions could have on end-of-season agricultural productivity. While not as accurate as post-harvest assessments, these precipitation estimates, NDVI, and crop-water models help to provide sources of information for early warning and timely action.

Product Directories:



A Concerning 2018-2019 Southern Africa Cropping Season

Laura Harrison, Chris Funk, Will Turner, and Juliet Way-Henthorne

Special thanks to Climate Hazards Center and NASA FAME members for their contributions

Main points:

  1. Below normal rainfall so far in many rainfed cropping areas
  2. Pessimistic forecast for early 2019 rainfall in some areas of concern based on multi-model forecast and expectation for El Niño
  3. Forecast hotter than normal temperatures poses additional hazard for crops

Poor growing conditions to start the 2018-2019 season

Thus far, the 2018-2019 cropping season has been lackluster in terms of rainfall performance across much of Southern Africa (Figure 1a). Rainfall accumulations from mid-October to late December are below the 1981-2017 average by 50 mm to 200 mm in areas south of northern Zambia and northern Mozambique. In southeast Angola, southern Zambia, and parts of Zimbabwe, Botswana, South Africa, and Madagascar, these anomalies are 1.5 to 2 standard deviations below average- and thus are very dry when compared to the same period for past 1981-2017 seasons. Based on December CHIRPS preliminary rainfall data and CHIRPS-GEFS short term forecasts, CHC Early Estimates indicate that more recent rains may be closer to expected climatology for the majority of this region. Timely improvement in South Africa would be a much-needed positive change, as the country is a major contributor to regional maize production. The December data indicate a swath of continued below normal rainfall stretching across southern Angola, much of Zimbabwe, and Madagascar.

Figure 1. Rainfall performance thus far in 2018-2019 season. (a) Climate Hazards Center Early Estimates show rainfall anomaly and standardized precipitation index (SPI) for recent ~2 month and 20-day accumulation periods. The last 10 days of each period is based on an unbiased GEFS forecast (CHIRPS-GEFS). (b) Probability of October to April 2018-2019 rainfall total being below normal, normal, or above normal, based on to-date accumulations and if historical mid-December to April rainfall were to complete this season.

Rainfall during the next two to four months will be an important determinant of production outcomes. Rainfall forecasts for this period are concerning (see next section) given expected El Niño conditions and the low rainfall seen thus far across the region. Time will tell the actual outcome. Impacts thus far include delayed planting, crop damage from heatwaves, and poor veld and livestock conditions, according to the Southern Africa Development Community (SADC)’s December 2018 Agromet Update.

What if rainfall for the rest of the season is the same as in previous years? This scenario can be helpful to gauge the importance of season-to-date rainfall for the seasonal total. Also, it incorporates to some extent historical tendencies for persistence or transience in sub seasonal rainfall, if such patterns exist. Based on current conditions and climatological outcomes, there appears to be a high probability (50-90%) that many critical maize growing areas in South Africa, and also in the Caprivi Strip and parts of Angola and Zambia, will experience low seasonal rainfall totals (Figure 1b). Seasonal rainfall forecasts are pessimistic for some for these areas, as discussed below.

Below normal rainfall in early 2019 is the most likely scenario

For a large part of southern Africa that has already seen rainfall deficits, climate model forecasts indicate a 40-50% chance of rainfall during January to March falling below normal. Below normal is defined here as falling into the driest 1/3 of years. This outlook is coming from a multi-model ensemble forecast from the North American Multi-Model Ensemble (NMME). Areas with forecast below-normal rainfall include South Africa, Zimbabwe, Botswana, southern Zambia, central and southern Mozambique, south and eastern Angola, and Namibia. (Figure 2a).

Figure 2. Current forecasts and El Niño assumptions. (a) NMME probabilistic rainfall forecast for January to March 2019 based on December 2018 initial conditions. (b) Root zone soil moisture forecasts from the NASA Forecasting for Africa and the Middle East (FAME) project. Maps accessed 19 Dec 2018. More information can be found at (c) Average October to March CHIRPS rainfall anomaly for previous El Niños of similar expected moderate strength. Selections based on Niño3.4 SST.

Most models agree with respect to below-normal rainfall as the most likely category that these areas will fall into. If one considers the rainfall forecast from one of these NMME models (for December onwards) and the soil moisture conditions that may have been present in November, elevated concerns for crop impacts appear justified. NASA FAME forecasts, based on the GEOS5 model, show root zone soil moisture (in the top 1 meter of soil) in low-percentile categories for January and February (Figure 2b) and through April.

These pessimistic forecasts and observed deficits align with the expectation for El Niño and outcomes of past El Niño seasons. Figure 2c shows the average rainfall anomaly for October to March across five previous moderate strength El Niños (1994-95, 2002-03, 2004-05, 2006-07, and 2014-15). According to the CPC/IRI December forecast, for January to March, there is a 90% chance of El Niño through the 2018-19 austral summer. You can check the current outlook here.

Another hot, dry season?

Heat stress and high evaporative demand due to above-normal temperatures also appear to be a hazard for this season. The current NMME forecast for January to March 2019 near-surface air temperatures is reminiscent of what the models forecast for the 2015-2016 season (Figure 3). In early 2016, hotter than normal days exacerbated longer-term effects of rainfall deficits and were a factor in reducing South Africa maize yields by 40% (Archer et al., 2017). Similar to 2015-2016, the 2018-2019 NMME forecast for January to March (December initial conditions) show +0.5°C to +1°C anomalies for most of southern Africa and a prominent epicenter with +1°C  to +2 °C anomalies. The forecast epicenter is smaller in 2018-2019 than in 2015-2016, which is consistent with the difference in expected 2018-2019 El Niño strength compared to the very strong 2015-2016 El Niño.

Figure 3. NMME 2-m air temperature anomalies forecast for upcoming January to March season based on (a) December 2015 (b) December 2018 initial conditions.

AGRO-CLIMATE ALERT: Very Poor Somali Deyr Harvest Highly Likely


Chris Funk, Gideon Galu, Greg Husak, Will Turner, Juliet Way-Henthorne


An analysis of mid-season rainfall, WRSI simulations, and crop statistics indicate a high probability of a very poor Somali Deyr 2018 growing season. Rainfall deficits in Kenya also indicate poor growing conditions.

FEWS NET food security outlooks rely on a series of information products. Before the growing season, an analysis of climate modes and climate forecasts provides an evidence-based foundation for seasonal rainfall outlooks. By mid-season, however, rainfall observations, crop and hydrology models, and other forms of monitoring provide more detailed and accurate sources of information. These information sources can be especially powerful in areas with short growing seasons (such as Somalia), where arid conditions create a short window for successful harvests. FEWS NET science partners have been steadily improving our monitoring resources, and here, we use some of these products to assess likely agroclimatic outcomes for the Somali 2018 Deyr growing season. Crop statistics, provided by the FEWS NET data warehouse, allow us to further leverage these resources to provide quantitative projections of Maize/Sorghum crop production values. While not as accurate as post-harvest assessments, these preliminary rainfall-based projections suggest that a poor cropping season is very likely. Poor pasture conditions also seem probable for some regions.

Poor seasonal rainfall performance

Dry conditions appear extremely likely over southern Somalia and most of Kenya. Note the large discrepancy between the ARC2 (Fig. 1-left) and CHIRPS (Fig 1-right) results over Eastern Ethiopia. CHIRPS incorporates additional station data provided by the Ethiopian Meteorological Agency and is much more representative of the current situation in eastern Somalia.  Rainfall estimates from CPC’s ARC2 archive and the USGS/CHC’s CHIRPS2 data sets converge on similar outlooks for a very dry October-December growing season for Somalia (Figure 1). The CPC provides Seasonal Rainfall Performance Probability (SPP) Analyses for October-December. These analyses use observed ARC2 data up to the present and then completes the remainder of the season with all previous historical ARC2 observations. This enables them to estimate the probable seasonal outcome (Figure 1, left).  Based on the SPP analysis through mid-November, most of Somalia (including the main crop growing areas) are very likely (>75% chance) to receive below-normal rainfall. The USGS/CHC has also recently started producing an “Early Estimate” product that combines observed CHIRPS data with CHIRPS-compatible 10-day GEFS rainfall forecasts. The right panel of Figure 1 shows Early Estimate rainfall anomalies for October 1 through November 25th. Dry conditions appear extremely likely over southern Somalia and most of Kenya.

Figure 1. Left – CPC Seasonal Precipitation Performance predictions for October-December 2018. Right – CHC Early Estimate seasonal rainfall anomalies.

Somalia Deyr Rainfall and Crop Production Projections

Next, we examine in detail rainfall averaged across crop-growing zones in Bay, lower, and middle Shabelle districts (Figure 2). These rainfall estimates combine October and the 1st two dekads of November, with CHIRPS-GEFS forecasts used to fill in dekad 2 of November. For 2018, these seasonal totals look very low, almost identical with previous signature drought years: 1998, 1999, 2003, 2005, 2010, and 2016.

Figure 2. October 1-November 20 CHIRPS observations for the main Somali crop growing regions.

Using historical sorghum and sorghum+maize crop production data from the FEWS NET data warehouse, we can use logistic regression to relate these seasonal rainfall totals to 1995-2016 Deyr seasonal production estimates (Figure 3). We treat sorghum separately because we found that it had a very strong relationship with rainfall (R2~0.7). We also provide maize+sorghum total production projections to support food security analyses. It should be noted that these projections are not intended to replace careful post-harvest crop assessments, but, rather, as an advanced outlook on the general pattern such assessments are likely to indicate. We have excluded 1997 and 2011 from our estimation procedure, because 1997 was an exceptionally wet year with modest production (presumably due to flooding), while 2011 was a normal rainfall year with low crop production (presumably due to conflict).

One of the most notable features of Figure 3 is the non-linear relationship between rainfall and crop production. Below 100 mm of rainfall, crop production drops very quickly. A 2018 rainfall total of ~60 mm indicates poor crop production. Using take-one-away cross-validation and regression, we have estimated the 2018 Deyr Sorghum and Sorghum+Maize production totals as 45 and 81 thousand metrics tons. These estimates correspond to ~50% and 60% of the 2012-2016 average production (i.e. production about half of the recent average production). Eighty percent confidence intervals, based on the cross-validation results, indicate ranges of 43-46 and 77 to 83 thousand metric tons for Deyr Sorghum and Sorghum+Maize production.  Note that the late November rainfall estimates are based on weather model forecasts, and there is still room for the season to improve; however, the overall outlook is very pessimistic.

Figure 3. Scatterplots of Rainfall and Somali Deyr Crop Production

Per Capita Sorghum+Maize Production

According to United Nation estimates, Somalia’s population has doubled since 1995, while Deyr crop production totals have not increased. Figure 4 shows per capita Deyr Maize+Sorghum production estimates (i.e. crop production per person). In the last 10-12 years, approximately half of the Deyr seasons were associated with very low per capita crop production outcomes. 2018 appears likely to be another such season, with per capita maize+sorghum production of around 5.4 kg per person.

Figure 4. Deyr Per Capita Sorghum+Maize Production

Concerns About Pasture Conditions

Next, we turn to an assessment of pastoral conditions, focusing on the same three regions (Bay, Lower Shabelle and Middle Shabelle). This analysis is not meant to be an exhaustive assessment of rangeland conditions for the entire country. Furthermore, rangeland conditions may be more variable than crop outcomes, since grasses can respond quickly to late-season rains. However, through the first dekad of November (Figure 5) vegetation conditions (as represented by the USGS eMODIS NDVI) appear quite low. Vegetation greenness in these regions typically peaks in the last dekad of November.

Vegetation conditions exhibit persistence, and robust lagged relationships with prior rainfall totals. Hence, the end-of-November values can be predicted very accurately based on 1st dekad of November values and October 1-November 20 rainfall estimates (R2 values of ~0.8). Times series of the observed and predicted regional NDVI are shown in Figure 6. According to these estimates, peak NDVI in these regions appears substantially below normal and similar to previous severe drought years. Note also the sequence of repeated low Deyr NDVI values. The 2016, 2017, and 2018 values are all quite low – indicating repetitive shocks. The 2017 season March-May Gu season was poor, while the 2018 Gu season was normal-to-above normal. So, four out of the last five seasons seem to have been associated with poor growing conditions: Deyr 2016, Gu 2017, Deyr 2017, and Deyr 2018.

Figure 5. Regional eMODIS NDVI time series. The red lines depict the median values for the 2003-2017 period. The blue line shows 2018 conditions.


Figure 6. Time series of November dekad 3 eMODIS NDVI, along with predictions based on dekad 1 NDVI and October-November rainfall

Conclusions and Context

The most effective drought early warnings systems rely on multiple indicators, transitioning from climate forecasts and analogs before the onset of a rainy season, to mid-season projections of likely food system disruptions. Here, we have used FEWS NET monitoring tools to focus on mid-season projections for Somalia. While these projections should not be mistaken for assessments, our analysis has shown that for Somalia, a high level of predictability can be obtained in mid-November. This skill rests on the short duration of the Deyr growing season, the general aridity of the region, and the persistent nature of vegetation, soil moisture, and crop conditions. Unfortunately, the results presented also suggest another poor crop-growing season for Somalia. For the Bay and Shabelle regions analyzed, poor vegetation conditions also appear likely – conditions which could quite likely persist until the March-May Gu rains of 2019. While not examined in great detail here, precipitation totals and WRSI (Figure 1 and 2) also appear quite poor over Kenya, and poor Kenyan short rainy season harvest could be problematic, ultimately contributing to regional food and price stresses.  Please note that it is probable that the marginal agricultural areas of the southeastern lowlands are likely to be more adversely affected by the current delayed onset, the quality of the seasonal rains, and the substantially reduced growing period.

The results concerning the Deyr harvest expressed here should be placed in the context of a good 2018 Gu harvest. Figure 7 below, taken from the 2018 post-Gu FSNAU-FEWS NET joint assessment (here), shows that the 2018 Sorghum+Millet Gu harvest was estimated to be about 147 thousand metric tons, the best harvest since 2010. On average, the Gu and Deyr harvests similar in magnitude, about 90 thousand metric tons, so the good Gu harvest will certainly help partially offset the poor 2018 cropping season. Current maize and sorghum prices are much lower than their peak in 2017, but historically prices also typically increase between December and April-May. It should also be noted, however, that the 2016 and 2017 Gu harvests were below normal, as were the 2016 and 2017 Deyr season harvests.

Key messages:

  • Rainfall-based early assessments of the 2018 Deyr harvest indicate substantial (40-50%) deficits with little chance of recovery.
  • Vegetation conditions in some areas are also quite poor, and predicted to remain so unless unseasonal rains arrive.
  • While the 2018 Gu season was very good, Somalia also faces the stress of repetitive shocks in four of the past five rainy seasons, compounded by increasing population stress.


Figure 7. Gu Season Cereal Production – figure taken from the 2018 post-Gu FSNAU-FEWS NET joint assessment

Concern for the “short rains” 2018 season in eastern Horn of Africa

Contributors: Laura Harrison, Chris Funk, Martin Landsfeld, Will Turner, Greg Husak, Juliet Way-Henthorne

As East Africa’s October to December “short” rainy season approaches its midpoint in many areas, parts of Kenya, southeastern Ethiopia, and much of southern Somalia have seen substantial rainfall deficits. Factors include a late start to the onset of seasonal rains, fewer than normal rain days, and long dry spells. Based on the Climate Hazards Center Early Estimate, a monitoring data resource that provides early indications of sub-seasonal to seasonal rainfall performance, central Kenya and Somalia’s cropping zones may end up with 50-100mm deficits for the October 1st to November 10th, 2018 period (Figure 1).

Figure 1: Climate Hazards Center Early Estimate for the 2018 short rains status made on November 5th, 2018. The CHC Early Estimate approach combines CHIRPS final and preliminary rainfall estimates (30-day or 60-day) with a compatible, unbiased version of the 10-day GEFS ensemble mean forecast (see Figure 3). Figure 1 is a rainfall anomaly composite of preliminary CHIRPS October 2018 rainfall and CHIRPS-GEFS, released November 1st for Nov. 1st-10th. Note that October data is preliminary and subject to change in the final version of CHIRPS, which will be available mid-November. In CHIRPS final, Somalia data includes FAO SWALIM reports and Ethiopia data includes Ethiopia National Meteorological Agency reports.

While some areas may receive relief in November, poor crop outcomes should be a concern in short growing season areas, such as Somalia. Concern for the performance of the short rains in the eastern Horn and Deyr crop production, in particular, is based on convergent evidence from independent data sets and a pessimistic outlook from the current rainfall forecast.

  1. Convergent evidenceUSGS’s expedited MODIS NDVI anomalies show vegetation productivity in a degraded state at the end of October. Like CHIRPS, NOAA CPC’s ARC2 data shows expansive rainfall deficits across the eastern Horn and similar patterns with respect to rainfall anomalies (Figure 2). These came during the typically wettest month of the short cropping season in Somalia’s main cropping zones.

Figure 2. Left: USGS/EROS eMODIS NDVI anomaly for last 10 days on October 2018. Figure from USGS FEWS NET Data Portal Right: NOAA CPC ARC2 October to December 2018 rainfall anomaly.  Figure from CPC Africa Desk.

2. Rainfall outlooks

Based on GEFS forecasts, relief is not expected during the first two weeks of November for most of Kenya and for at least two of Somalia’s main cropping zones (Figure 3). National Multi-Model Ensemble (NMME) forecasts for November rainfall vary widely. Outlooks from NOAA CPC, based on historical November and December rainfall, indicate a higher than 60% chance of below normal OND totals in southern Somalia and southeastern Ethiopia.

Figure 3. CHIRPS-GEFS rainfall anomaly forecast for the 10-day period beginning November 4th, 2018. More can be found about this data product and how to access it a blog on the UCSB Climate Hazards Center website

Blending CHIRPS Data and GEFS Forecasts for an Enhanced Rainfall Forecast Product

Contributors: Martin Landsfeld, Laura Harrison, Chris Funk, Juliet Way-Henthorne

Background – CHIRPS    

The Climate Hazards Group Infrared Precipitation with Stations data set (CHIRPS) is a land-based, quasi-global (latitude 50°N-50°S), 0.05 degree resolution precipitation data set. It has a relatively long-term period of record (1981 – near present). CHIRPS is based on a well-developed climatology and perturbed with infrared satellite measurements and in-situ observations to estimate gridded precipitation in near-real time. A final monthly precipitation product is created about 2 weeks after the end of the month when all the station data (from over 10 different sources) has been collected and blended with the satellite estimates. A preliminary product, CHIRPS Prelim, is available on the dekadal time scale and available 2-3 days after each dekad and month. It is blended with only 2 sources of station data— WMO GTS and CONAGUA.

You can find more information about CHIRPS at:

Figure 1.  CHIRPS rainfall estimates product.


Background – GEFS

NCEP’s Global Ensemble Forecast System (GEFS) is a weather forecast system that provides daily forecasts out to 16 days at 1 X 1 degree resolution at 6-hour intervals.

Figure 2. GEFS rainfall estimates product.

This forecast product can be very useful to early warning famine and hydrological monitoring efforts, so the Climate Hazards Group (CHG) creates forecast precipitation fields at the dekadal (10 day) time scale and makes those available to researchers and decision makers in the EWX data viewer at:

The data is also available for download in GeoTIFF format through our FTP service at:

Precipitation is accumulated to 10-day intervals, every day. The precip_mean, anom_mean, and zscore_mean directories contain the means of the ensemble runs, and the file names are comprised of the dates of the first day of the forecast, followed by the last day of the forecast.


Blending GEFS Forecasts with CHIRPS

To make GEFS forecasts interoperable with CHIRPS, we bias-correct and downscale GEFS with respect to CHIRPS. The method that is followed is similar to the standard cumulative density function matching process of bias-correction. For a given target 10 day period’s GEFS forecast, its rank with respect to its climatology (1985 – present) is identified, and then a value of the same rank from the CHIRPS climatology is used to replace GEFS forecasts for each ~5km pixel. A comparison is shown below.

Figure 3. Comparison of GEFS 1 degree resolution forecast (left) with the blended CHIRPS-GEFS 0.05 degree forecast (right). 


Accuracy assessment:

GEFS and CHIRPS-GEFS 10-day accumulations were compared to historical Ethiopian station measurements. Ethiopia maintains many stations through the time period since 1985 and has a varied topographic relief. Significant improvements were made, as seen below (Figure 4).


The correlation, mean bias, and root mean squared error are shown in Table 1. Significant improvements in the correlation to station values were improved, and the mean bias and RMSE were reduced, demonstrating the increased accuracy when the GEFS forecasts are blended with the CHIRPS climatology and long-term time series.

                             Corr. Bias RMSE

               GEFS: 0.51 1.45     63.54

CHIRPS-GEFS:   0.68 1.05 37.26

Table 1. Comparison coefficients of GEFS and CHIRPS-GEFS estimates with rain gauge stations.

Spatial correlations where calculated between CHIRPS-GEFS forecasts and CHIRPS Final estimates for dekadal time periods over the entire time series. High correlations exist over significant portions of Africa for the first dekad of April (Figure 5). Other correlations, biases, and mean-biased errors for other time periods can be found at: 

Figure 5. Correlation between CHIRPS-GEFS and CHIRPS Final for the 1st dekad of April (1985-2016). 



Real-time precipitation forecasts are critical for predicting flooding events and protecting property and lives. When torrential rains hit Kenya in mid to late April 2018, major flooding occurred, overwhelming drainage systems and collapsing dams. The Kenya Meteorological Department wrote:

Heavy rain has been affecting the central, the south-west and south-east areas of the country, including the capital Nairobi, since the beginning of the month, causing floods, flash floods, and casualties. According to media, as of 20 March, the death toll has reached at least 15 people in the provinces of Central, Nyanza, and Eastern. They also reported that around 1000 people were evacuated in the counties of Makueni (Eastern province), Kilifi and Tana (Coast province). Over the next 24 hours, more heavy rain with local thunderstorms is forecast for the affected areas. 

The impacts of this flooding can be seen below (Figures 6 and 7).

Figure 6. A passenger is rescued from his submerging vehicle following heavy downpour in Nairobi on April 15, 2018.

Figure 7. Submerged vehicle in Nairobi, April 24th, 2018. 


The CHIRPS-GEFS 10-day forecast, made on April 10th, 2018, could have been used to help predict these events and warn residents of impending flooding. As an example, consider the left image in Figure 8 below. This map, obtained from the CHG experimental Early Warning Explorer ( using the CHIRPS GEFS latest under Global Datasets, shows a downscaled GEFS forecast for the 10th – 20th of April. Below is a comparison of  the anomalies of the CHIRPS-GEFS forecast with the subsequent CHIRPS Final for the same time period. The right image shows the observed CHIRPS data for that same time period. While the agreement is not perfect, the CHIRPS GEF forecasts did a very good job of capturing the potential high rainfall amounts. Figure 9 shows forecasts and observations for the next 10 days April 21-30.

As these maps illustrate, the CHIRPS GEFS forecasts can be very useful tools for flood prediction. Current efforts at the CHG are setting up daily updates of these predictions. These results will be available via ftp at,, and within the EWX.

CHIRPS-GEFS Forecast, 2018/4, dekad 1       CHIRPS Estimates, 2018/4, dekad 1

Figure 8. CHIRPS-GEFS forecast for dekad 2 of April, 2018 (left) and the CHIRPS Final estimates (right) for the same time period. 


The period of the 21st -30th of April, 2018 shows a similarly good comparison between the prediction and the CHIRPS estimate.


Figure 9. CHIRPS-GEFS forecast for dekad 3 of April, 2018 (left) and the CHIRPS Final estimates (right) for the same time period. 

Update on 2018 Blue Nile forecast

Update to the guest blog by the Ad hoc Blue Nile Forecast Group (listed alphabetically): Sarah Alexander (1), Paul Block (1), Annalise Blum (2), Shraddhanand Shukla (3), Shu Wu (1), Temesgen Yimane (2), Ben Zaitchik (2)*, and Ying Zhang (2).

  1. University of Wisconsin-Madison, Madison, WI, USA

2. Johns Hopkins University, Baltimore, MD USA

3. University of California Santa Barbara, Santa Barbara, CA, USA

*Correspondence can be addressed to

The end of July represents the midpoint of the Kiremt rainy season in the Blue Nile basin. On average, just over half of the June-September rainfall total is realized by the end of July. This is also true for the annual total rainfall, January-December (Figure 1; shown spatially in Figure S1). This makes the end of July an opportune time to provide an update on the seasonal forecasts of 2018 Blue Nile rainfall and river flow that we posted at the outset of the rainy season.

Figure 1: Cumulative Blue Nile basin rainfall, January-December, according to CHIRPSv2 1981-2017 climatology. On average, 54% of annual rainfall occurs by the end of July (month 7).
At the time of our first post, there were already reports that rains had come early to portions of the Blue Nile basin, and we found near unanimity across statistical and dynamically-based seasonal forecasts that the June-September Kiremt rainfall would be average to above average. Forecasts of Blue Nile flow showed the same tendency. Consistent with these forecasts, rainfall in the basin through the first dekad of July was well above average (Figure 2). Positive rainfall anomalies for early season rainfall exceeded 50 mm over the majority of the basin, with significant areas showing anomalies greater than 100 mm. For context: according to CHIRPS estimates, average June rainfall for the basin is 195 mm, and average June-September total rainfall is 942 mm. So anomalies on the order of 50-100 mm through only the first third of the rainy season are quite meaningful.
Figure 2: CHIRPS-prelim cumulative seasonal rainfall anomalies for June 1 – July 15, 2018. Information on the CHIRPS-prelim product is available at

Interestingly, however, the North American Multi-Model Ensemble (NMME) forecasts launched in July show a more mixed outlook than the same ensemble had offered in May. At the time of the May forecast, every NMME model had a mean prediction of average to above average rainfall for June-September (Figure 10 in our original post). Looking at forecast July-September rainfall in the July initialized NMME simulations, we see that there are now several models that predict below average rainfall, albeit with only a modest negative anomaly (Figure 3). This breaks the consensus that existed in May. This shift to drier forecasts might reflect the influence of a shifting El Niño outlook, as the predicted probability of an El Niño forming by the end of the season is higher now than it was in May (Figure 4; compare to Figure 9 in the original post).

Figure 3: Forecast of July-September rainfall anomaly for the Blue Nile basin (mm) in July initialized forecasts of NMME models. Mean is shown by (x), circles are individual members, and boxplots show quartiles. The “Average” boxplot consists of the average forecast of each participating model.
Figure 4: Oceanic Niño Index forecast, generated by NOAA CPC and the International Research Institute (IRI) and issued on July 23, 2018.

Notwithstanding this somewhat drier outlook from NMME, the strong rains observed in June and early July got the basin off to a wet start. This, combined with an NMME forecast that still, across all models, points to average rainfall conditions for the remainder of the season, suggests that the Kiremt rains will be average to above average overall. The probability of a dry year is low, though some NMME models now suggest that late season rains might be lower than average.

Average to above average Blue Nile River flow expected in 2018

A guest blog by the Ad hoc Blue Nile Forecast Group (listed alphabetically): Sarah Alexander (1), Paul Block (1), Annalise Blum (2), Shraddhanand Shukla (3), Shu Wu (1), Temesgen Yimane (2), Ben Zaitchik (2)*, and Ying Zhang (2).

  1. University of Wisconsin-Madison, Madison, WI, USA

2. Johns Hopkins University, Baltimore, MD USA

3. University of California Santa Barbara, Santa Barbara, CA, USA

*Correspondence can be addressed to


Ethiopia will soon begin filling the reservoir of the largest hydropower dam in Africa, the Grand Ethiopian Renaissance Dam (GERD). Construction of the GERD has been highly controversial because it sits on the Blue Nile River (Figure 1), which provides about 60% of total natural Nile flow into Egypt. Egypt relies almost exclusively on the Nile River for its renewable freshwater supply. When complete, the GERD will rise 155 meters tall, have an installed generating capacity of over 6,000 megawatts, and create a 74 cubic kilometer reservoir. This dam will be the first major infrastructure project in Ethiopia on the main stem of the Blue Nile.

Figure 1: The Nile Basin (yellow), including the GERD site (orange star) and GERD catchment (green)

The history, politics, and current diplomatic impasse regarding the construction and operation of the GERD have been covered extensively and sometimes breathlessly in media outlets around the world (for example: here, there and everywhere). We won’t get into those issues here, but suffice it to say that there is reason for concern. This is true both because of the actual hydrological impacts that filling of the GERD reservoir might have on downstream countries and because the dam has become a focal point for broader geopolitical tensions. In this context, we ask a simple forecast question: what is the status of seasonal forecasts of Blue Nile River flow? Few would disagree that filling the reservoir during wet years would lead to fewer acute impacts on downstream countries. While it is not clear whether Ethiopia will pursue a forecast-based filling policy (valuable as that might be!), we believe it would be useful for interested Parties to share a similar understanding of how the rainy season is likely to unfold in each year of the filling period if and when they engage in discussions of filling plans and responses.

With this as motivation, this post will: (1) offer a brief review of the drivers of rainfall variability in the Ethiopian Blue Nile basin; (2) describe existing statistical and dynamically-based forecast systems; and (3) offer an informal ensemble prediction for 2018 flows. The objective of this prediction is not to supplant existing forecast systems in the region. High quality meteorological seasonal forecasts are available from the Greater Horn of Africa Climate Outlook Forum (GHACOF) as well as from the Ethiopian National Meteorological Agency and other nations’ forecast centers. Rather, we aim to present a broad range of predictions generated using different methods, and generated in the absence of any perception of subjective input.


For those in a hurry: our multi-method seasonal forecast ensemble shows a strong likelihood of average to above-average Blue Nile basin rainfall and associated Blue Nile River flows at the GERD site in 2018 (Figure 2). The result holds true across statistical forecasts and dynamically-based forecasts (Figure 3). This is consistent with regional seasonal outlooks issued by several independent modeling centers. If the Ethiopian government does begin to fill the reservoir this year, then, these forecasts suggest that there is a low probability that reasonable filling activities would lead to catastrophically low flow in Nile River downstream. This seasonal outlook does carry substantial uncertainties, as we describe below.

Figure 2: Percentage of forecasts in this study predicting below normal, near normal, and above normal June-September rainfall (left) and June-December streamflow at the GERD site (right) for 2018, including eight NMME models (ensemble mean for each model) and eight statistical models. NMME forecasts are adjusted for mean biases. No variance adjustment was applied to any of the model forecasts.

Figure 3: (A) Boxplots of June-September historical rainfall, 1982-2017, according to CHIRPS rainfall estimates (grey), and of the 2018 forecasts for dynamically-based models participating in the North American Multimodel Ensemble (NMME; green) and for statistically-based models applied in this study (blue). Dashed lines show the upper and lower terciles of historical rainfall totals. (B) As in (A), but for the June-December forecast Blue Nile flow at the GERD site.

Drivers of rainfall variability in the Blue Nile basin

The Ethiopian Blue Nile (or Abay River, to Ethiopians) is located in the western portion of the country, in a region that gets the vast majority of its rainfall in the summertime kiremt season, between June and September. The flow of the Nile at the GERD site is highly seasonal, following the rains. Peak flows are most commonly seen in October (Figure 4A). Precipitation in the Blue Nile is variable from year to year (Figure 4B), and this rainfall variability is associated with strong interannual variability in Blue Nile flows (Figure 4C). The coefficient of variation for annual rainfall in the basin is on the order of 6%, while the coefficient of variation in annual flow near the GERD is close to 20%.

Figure 4: (A) climatological monthly average rainfall for the GERD catchment area (CenTrends data) and climatological monthly average Blue Nile River flow at El Diem, near the GERD site; (B) density function of annual total CenTrends rainfall for the GERD catchment; (C) as in (B), but for El Diem flows. For all plots rainfall data are for the period 1965-2009, constrained by our access to flow data


The effort to characterize, explain, and predict this variability dates back thousands of years, to the Nilometers of Ancient Egypt and the stories of the Bible. It has also been an active focus of modern climate research, and we won’t attempt a review here; see Berhane et al. (2013) or Nicholson (2017), among others, for more comprehensive treatment. Some prediction-relevant highlights from the literature on drivers of Blue Nile variability are:

  1. The identification of a tropical Indian Ocean influence on East Africa summer rains, in which low pressure in the Indian Ocean enhances the flow of moist near-surface westerly winds across the Congo Basin and into Eastern Ethiopia.
  2. Complementary work on the westerly wind influence that identifies potential predictors in surface or near-surface pressure anomalies in the tropical Atlantic and the Arabian Peninsula, or the gradient between the two.
  3. Evidence that the southern hemisphere subtropical high pressure centers—the Mascarene in the Indian Ocean and the St. Helena in the Atlantic Ocean—influence the inflow of water vapor to the region.
  4. Studies showing that the strength and location of the Tropical Easterly Jet (TEJ) and, to some extent, the African Easterly Jet (AEJ) modify convection in the region. Links between the TEJ and the Quasi Biennial Oscillation (QBO) have raised the possibility of QBO as a predictor of Blue Nile flow.
  5. Perhaps most prominently, extensive study of the influence that the El Niño Southern Oscillation (ENSO) has on summertime Ethiopian rainfall, both synchronously and with a time lag. The ENSO influence is preeminent in operational statistically-based prediction systems, and its impact is clear in almost any analysis of rainfall variability (e.g., Figure 5). During the summer rains, El Niño events are associated with below average rainfall; this is the opposite of the well-known influence ENSO has on the eastern Horn of Africa and Equatorial East Africa during the October-December short rains.  The mechanism of the summertime ENSO influence, however, is still debated, and could be multifaceted. Influences of ENSO on subtropical highs, on water vapor advection, and on the TEJ and AEJ have all been proposed.

Figure 5: annual rainfall anomaly (mm) associated with El Niño relative to non-El Niño years. Precipitation data are from CHIRPS for the period 1981-2014. El Niño years are defined in terms of Niño 3.4 anomaly.

This partial list of drivers of variability points to the potential for teleconnection-based prediction in the region. But it also highlights the complexity of interacting climate dynamics that affect the Blue Nile basin (Figure 6). These influences also appear to vary over the course of the season. Berhane et al. (2013) find that the influence of ENSO and of Indian Ocean predictors tends to be strongest late in the rainy season and non-significant early in the season. Further, the influences are non-stationary in time. The ENSO teleconnection, for example, which is central to many prediction systems, was stronger prior to 1976 than afterwards, and recent warming in the western Pacific and Indian Oceans has had a substantial impact on rainfall across East Africa.

Figure 6: schematic map of some of the leading proposed influences on the Ethiopian summer rains. ENSO = El Niño Southern Oscillation, AEJ = African Easterly Jet, TAH = Tropical Atlantic High, SHH = St. Helena High, green arrow = Mediterranean Sea water vapor import, AL = Arabian Low, SAM = South Asian Monsoon, TEJ = Tropical Easterly Jet, IOSST = Indian Ocean SST, MH = Mascarene High.

Existing seasonal forecast models

The importance of rainfall in East Africa has inspired many efforts at seasonal prediction, including a number that are directly relevant to Blue Nile flow forecast. The majority of these efforts were undertaken for research and are not currently operational, while a smaller number are available as operational forecast systems. We divide these efforts into statistically-based forecast models, which are based on observed relationships between rainfall (or river flow) and time lagged large scale predictors in the climate system, and dynamically-based forecast systems, which employ global climate models to forecast atmospheric conditions and surface meteorology at lead times of weeks to months.

Most of the published statistical models of Ethiopian summertime rainfall–the season that matters most for Blue Nile flows–derive their skill all or in part from associations with ENSO and/or Indian Ocean surface conditions (Table 1). Gissila et al. (2004) leveraged these teleconnections to generate a nine predictor linear regression model in which summertime Ethiopian rainfall is positively associated with western Indian Ocean SST and negatively associated with SST in the tropical Pacific Ocean (Niño 3.4 region). This is consistent with empirical relationships and proposed mechanisms described above.  Korecha & Barnston (2007) take a slightly different approach, using the evolution of Niño 3.4 SST and tropical Atlantic Ocean SSTs in the months leading up to the rainy season as predictors, alongside the late spring Niño 3.4 SST. Their predictive regression model indicates that springtime cooling in the Niño 3.4 region is associated with more precipitation, perhaps because this metric captures a trend away from El Niño (or a trend into La Niña) conditions in the months leading up to the onset of rains.

Nicholson (2014) included atmospheric fields in her identification of predictors. This led to a three variable linear regression model in which summertime rains in East Africa are a function of zonal wind strength in the TEJ exit region, SST gradients in the tropical and subtropical Pacific Ocean, and tropical Indian Ocean SLP. Segele et al. (2015) cast a wide net for predictors, generating an ensemble of multivariate regression models from a family of potential atmospheric and SST predictor fields. The Ethiopian National Meteorological Agency use a statistical analog approach based on ENSO indices. Historical analog years are selected based on similarity in a family of ENSO indicators diagnosed both from observed and forecast ENSO conditions for the coming rainy season. These analog years are then applied to forecast seasonal total rainfall terciles for homogenous regions across Ethiopia and also to forecast second order properties of the rainy season such as onset and cessation of rains and probability of extreme events (Korecha & Sorteberg, 2013).

A number of global dynamically-based seasonal forecast systems are also available for application to Blue Nile forecast. The raw output of some of these dynamically-based systems are publicly available (notably all models participating in the North American Multi-Model Ensemble (NMME); Kirtman et al. 2014), and all centers generate products that can be incorporated into value-added analyses by operational forecasting organizations. Analysts with the Famine Early Warning System (FEWS), for example, generate outlooks that take into account dynamically-based predictions from NMME, the European Centre for Medium-range Weather Forecasts (ECMWF), and the United Kingdom Meteorological Service (Met UK), along with ancillary climate data and expert opinion. The Greater Horn of Africa Climate Outlook Forum (GHACOF) generates probabilistic consensus forecasts by merging multiple dynamically-based and statistical forecasts, based heavily on input from participating national meteorological agencies and the IGAD Climate Prediction and Application Center (ICPAC). The International Research Institute for Climate and Society (IRI) produces a widely used forecast based on bias corrected and weighted ensembles of dynamically-based forecast systems.

Here we examine as many of these statistically and dynamically-based forecasts as we had information and time to pursue, both to see how the models have performed in recent years and to provide an outlook for the 2018 rainy season. For statistical models that are not operational we adopted the predictors and model structure from the literature but refit all coefficients based on a consistent suite of datasets, updated to as recently as the data allowed. For operational systems that have issued their own 2018 predictions we simply report the publicly issued forecasts.

Models used in this study


The statistical models used in our analysis are listed in Table 1. In all cases we used June-September total precipitation from CHIRPS as the predictand. Models were fit using a combination of predictors from NCEP/NCAR reanalysis product and NMME lag-0 forecast products (forecasts for May, made in May). The NMME lag-0 fields were used for any May predictors required for 2018 June-September prediction, since latency in the reanalysis product makes it impossible to use May reanalysis fields for an operational prediction that is issued by the beginning of the rainy season. Leave-one-out cross-validation was used when fitting these models to optimize for out-of-sample predictive accuracy.  The period of retrospective analysis was 1982-2017, the longest record for which we could obtain output from both statistical models and dynamically-based prediction systems. Recognizing that teleconnections might have shifted over this 36 year period, we also fit one model using the shorter 2001-2017 record in order to see how changes post-2000 might influence model skill (Blum-2 in Table 1). Our analysis of dynamically-based forecast systems is limited to models in the NMME ensemble, as those were the only models for which we could easily obtain both historical and operational forecasts. The years 2011 and 2017 were not available for all NMME models included in the analysis, so the evaluation period was restricted to 1982-2010 plus 2012-2016. Precipitation from NMME models was bias corrected to CHIRPS prior to evaluation.

Table 1: Statistical models used in this study. P_JJAS=Sum of precipitation for June, July, August and September; SST=Sea Surface Temperature; All variables (predictors and predictand) were standardized (subtract mean and divide by standard deviation 1982-2017). Note that in applying previously published models we retain their structure and list of candidate predictors, but coefficients and, in some cases, selected variables are different from the originally published versions.

Model/Paper Model Structure Predictor variables
Gissila et al. (2004) Multivariate linear regression March, April, May SST for tropical western Indian Ocean, 10ºS-10ºN, 50-70ºE, tropical eastern Indian Ocean, 10ºS-0,90º-110ºE and Nino3.4
5ºS-5ºN, 170ºW-120ºW
Korecha & Barnston (2007) Multivariate linear regression The difference of May minus February–March SSTs over the south Atlantic, 30°S–40°S, 30°W–15°W
The difference of May minus the February–March Niño-3.4 SST, 5°N–5°S, 170°W–120°W
May Niño-3.4 SST, 5°N–5°S, 170°–120°W
Nicholson (2014) Multivariate linear regression May u 200 hPa, 30ºE-50ºE and 0º -10ºN
May difference in SST:

(170º -265ºE, 5ºS – 5ºN) – (137ºE-160ºE, 18ºN – 28ºN)

May sea level pressure (SLP) for 80º -90ºE,  5º 15ºN
Segele et al. (2015) Mean of 11

Multivariate linear regression models (Fig.5 in Segele et al. 2015)

20 predictors from Table 1 in Segele et al .(2015)
Wu – this study Linear Inverse Model May, June, July, Aug and Sep EOFs of Local Precipitation, Tropical Pacific SST, Tropical Atlantic SST, Indian Ocean SST, North Pacific SST, North Atlantic SST,

Southern Ocean SST, Global 850 hPa and 200 hPa Geopotential height

Alexander – this study Principal Component Regression   First two principal components of May Tropical Pacific Ocean SST, March-Feb Central Indian Ocean SST, April Tropical Atlantic SST, and Indian Monsoon SLP Index
Blum-1 – this study Multivariate linear regression May SST for tropical western Indian Ocean, 10ºS-10ºN, 50ºE-70ºE

May sea level pressure (SLP) for 80ºE – 90ºE, and 5ºN – 15ºN

May difference in SST: (170ºE – 265ºE, 5ºS – 5ºN) – (137ºE-  160ºE, 18ºN – 28ºN)

Southern Europe air temperature at 100hPa, 10ºE-15ºE, 45ºN-50ºN

North Arabian Sea March meridional temperature flux at 925 hPa, 60ºE-65ºE, 20ºN-25ºN

Blum-2 – this study Multivariate linear regression (fit using 2001-2017 data) Eastern Indian Ocean zonal geopotential height flux at 300 hPa, 80º-85ºE, 5ºS-0º

Arabian Sea meridional moisture flux at 925 hPa, 60º-65ºE,10º-15ºN

North tropical Pacific SST, 171ºE-175ºE, 16ºN-18ºN

Southeastern Europe zonal temperature flux at 100hPa, 25ºE-30ºE, 40ºN-45ºN

Historical model performance for prediction of June-September precipitation is summarized in Table 2, corresponding to the hindcasts plotted in Figure 7. This summary is intended to give a general perspective on performance rather than a detailed or exhaustive assessment of model behavior. We consider the anomaly correlation coefficient (ACC) as a measure of linear relationships between predictions and observations independent of mean and variance, root mean square error (RMSE) of the anomalies to quantify errors in variance, and bias. We also quantify the Hit Score for predictions, which is calculated by dividing the observational record into three terciles (below normal, near normal, and above normal) and calculating the fraction of predictions that fell into the correct category over the period of analysis. To focus on extremes, we include both a standard Hit Score and a Hit Score for Extremes, which considers only the hit rate for top and bottom tercile events. All scores are for cross-validation out-of-sample predictions, rather than for model fit.

Figure 7: 1982-2017 hindcasts for June-September Blue Nile basin precipitation. Grey lines are statistical models, with the solid black line showing the ensemble average of statistical models applied in this project. Blue lines are NMME forecasts initialized in May of each year–each light blue line is the ensemble average of a single model and the dark blue line is the full NMME ensemble average for all models that had complete data records. Dashed black line is CHIRPS June-September precipitation anomaly averaged across the Blue Nile basin.

Table 2: Evaluation of statistical and dynamically-based hindcasts for the period 1982-2017 (for NMME models only 1982-2010 and 2012-2016 were available for all models).

Models ACC RMSE (mm) BIAS (mm) HitScore HitScore_Extremes
Gissila2004 0.36 72.7 -2.1 0.50 0.52
Korecha2007 0.52 62.3 -0.1 0.44 0.52
Ncholson2014 0.56 60.1 0.4 0.47 0.52
Segele2015 0.47 64.8 1.0 0.47 0.52
Alexander – this study 0.70 52.2 -3.5 0.61 0.70
Blum-1 – this study 0.70 51.7 1.0 0.58 0.65
Blum-2 – this study 0.43 67.1 1.1 0.39 0.48
Wu – this study 0.66 55.7 -1.4 0.64 0.70
Mean Statistical Model 0.66 56.7 -2.8 0.58 0.65
CMC1-CanCM3 0.23 89.5 485.8 0.35 0.45
CMC2-CanCM4 0.58 95.3 547.8 0.59 0.64
COLA-RSMAS-CCSM4 -0.08 80.3 -284.8 0.29 0.27
GFDL-CM2p1-aer04 0.29 73.5 -113.5 0.38 0.41
GFDL-CM2p5-FLOR-A06 0.19 72.3 98.3 0.47 0.50
GFDL-CM2p5-FLOR-B01 0.41 65.6 84.3 0.56 0.59
NASA-GEOSS2S 0.49 105.7 -281.1 0.44 0.55
NCEP-CFSv2 0.33 84.4 103.6 0.44 0.50
Mean NMME 0.55 57.9 80.1 0.62 0.68


Table 3: Evaluation of statistical and dynamically-based hindcasts for the period 2001-2017 (for NMME models only 2001-2010 and 2012-2016 were available for all models).

Models ACC RMSE (mm) BIAS (mm) HitScore HitScore_Extremes
Gissila2004 0.12 77.3 -12.9 0.39 0.36
Korecha2007 0.37 62.9 -18.1 0.28 0.36
Ncholson2014 0.4 61.3 -12.5 0.33 0.36
Segele2015 0.14 67.0 -18.0 0.44 0.45
Alexander – this study 0.77 42.3 -15.2 0.67 0.73
Blum-1 – this study 0.49 58.7 1.3 0.39 0.36
Blum-2 – this study 0.45 60.5 -9.1 0.33 0.36
Wu – this study 0.69 48.5 6.1 0.61 0.64
Mean Statistical Model 0.72 51.3 13.0 0.44 0.55
CMC1-CanCM3 0.57 71.2 461.8 0.38 0.5
CMC2-CanCM4 0.54 85.3 536.2 0.62 0.7
COLA-RSMAS-CCSM4 -0.4 81.4 -306.9 0.31 0.2
GFDL-CM2p1-aer04 -0.05 82.1 -133.3 0.62 0.6
GFDL-CM2p5-FLOR-A06 -0.2 73.5 82.5 0.38 0.3
GFDL-CM2p5-FLOR-B01 0.32 63.8 69.9 0.44 0.5
NASA-GEOSS2S 0.31 95.3 -245.9 0.5 0.5
NCEP-CFSv2 0.32 87.9 68.2 0.31 0.4
Mean NMME 0.54 56.3 66.6 0.63 0.7


Our evaluation of NMME members confirms previous work with ensembles of dynamically-based prediction systems, in that performance is quite mixed across different models. The NMME ensemble average prediction, however, shows predictive skill that is comparable to many of the statistical models, and the NMME hit rate is as good as the top performing statistical models. Table 2 shows large bias values for some NMME models prior to bias correction, but this does not affect the application of the models to study anomalies. We note that these are direct NMME forecasts of precipitation in the GERD watershed. We did not consider “hybrid” approaches here, in which dynamically-based forecast systems are used to predict the predictors for a statistical model (e.g., Shukla et al. 2014, Gleixner et al. 2017), and which represent another powerful way to make use of dynamically-based models.

For the statistical models, we find that skillful predictions can be obtained using a number of different model structures and sets of predictors. All models rely to some extent on associations with surface conditions in the Tropical Pacific and/or Indian Ocean, reflecting the known importance of lagged ENSO connections and Indian Ocean conditions to the region. Skillful models also generally include some combination of predictors that capture the influence of Tropical Atlantic conditions or of atmospheric predictors that capture synoptic dynamics affecting East Africa, though the combination of these predictors varies by model. This is consistent with the recent work of Nicholson (2014) and Segele et al. (2015) who showed that diversifying predictors beyond the traditional ENSO and Indian Monsoon zones can enhance predictive skill. The fact that skill can be obtained using different combinations of these predictors makes it difficult to apply these prediction-oriented models to advance fundamental understanding of the drivers of summertime rainfall variability. These mechanisms, including the relationship between correlated predictors in these models, is the subject of ongoing research in the climate community. The models included here show some range in statistical performance. In general, the models that were customized for Blue Nile basin rainfall and developed using more recent data records show slightly higher skill for this application. A prediction based on the mean forecast of all statistical models fares well in both evaluation time periods.

While the 1982-2017 baseline has the advantage of offering a 36 year record for model evaluation, we do not necessarily expect that model skill is stationary over this full period. Indeed, the strength of various teleconnections to East Africa is known to vary as a function of climate variability, and over the past 37 years it is also possible that global warming has altered climate dynamics relevant to prediction. For this reason we also evaluate all models over the shorter 2001-2017 period (Table 3). Evaluation for this shorter record is not as statistically robust, but it offers a view of how well models are doing under more current conditions. All of the previously published statistical models show poorer performance for 2001-2017 than they do for the whole record. This is not surprising, as these models were developed without using the most recent years that are now available; we recalculated the coefficients using these recent years, but we did not revisit model structure. The magnitude of the drop-off is rather large, however, which is consistent with the observation that correlation with previously reliable predictors in the Pacific Ocean has been inconsistent since 2000. Several of the models we developed specifically for this study maintain their level of skill when tested against this more recent period. NMME models show some changes in skill for the 2001-2017 period, but these changes show no evidence of being systematic, at least within the limits of this cursory analysis. In sum: statistical approaches provide meaningful skill for seasonal forecast of Blue Nile basin rainfall, though there is considerable unexplained variance for all modeling approaches. The NMME models vary widely in their performance, but the NMME ensemble average offers respectable performance relative to many statistical approaches.

Blue Nile flow

Both the statistically and dynamically-based models described above predict rainfall. To convert a rainfall forecast to a Blue Nile streamflow forecast we adopted a variation of the Water Balance model (WatBal; Yates 1996). It takes monthly precipitation (P), and climatological monthly mean temperature (T) and diurnal temperature range (DTR) to produce monthly streamflow within a river basin. CRU TS data are used for climatological T and DTR. CenTrends and CHIRPS data are used for precipitation. Note that in order to take the predicted JJAS total precipitation as the updated inputs to produce the predicted streamflow, the model is calibrated using the disaggregated historical JJAS monthly P and the climatology of T and DTR. Climatological rainfall was used for Oct-Dec of 2018, as there is little rain in these months.

The calibrated WatBal model offers strong performance in the historical record. The monthly time series show a correlation of 0.96 (p-value <.0001) between the observation and calibration (Figure 6A). For interannual variability on  total flow the correlation is 0.77 (p-value <.0001). The scatter plot of monthly observation and calibration also aligns with the 45 degree line (Figure 6B). However, in general the calibrated streamflow shows an overestimation during early months of the calendar year while an underestimation during the later months (Figure 6C). Accumulatively, the calibrated streamflow is approximately 4% higher than the observed streamflow annually. Flow estimates generated by applying precipitation hindcasts to the hydrological model are shown in Figure 6D.

Figure 8: (A) comparison of observed streamflow at El Diem and simulated calibrated streamflow from WatBal, in units of million cubic meters (Mm3); (B) scatter plot of observed monthly streamflow at El Diem (x-axis) and calibrated streamflow from WatBal (y-axis); (C) monthly average streamflow based on observations at El Diem and calibrated streamflow from WatBal. Data for all plots is 1965-2009; (D) hindcasts of June-Dec total Blue Nile flow at the GERD site (Mm3) generated by applying statistical (grey lines) and dynamically-based (blue lines) precipitation hindcasts to the WatBal model. Dashed black line in (D) is the result of historical WatBal simulation driven with CHIRPS rainfall.

The outlook for 2018

The outlook for 2018 is highly informed by the fact that we are currently in the waning months of a La Niña event, which is expected to fade to neutral conditions that will persist through the summer rains (Figure 8). Since most droughts in the Blue Nile basin occur during El Niño events, current conditions and the majority of projections through the season suggest that drought is unlikely to occur. There is, however, the potential for El Niño conditions to emerge before the end of the season, which might affect late season rains in a manner that the statistical models considered here do not directly capture. Many factors play into climate variability in the Blue Nile basin, but the absence of El Niño provides a strong precondition towards average or above average Blue Nile flows.

Figure 9: Oceanic Niño Index forecast for 2018, generated by NOAA CPC and the International Research Institute (IRI) and issued on May 14, 2018.

As shown in Figure 2, both statistical and dynamical models suggest that there is a strong likelihood of average to above average rainfall in the Blue Nile basin in the 2018 summer rainy season. The predictions are not uniform, however. There is a distribution in statistical models that includes three models with a median prediction that is in the upper tercile of the historical record, four models that are in the middle tercile, and one that is in the lower tercile (Figure 10). For NMME, predictions for seven of the eight models we were able to obtain have a median forecast in the top tercile. There is, however, considerable spread in the absolute forecast of the NMME models.

Figure 10: Boxplots illustrating range of NMME ensemble predictions and statistical models with re-sampled error added back in. Dashed lines show below/above normal conditions. Statistical boxes represent range of 2018 forecast plus 100 resampled errors (from 1982-2017 leave-one-out predictions).

The spread shown in Figure 10 is a combined result of differences in rainfall variability between models–GEOS_S2S, for example has a large interannual standard deviation in precipitation, and Figure 10 shows bias corrected but not standardized forecasts–and of differences in the forecast relative to each model’s historical record. Standardized anomalies range from 0.5 to > 2.0 standard deviations above the historic mean. We chose not to standardize when presenting results because the majority of models have variability that is smaller than CHIRPS observed interannual variability. This means that for a wet forecast like 2018, presenting the absolute rather than variance-adjusted anomalies represents a conservative error when assessing the potential volume of excess flow predicted to be available. Ensemble spread across realizations of each NMME model is also substantial, with select ensemble members of several models yielding below average precipitation forecasts. The uncertainty in both statistical and dynamically-based models is clear in the smoothed histograms shown in Figure 11, which indicate that both families of forecasts include some members that allow for below average conditions to develop in 2018, even if it is a low probability.  Figure 12 shows the same smoothed distribution for streamflow estimates.

Figure 11: Smoothed histograms showing climatology (1982-2017) compared to statistical and NMME predictions for 2018 June-September rainfall. Dashed lines show below/above normal conditions. Note: climatology (n=37),NNME (n=8),statistical (n=8)

Figure 12: As in figure 11, but for June-December streamflow at the site of the GERD. Note: climatology (n=37),NMME (n=8),statistical (n=8)

These precipitation forecast results can be compared to those issued by operational forecasting institutions. The dynamically-derived IRI forecast of May 15, 2018, for example, shows somewhat elevated probability of high rainfall over much of the Blue Nile basin (Figure 13). The NOAA Climate Prediction Center (CPC) categorical forecast, calculated from the full, unweighted NMME ensemble, also shows a slightly elevated probability of above average rainfall in Ethiopia (Figure 14). ECMWF June-August precipitation forecasts are for normal rainfall conditions in the Blue Nile region. The GHACOF and Ethiopia NMA forecasts were not available at time of writing, but we will update to include them when they go online.

Figure 13: IRI July-September precipitation forecast, derived from weighted analysis of dynamically-based forecast systems. Red box indicates approximate location of the GERD watershed.

Figure 14: CPC categorical predictions for July-Sep precipitation in the May NMME ensemble.


While seasonal precipitation and hydrological forecasts of the Blue Nile River carry substantial uncertainties, the relative consistency of the 2018 prediction across models of different origin and structure provides some confidence that there is a high probability of average to above average flow in the coming season.  The sampling of models considered here suffer from inconsistent performance in recent years, sensitivity to calibration period (for statistical models) and to uncharacterized spatial biases (for dynamically-based systems), and the inherent limitations that come when forecasting an imperfectly understood system. Nevertheless, the consensus outlook is consistent with (and, indeed, driven in large part by) the understanding that drought in this region usually coincides with El Niño events, and we are currently experiencing a fading La Niña, with ENSO neutral conditions likely in the coming months. We do note that the ENSO teleconnection to the Blue Nile basin is not fully understood and appears to be inconsistent over time. This contributes to forecast uncertainty, and is an important area of ongoing research.

This exercise has been motivated in large part by the fact that Ethiopia will soon begin to fill the reservoir of the Grand Ethiopian Renaissance Dam. This is a major hydrological activity with considerable economic, social, and political implications. Decisions regarding the timing and rate of filling clearly depend on many factors that fall beyond the scope of seasonal forecast. Nevertheless, we feel that it it is valuable for Parties involved in filling decisions to have a common and realistic set of expectations for water availability in each year of the filling period. Consensus seasonal hydrological forecasts can inform those expectations on a year by year basis, perhaps easing one source of tension as riparian countries engage in the shared challenge of managing a transboundary waterway under rapid development.


Berhane, F., Zaitchik, B., & Dezfuli, A. (2014). Subseasonal analysis of precipitation variability in the Blue Nile River Basin. Journal of climate, 27(1), 325-344.

Gissila, T., Black, E., Grimes, D. I. F., & Slingo, J. M. (2004). Seasonal forecasting of the Ethiopian summer rains. International Journal of Climatology, 24(11), 1345-1358.

Gleixner, S., Keenlyside, N. S., Demissie, T. D., Counillon, F., Wang, Y., & Viste, E. (2017). Seasonal predictability of Kiremt rainfall in coupled general circulation models. Environmental Research Letters, 12(11), 114016.

Kirtman, B. P., Min, D., Infanti, J. M., Kinter III, J. L., Paolino, D. A., Zhang, Q., … & Peng, P. (2014). The North American multimodel ensemble: phase-1 seasonal-to-interannual prediction; phase-2 toward developing intraseasonal prediction. Bulletin of the American Meteorological Society, 95(4), 585-601.

Korecha, D., & Barnston, A. G. (2007). predictability of June–September rainfall in Ethiopia. Monthly weather review, 135(2), 628-650.

Korecha, D., & Sorteberg, A. (2013). Validation of operational seasonal rainfall forecast in Ethiopia. Water Resources Research, 49(11), 7681-7697.

Nicholson, S. E. (2014). The predictability of rainfall over the Greater Horn of Africa. Part I: Prediction of seasonal rainfall. Journal of Hydrometeorology

Nicholson, S. E. (2017). Climate and climatic variability of rainfall over eastern Africa. Reviews of Geophysics.

Segele, Z. T., Richman, M. B., Leslie, L. M., & Lamb, P. J. (2015). Seasonal-to-interannual variability of ethiopia/horn of Africa monsoon. Part II: Statistical multimodel ensemble rainfall predictions. Journal of Climate, 28(9), 3511-3536.

Shukla, S., Funk, C., & Hoell, A. (2014). Using constructed analogs to improve the skill of National Multi-Model Ensemble March–April–May precipitation forecasts in equatorial East Africa. Environmental Research Letters, 9(9), 094009.

Climate Change Likely to Contribute to Another East African drought in 2018

by Chris Funk


I needed to come in to the office today (Martin Luther King Day) to work on another blog because I am worried about the strong probability of yet another East African drought in March-May of 2018. I wanted to come in to work today because it is Martin Luther King Junior’s birthday, and this is one way of honoring his life. Dr. King has always been a personal hero, teaching us to care deeply for each other, fighting for our shared humanity in a non-violent way.

It is not our genes that make us human, but rather our highest aspirations expressed in coherent action.

I would like to dedicate this post to all the first responders in Africa and California, who have helped so many face the ravages of drought, fire, and flood.

The main objective of this post is to summarize some of our climate change research, and link it to a pessimistic outlook for the March-May East African rains. But I can’t not mention that since the Whittier Fire broke out near my house in the woods on July 8th of 2017, it seems like disasters in Santa Barbara have been rampant. The Whittier fire burned through July, August, and September. Then October, November, and December were exceptionally dry (Figure 1) and hot (Figure 2).  Many regions of the southwest have received less than half the amount of rain typically received by this point in winter. Even more striking, however, is the exceptional warmth, shown in Figure 2 averaged over the South Coast Drainage near Santa Barbara. Recent average air temperatures have been about 57° Fahrenheit. The 2017 air temperatures were about 61.5°F. This very warm weather and an almost total lack of rainfall resulted in very dry vegetation conditions. Everywhere I hiked this winter branches were brittle-dry, like matchsticks.

Figure 1
Figure 1. October 1 2017-January 14th precipitation, expressed as a percent anomaly.
Figure 2
Figure 2. October 1 2017-January 14th precipitation, expressed as a percent anomaly.

When fall transitions to winter, the Santa Barbara area starts to get ‘sundowner’ and ‘Santa Ana’ weather conditions, which blow rapidly down from the mountains towards to the sea. Typically, these occur after it rains and pose little risk. This year, however, these winds were extremely rapid, up to 70 miles per hour, and raced across tinder-like branches. When a fire broke out behind the city of Ventura on December 4th, these intense winds combined with abundant super-dry vegetation and a spark, producing the massive Thomas Fire. While aperiodic, the spread of the fire over the remainder of the month was at times insane, advancing an acre a second, leaping up the coast towards Carpentaria and then Montecito and Santa Barbara.

Then on Tuesday January 9th, a rainstorm struck the coastal mountains behind Montecito. Under normal conditions, such a storm would have had little impact. But the fire has been so recent the hills were bare, and the soil crusty; literally baked by the intense flames of the Thomas Fire.

Today the Santa Barbara/Montecito community feels crushed by the weight of the resulting mudflow which tore down through Montecito in the very early morning at speeds of up to 20 miles per hour. The Thomas Fire had been horrible. It caused many of us to evacuate, and all of us to breathe ash-filled air for weeks. More than a thousand structures had burned in what had quickly become the largest wildfire in recorded history. But the fire response had been massive and very successful. More than 8,000 courageous firefighters, 100 fire engines, and dozens of helicopters, sent from all across the western US had come together to defeat the Thomas Fire. On January 9th we lost at least twenty lives in a terrible flood that ripped all the way down to the coast. The Thomas Fire was not even 100% contained until January 12th.

The Walker Circulation links us with East Africa

While I am not an expert on California climate, I do think that there is an important climate feature that links California with East Africa: the Walker Circulation. Understanding this climate pattern helps inform successful predictions of drought, such as our group’s successful predictions of the 2016 October-December, 2017 March-May, and 2017 October-December droughts in East Africa. A recent assessment by the World Food Programme highlights the extent and some of the impacts of these sequential droughts: “This relentless sequence of climatic shocks together with insecurity and high food prices is having a profound negative impact on vulnerable populations”. Assessments by the Famine Early Warning Systems Network (FEWS NET) identify very large food insecure populations, especially in Somalia and Eastern Ethiopia.

To understand how fragile some of these populations are it helps to understand how deep poverty impacts how people eat. World Bank data for Kenya and Ethiopia suggest that the poorest 20% of households live on $330 and $251 dollars per year per person. In very poor households, most of this money (~60%) will typically be spent on food. If you live among a very vulnerable pastoral community, making your living off the itinerant grazing of yours flocks, you have to depend heavily on food purchases. Droughts can inflict a triple threat – food prices go up, purchasing power and savings erode as livestock die, and weak and dying herds also stop producing milk and meat, important sources of calories.

Figure 3
Figure 3. Schematic drawing showing the over-turning cells of the Walker Circulation.

While there are a lot of complexities involved in East African climate, the predictable components on climatic time scales tend to be associated with a large circulation feature called the ‘Walker Circulation’ (named after another personal hero of mine, Sir Gilbert Walker). Figure 3 shows a schematic depiction of the Walker Circulation. The 2D drawing goes from the Earth’s surface to the top of the troposphere. Moving from left to right we start in Uganda and end up in the eastern Pacific. The Walker Circulation is centered on the equator, though across the east Pacific the sub-tropical high pressure cells extend poleward, keeping California sunny and dry. Near Indonesia (~130°E), where the Indian and Western Pacific oceans meet, we find extremely warm ocean waters. Above these warm waters we find lots of rainfall and ascending air. In the lower atmosphere, winds blow into this region from both the equatorial Pacific and Indian Ocean, helping to produce moisture convergence and ascending air motions. In the upper atmosphere (near 15 km) the atmosphere becomes very stable, and the ascending air parcels are shunted to the east or west. Most of the energy carried by these parcels ends up reinforcing the east Pacific sub-tropical highs. These high pressure cells, in turn, drive the Pacific trade winds that blow west towards Indonesia, reinforcing the Walker Circulation. It’s these cells that help make it dry and sunny in California.

Over the Indian Ocean, there is also an Indian Ocean branch of the Walker Circulation. This circulation is associated with dry subsiding air over eastern East Africa, and dry hot surface conditions. To a large extent, eastern East Africa is dry because of the moisture and energy convergence over Indonesia. This convergence drives rising air, some of which sinks over East Africa.

Climate variations like El Niños, La Niñas, and warming trends in the oceans can modulate the strength of the Walker Circulation, increasing or reducing the probability of East African droughts. Research by the Climate Hazards Group and FEWS NET (discussed below) has suggested that climate change is increasing the strength of the Walker Circulation, reducing rainfall during the March-May long rainy season, and increasing the impact of La Niña events.

 Why climate change is increasing the frequency of East African droughts
The Climate Hazards Group Perspective

Declines in East African rainfall were first identified as part of our efforts to improve drought early data sets for the FEWS NET. Our early climate analyses correctly isolated the immediate cause of the drying (an intensification of the Indian Ocean branch of the Walker Circulation), but probably over-emphasized the role played by the Indian Ocean. In the early 2010s, we doubled down on our efforts to collect and analyze the best possible rainfall data, and published a series of reports for the US Agency for International Development. Our Kenya report, for example, showed the strong correspondence between a measure of low frequency warming, the Indo-Pacific Area Index and declines in Kenya rainfall. Most of this work has focused on the March-May rainy season. We have refined our ideas over time, going through the classic science progression of ‘description’, ‘explanation’, and ‘prediction’. Here I just recap some of the description, explanation, and prediction efforts, with the caveat that many have contributed to these papers and many other good studies have also been done (this is not an exhaustive literature review). A long list of publications is shown at the end of this post.


With support from the FEWS NET, we have been able to invest heavily in building the best possible international data sets for East Africa (see A and B). While there is essentially no debate about the recent declines of the East African March-May ‘long rains’ these data sets let us say with confidence that there has been a large persistent increase in the frequency of poor East African seasons. Figure 4 shows a long time series of standardized rainfall averaged over eastern/central Kenya, southeastern Ethiopia, and Somalia. This image is from a recent paper, currently in final review.

Figure 4
Figure 4. A long time series of rainfall for the region outlined in the left panel show a very large increase in the frequency of dry seasons. Since 1998, only 2010 and 2013 exhibited healthy long rains. Based on CHIRPS and Centennial Trends rainfall as described here:

Since 1998, only 2010 and 2013 exhibited above normal rainfall performance. Paleoclimate analyses by Jessica Tierney and co-authors also indicate a long term decline and a negative relationship to global warming. It should be noted that the October-December ‘short’ rains have not exhibited similar declines, probably due to warming in the eastern Indian Oceans. Nonetheless, the dramatic increase in the frequency of East African March-May droughts threatens the economic stability of poor households. More frequent climate shocks make it very hard to build up reserves and escape from a cycle of poverty.


Our ability to explain the mechanisms behind the recent declines has been advancing quickly. This work has tended to emphasize the impact of the ‘West Pacific Warming Mode’ – a low frequency warming trend mode found in both the observed sea surface temperature record and climate change simulations. The warming mode pattern is associated with a characteristic ‘Western V’ sea surface temperature pattern over the Pacific. This pattern is also quite similar to the ocean temperature anomalies found during recent East African March-May droughts.

Figure 5
Figure 5. Standardized March-May sea surface temperature composites based on signature East African drought years (1984, 2000, 2004, 2009, 2011).

We can characterize the sea surface temperature patterns associated with these droughts by averaging conditions during recent severe dry seasons (Figure 5). This figure shows an area where warming is associated with less rainfall over East Africa (primarily the Western North Pacific, outlined with a yellow box), and an area in the eastern equatorial Pacific where warm (El Niño-like) sea surface temperatures are associated with more rainfall over East Africa. The area over the eastern equatorial Pacific is strongly influenced by El Niños and La Niñas – the major form of inter-annual climate variability. El Niños warm the east Pacific and slow the Walker Circulation. La Niñas do the opposite. At decadal time scales, sea surface temperatures in the east Pacific have not warmed substantially, and this is a major difference between the observed climate and predictions made by climate change models.

The western Pacific, on the other hand, has tracked very closely with climate change predictions. This is an area where sea surface temperatures are strongly controlled by downwelling radiation. La Niñas also tend to produce warming as well. The combination of these two influences are shown in Figure 6 (details here). Western North Pacific Ocean temperatures have increased dramatically, in line with climate change predictions. It is important to note, however, that this region warms during La Niña events as well. The combination of long term warming and the responses to La Niña events creates a strong gradient between the western and central Pacific amplifying the impact of La Niña events.

Figure 6
Figure 6. The red line shows west Pacific sea surface temperatures from the yellow box in the previous figure. The blue line shows the same from a large set of climate change simulations. The green line shows estimates based on a combination of climate change averages and the influence of La Ninas and El Ninos.

We can now explain fairly well why a ‘Western V’-like pattern amplifies the Walker Circulation, linking increases in sea surface temperatures (Figure 6) with declines in the East African long rains (Figure 4). Near the surface, equatorial sea level pressure values track closely with ocean temperatures. As shown by many recent studies, warm west Pacific Ocean conditions are associated with lower pressures and convergent low level winds that blow from the Indian and Pacific into the area around Indonesia, enhancing the Walker Circulation. Figure 7 quantifies the effects of the warming west Pacific by contrasting new (1981-2016) west Pacific warm events with old (1921-1980) warm events. This figure depicts rainfall and low level circulation changes based on a large set of atmospheric model simulations. The details of the drying signal over East Africa are not reproduced too well, but note the strong modulation of the moisture bearing winds over the Indian Ocean (blowing away from Africa). These simulations identify large (more than one and a half standardized anomaly) increases in precipitation over Indonesia. This indicates a large increase in the strength of the Walker Circulation (Figure 3), which tends to increase subsidence over East Africa, increasing the frequency of droughts.

Figure 7
Figure 7. Composites of a large (20 member) ensemble of CAM5 atmospheric model simulations to diagnose changes in precipitation and low level (850 hPa) wind anomalies by contrasting 1981-2016 and 1921-1980 warm WNP events. Image produced by Laura Harrison.
Figure 8
Figure 8. Here we present estimates based on ensemble climate change averages (blue line) and estimates based on climate change averages and the influence of ENSO, based on regressions with NINO3.4 sea surface temperatures.

Looking further up in the atmosphere, things become even more interesting. Examining similar atmospheric change composites (Figure 8) we can see the full implications of a ‘Western V’ warming pattern. This map shows changes in upper level geopotential heights (i.e. the maps of high and low pressure systems we see in weather reports) along with changes in upper level winds. Over the northern and southern mid-latitudes, we see very large (~+50 m) increases in heights. These high pressure cells sit right underneath the sub-tropical westerly jet. Smaller, but equally important, are the increases in upper level heights found near Indonesia. Taken together, this pattern of high pressure sends upper level winds first to the west across the northern and southern Pacific near 30°N/S, and then turns those winds towards the central Pacific near the equator. These anomalous flows from the north and south converge east of 150°E, producing subsidence, and enhancing the Walker Circulation, creating a La Niña-like response. Over the western Indian Ocean we see an enhancement of the Indian Ocean branch of the Walker Circulation. When warm Western V conditions combine with cool La Niña conditions in the East Pacific, we are set to have an elevated chance of East African drought.

Attribution Studies

Since 2012, we have been active participants in the special climate attribution issue of the Bulletin of the American Meteorological Society’s annual issue on extreme event attribution. This special issue examines each year’s extreme events, and formally assesses questions of climate attribution. Articles in these special issues ask: can extreme events from each year be attributed to anthropogenic influences?

Our analysis of the 2014 East African March-May drought found: “Anthropogenic warming contributed to the 2014 East African drought by increasing East African and west Pacific temperatures, and increasing the gradient between standardized western and central Pacific SST causing reduced rainfall, evapotranspiration, and soil moisture”. Our analysis of the 2015 Ethiopian drought found: “Anthropogenic warming contributed to the 2015 Ethiopian and southern African droughts by increasing El Niño SSTs and local air temperatures, causing reduced rainfall and runoff, and contributing to severe food insecurity”. In general, our work suggests that both El Niño and La Niña-like climate disruptions may be made more intense by climate change, leading to opportunities for prediction.


Our ability to predict long rains droughts has moved through three stages. During Stage 1 (~2004-2009) an increased drought frequency had been identified and related to the Walker Circulation. At this point we had little predictive skill. During Stage 2 (2010-13) we were identifying the link between Indo-Pacific warming and an increase in La Niña impacts on the Greater Horn of Africa. This helped support a successful prediction of the severe famine-producing 2010-11 drought. During Stage 3, we then started making the link between the west Pacific gradient and using statistical models to predict East African droughts. These latter efforts are quite similar to the successful Climate Outlook Forum process used by East African scientists. This work is important, because dynamic forecast models still continue to miss major drought events, such as the March-May and October-December droughts from 2017 (Figure 9 and 10). Partnering with many agencies we were able to use ‘Stage 3’ type forecasts to provide an effective early alert before the poor March-May 2017 season. In the early spring of 2017, timely humanitarian response helped prevent a repeat of the catastrophic runaway food prices seen in 2011. Unfortunately, we may be looking at yet another poor rainy season. The rest of this post describes a statistical forecast for the March-May 2018 season.

Figure 9
Figure 9. International Research Institute rainfall forecast for March-May 2017.
Figure 10
Figure 10. International Research Institute rainfall forecast for October-December 2017.

We begin by looking at the correlation between observed 1998-2017 CHIRPS rainfall over central/Eastern Kenya, all of Somalia, and central/southeastern Ethiopia (see Figure 3 from here for the precise region) and observed December NOAA Extended Reconstruction sea surface temperatures (Figure 11). We find a fairly strong negative correlation with the ‘Western V’ and a fairly strong positive correlation with the equatorial eastern Pacific. This relationship holds up reasonably well back to about 1993, but then seems to disappear, leading some climatologists to claim that the long rains are weakly linked to Indo-Pacific sea surface temperatures and relatively unpredictable. Our view on this subject is quite different.

Figure 11
Figure 11. 1998-2017 correlations between observed March-May East African rainfall and December sea surface temperatures.

Following the very large 1997/98 El Niño, sea surface temperatures in the west Pacific warmed substantially (Figure 6). La Niña-like March-May seasons have been associated with much larger Walker Circulation intensifications (Figure 7), that draw near surface winds over the Indian Ocean away from Eastern Africa (Figure 8). Looking at the upper-troposphere (Figure 8), we see large increases in the westward winds over East Africa, characteristic of an enhancement of the Indian Ocean branch of the Walker Circulation (Figure 3). This enhancement can help explain the emergent correlation structure plotted in Figure 11 and the fact that ‘new’ La Niñas appear to have a stronger negative impact on the East African long rains.

Figure 8 also helps understand the correlation pattern shown in Figure 11. ‘Western V’ warming in the blue areas of Figure 11 tends to increase upper level heights with the ‘U’ shaped pattern shown in Figure 8. Cooling in the red/orange areas of Figure 11 tends to reduce upper level heights within the two central Pacific cyclones shown in Figure 8.

So, dynamically, the Western V warming and central Pacific cooling patterns tend to fit ‘hand-in-glove’. The upper-level cyclones produced by cool central Pacific conditions slot neatly within the surrounding height increases associated with warmer Western V sea surface temperatures, produce intense convergent wind patterns and subsidence across the eastern equatorial Pacific. This enhances the Walker Circulation and increases the frequency of droughts over East Africa.

Figure 12
Figure 12. December 2017 NOAA Extended Reconstruction Version 4 sea surface temperature anomalies. Based on a 1981-2010 baseline.

Conditions in December of 2017 (Figure 12) appear conducive to below-normal March-May 2018 rains. We see generally warm conditions across the ‘Western V’ region and La Niña sea surface temperatures across the eastern equatorial Pacific. We use the magenta and cyan regions from Figure 12 to develop predictive relationships.

Figure 13
Figure 13. Scatterplot between December Western V sea surface temperatures and March-May East African precipitation.
Figure 14
Figure 14. Scatterplot between December East Pacific sea surface temperatures and March-May East African precipitation.

Figures 13 and 14 show, respectively, scatterplots between 1998-2017 December Western V and eastern Pacific sea surface temperatures and standardized March-May East African rainfall. Both predictive relationships are similar in magnitude (R=-0.70 and +0.68). Red dots in Figure 13 and 14 show estimated outcomes for March-May 2018, based on each individual predictor. The results are quite similar, with forecasts of about a -1 standard deviation rainfall deficit.

Figure 15
Figure 15. Scatterplot of 1998-2017 Western V and East Pacific sea surface temperature values

Figure 15 shows the 1998-2017 Western V and East Pacific sea surface temperature values. Note that these time series are also inversely correlated (R=-0.70). This year appears typical for a recent moderate La Niña/East Africa drought season. Some Decembers looked worse (1998, 1999, 2010 and 2011), but many (13) Decembers looked better, so we might expect to find ourselves in the below normal rainfall tercile. The 2007, 2000, and 2016 seasons might be close analogs. The 2017 conditions appear much more characteristic of new drought years than 2016. That year was atypical, in that we had very warm Western V sea surface temperatures and fairly neutral East Pacific conditions.

Figure 16 (not included)
Figure 16. Scatterplot of predicted and observed East African precipitation values. Green circles show estimates based on take-one-away cross validation. The red circle shows the 2018 forecast, along with 80 percent confidence intervals.

We can use cross-validation to produce 1998-2017 forecasts for March-May 2018 (Figure 15). This forecast indicates a high probability of below normal rains. Note that this scatterplots suggests that we can identify most (eight out of nine) poor rainy seasons with a very low false alarm rate. We also see this in our Western V and East Pacific scatterplots (Figures 13 and 14); when Western V or East Pacific sea surface temperatures a particularly warm or cold (respectively) we do not see above normal Eastern East African March-May rains. Our below normal forecast appears broadly consistent with the most recent National Multi-Model Ensemble climate model forecasts. These forecasts also call for an end to La Niña conditions by mid-spring. This could alter the seasonal outlook substantially. This transition might suggest a more pessimistic outlook for the early part of the rainy season. Such transitional conditions tend to translate into more certainty for farmers in marginal areas. In semi-arid crop growing regions of central-eastern Kenya, Southern Somalia and Belg-dependent regions of Ethiopia, typical growing seasons are already short, so poor rains during the first half of the rainy season can be very disruptive. A poor Somali Gu harvest outlook may be one of the impacts of these predicted dry conditions, since this growing season is so short. Unfortunately, many of the La Niña high risk regions correspond to regions that received very poor rainfall totals over the prior three seasons: October-December 2016, March-May 2017, and October-December 2017 (Figure 17). Some regions of eastern Kenya and southern Somalia also experienced poor March-May 2016 conditions, and may be looking at five poor rainy seasons in a row.

Figure 17
Figure 17. Total CHIRPS rainfall for October-December 2016, March-May 2017 and October-December 2017 expressed as an anomaly from the 1981-2016 average.


Relevant Recent Research

Williams P. and Funk C. (2010) A Westward Extension of the Tropical Pacific Warm Pool Leads to March through June Drying in Kenya and Ethiopia, USGS Openfile Report 1199,

Funk, C., Eilerts, G., Davenport, F., and Michaelsen, J., (2010) A Climate Trend Analysis of Kenya-August 2010, USGS Fact Sheet 2010-3074: 

Funk, C. (2011) We thought trouble was coming, Nature Worldview, 476.7.

Funk C, J. Michaelsen and M. Marshall (2012) Mapping recent decadal climate variations in precipitation and temperature across Eastern Africa and the Sahel, Chapter 14 in “Remote Sensing of Drought: Innovative Monitoring Approaches”, edited by B. Wardlow, M. Anderson and J. Verdin, Taylor and Francis, 25 pages.

Galu, G., J. Kere, C. Funk, and G. Husak (2011) Case study on understanding food security trends and development of decision-support tools and their impact on a vulnerable livelihood in East Africa. Global Assessment Report on Disaster Risk Reduction 2011, United Nations, International Strategy for Disaster Risk Reduction, Geneva, Switzerland.

Liebmann, B., Bladé I., Kiladis G. N. , Carvalho L. M. V., Senay, G., Allured D., Leroux S., Funk C (2011) African Precipitation Seasonality Based on Daily Satellite Data from 1996-2009, J. of Climate, Journal of Climate 25, no. 12 (2012): 4304-4322.

Funk, C., (2012) Exceptional warming in the western Pacific-Indian Ocean Warm Pool has contributed to more frequent droughts in Eastern Africa, Bull. Amer. Met. Society,v7(93) p. 1049-1051.

Hoell, A. and C. Funk (2013): The ENSO-related West Pacific Sea Surface Temperature Gradient, Journal of Climate. J. Climate, 26, 9545–9562.      doi:

Funk C., G. Husak, J. Michaelsen, S. Shukla, A. Hoell, B. Lyon, M. P. Hoerling, B. Liebmann, T. Zhang, J. Verdin, G. Galu, G. Eilerts, and J. Rowland, 2013: Attribution of 2012 and 2003-12 rainfall deficits in eastern Kenya and southern Somalia [in “Explaining Extreme Events of 2012 from a Climate Perspective”]. Bull. Amer. Meteor. Soc., 94, S45-S48.

Hoell A. and C. Funk (2013) Indo-Pacific Sea Surface Temperature Influences on Failed Consecutive Rainy Seasons over Eastern Africa, Climate Dynamics, DOI: 10.1007/s00382-013-1991-6. p1-16.

Hoell, A., Funk, C., & Barlow, M. (2013). The regional forcing of Northern hemisphere drought during recent warm tropical west Pacific Ocean La Niña events. Climate Dynamics, 1-23.

Liebmann, B., Hoerling M., Funk C., Dole, R. M., Allured A., Pegion, P., Blade I., Eischeid, J.K, (2014) Understanding Eastern Africa Rainfall Variability and Change, Journal of Climate.

Funk, C., Hoell, A., Shukla, S., Bladé, I., Liebmann, B., Roberts, J. B., and Husak, G. (2014). Predicting East African spring droughts using Pacific and Indian Ocean sea surface temperature indices. Hydrology and Earth System Sciences Discussions, 11(3), 3111-3136.

Shukla, S., McNally, A., Husak, G., & Funk, C. (2014). A seasonal agricultural drought forecast system for food-insecure regions of East Africa. Hydrology and Earth System Sciences Discussions, 11(3), 3049-3081.

Shukla S., Funk C. and Hoell A. (2014) Using constructed analogs to improve the skill of March-April-May precipitation forecasts in equatorial East Africa , Env. Res. Letters, Environmental Research Letters 9.9 (2014): 094009.

Funk, C.,  Hoell A., Husak G., Shukla S. and Michaelsen J. (2015) The East African monsoon system: seasonal climatologies and recent variations. Chapter for “The Monsoons and Climate Change”, Leila M. V. Carvalho and Charles Jones (Eds.)

Funk, C. and Hoell A. (2015) The leading mode of observed and CMIP5 ENSO-residual sea surface temperatures and associated changes in Indo-Pacific climate, J. Climate, 28, 4309-4329,

Shukla S., Safeeq M., Aghakouchak A., Guan K. and Funk C. (2015) Role of Temperature in the Water Year 2014 Drought in California, Geophysical Research Letters, 42(11), 4384-4393. DOI: 10.1002/2015GL063666

Funk C., Nicholson S. E., Landsfeld M., Klotter D., Peterson P. and Harrison L. (2015) The Centennial Trends Greater Horn of Africa Precipitation Dataset, Scientific Data, 2, 150050. DOI:  10.1038/sdata.2015.50. doi:10.1038/sdata.2015.50.

Funk, C., Shukla S., Hoell A. and Livneh B. (2015) Assessing the contributions of East African and west Pacific warming to the 2014 boreal spring East African drought, BAMS Climate Attribution Issue. 96.12 (2015): S77-S82.

Shukla S., Roberts, J., Hoell A., Funk C., Robertson F. and Kirtmann, B. (2016) Assessing North American Multimodel Ensemble (NMME) Seasonal Forecasts, Climate Dynamics. 1-17.

Funk, C., Harrison L., Shukla S., Hoell A., Korecha D., Magadzire T., Husak G., and Galu G., 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, December 2016, S75-S77.

Funk, C and Hoell A. (2017) Recent Climate Extremes Associated with the West Pacific Warming Mode, AGU Monograph-Climate Extremes: Patterns and Mechanisms. Ed. by Simon Wang,  Jin-Ho Yoon, Chris Funk and Robert Gillies, published by Wiley Press (2017): 165. ISBN: 978-1-119-06784-9.

Liebmann B., Bladé I., Funk C., Allured D., Hoerling, M., Hoell, A., Peterson, P., Thiaw, M.T. (2017) Climatology and Interannual Variability of Boreal Spring Wet Season  Precipitation in the Eastern Horn of Africa and Implications for its Recent Decline, J. Climate, Journal of Climate 30.10 (2017): 3867-3886,

Brown M.E., Funk, C., Pedreros, D., Korecha D., Lemma M., Rowland J., Williams E. and Verdin, J. (2017) Climate Trend Analysis of Ethiopia-Examining Subseasonal effects on Crops and Pasture Conditions, Climatic Change, DOI 10.1007/s10584-017-1948-6.

Funk C, Davenport F, Harrison L, Magadzire T, Galu G, Artan G, Shukla S, Korecha D, Indeje M, Pomposi C, Macharia D and Husak G (2017) Anthropogenic enhancement of moderate-to-strong El Niños likely contributed to drought and poor harvests in Southern Africa during 2016, Bulletin of the American Meteorological Society, 37.S1-S3, DOI. 10.1175/BAMS-D-17-0112.2.

Philip S, Kew S.F., Jan van Oldenborgh G, Otto F, O’Keefe S, Haustein K, King A,  Zegeye A, Eshetu Z, Hailemariam K, Singh R, Jjemba E, Funk C,  Cullen H, Attribution analysis of the Ethiopian drought of 2015, J. Climate,

Funk C, McCormick S, Galu G, Massawa, E., McCormick S, Omondi P, Sebina E, Shitote S, White L (2017), Climate Change Vulnerability Impact Assessments and Adaptation in East Africa, Summary for Policy Makers, USAID PREPARED Project.

Nicholson, SE, Funk, C. and Fink A (2017) Rainfall over the African continent from the 19th through the 21st century, J Global and Planetary Change, In Press.

Funk C, Harrison L, Shukla S, Pomposi C, Galu G, Korecha D, Husak G, Magadzire T, Davenport F, Hillbruner C, Eilerts G, Zaitchik B and Verdin J (2017) Examining the role of unusually warm Indo-Pacific sea surface temperatures in recent African droughts, Q. J of the R. Meteorological Society. In Review, minor revisions.

Reformulated NMME simulations indicate probable dryness for the 2017 and 2018 short and long rains in eastern East Africa

Chris Funk, Andy Hoell, Wassila Thiaw, Diriba Korecha, Gideon  Galu, Miliaritiana Robjhon, Endalkachew Bekele, Shraddhanand Shukla, Pete Peterson, and James Rowland


This analysis has two objectives:  to i) provide a timely multi-agency assessment of the forthcoming short and long rainy seasons for eastern East Africa, while also ii) advancing towards a multi-agency paper describing a ‘FEWS NET Integrated Forecast System’ (FIFS). The main ideas behind FIFS are to improve the accuracy of seasonal precipitation forecasts by using statistical ‘Model Output Statistic’ (MOS) reformulations based on output from coupled ocean-atmosphere models, such as the North American Multi-Model Ensemble (NMME).  Note that while the work presented here is ‘experimental’, it is backed by years of FEWS NET research and successful forecasts based on similar approaches. The basic relationships used to drive these forecasts involve La Niña-like modifications of the Walker Circulation, which FEWS NET research, motivated by the 2011 famine in Somalia, has shown to be a key driver of droughts during both the October-November-December ‘short’ rains and the March-April-May ‘long’ rains.  Please see the list of papers at the end of this report.

The method explored here is called ‘adaptive regression’ (AdReg), which is an outgrowth of the modified constructed analog approach explored by Shukla and Funk.  AdReg is also similar to the process that an expert statistical climate forecaster might carry out, i.e. similar to the approaches used to derive outlooks for climate outlook fora.  AdReg has two steps: selection and estimation. AdReg employs three selection criteria: correlation, activity, and dynamic significance. In this application, ensemble average NMME precipitation fields will be used as predictors. We will use October-November-December (OND) NMME forecasts to predict OND eastern East African (EEA) rainfall and OND NMME forecasts to predict MAM EEA rainfall.


This work has been motivated by the severe food crisis in EEA. As announced in a recent FEWS NET alert, the prospect of a fourth, and perhaps even a fifth, consecutive drought looms over East Africa due to the emerging La Niña-like sea surface temperature conditions. At present, the world faces an extremely high level of food insecurity, with an estimated 81 million people requiring urgent assistance. In East Africa, consecutive droughts have pushed millions to edge of starvation. In Somalia, food insecurity remains very high, with close to a million people (more than 800,000) facing near-famine (Emergency) conditions, another 2.3 million people in a severe food crisis, and another 3.1 million facing substantial food stress; an estimated 388,000 children under the age of five are acutely malnourished. In Kenya, persistent high food prices continue to drive high levels of food insecurity. In the Somali region of Eastern Ethiopia food security conditions are approaching famine conditions, and over all there are ~5-10 million Ethiopians in facing acute food insecurity.

Figure 1 shows the early October Integrated Phase Classification (IPC) food insecurity outlook map from Orange and red areas in this map indicate areas experiencing a severe food crisis (orange) or emergency (red). These shades indicate severe food stress. For this study, we have selected the area outlined in black. This region was selected because of its high level of food insecurity, regional homogeneity, and fairly good data support.

Figure 1. FEWS NET Food Security Outlook from Early October
Figure 1. FEWS NET Food Security Outlook from Early October

Over the past thirty years, rainfall in this region during the OND season has remained fairly steady, probably due to the beneficial influence of warming western Indian Ocean SSTs. Rainfall during the MAM long rains, however, has declined substantially. Figure 2 shows a 1900-2017 March-April-May Standardized Precipitation Index (SPI) for this area, based the Centennial Trends (1900-1980) and CHIRPS (1981-2017) data sets, centered on a 1981-2010 baseline. The Centennial Trends and CHIRPS datasets are highly correlated (0.93) over the 1981-2014 time period.  What is striking about Figure 2 is large frequency of dry seasons since 1999. Only 2010 and 2013 experienced good rainfall performance. The onset of La Niña-like climate conditions combined with warm SSTs in the western Pacific could set the stage for another sequence of back-to-back droughts during the OND 2017 and MAM 2018 long rains.

Figure 2. Eastern East Africa MAM Standardized Precipitation Index time series.
Figure 2. Eastern East Africa MAM Standardized Precipitation Index time series.

Here, we make AdReg forecasts based on the ensemble average of the latest NMME model simulations. Bias corrected precipitation simulations have been provided by Brent Roberts and the NASA SERVIR program. These forecasts have been bias corrected using Global Precipitation Climatology Center precipitation fields. In general, the NMME models predict moderate La Niña conditions (Figure 3) during the fall and winter of 2017/18, followed (in most models and simulations). In Figure 4 we see the NMME OND precipitation forecast. Consistent with a moderate La Niña forecast and moderately warm West Pacific SSTs, we see strong subsidence at the dateline near the equator. Such subsidence is typically a robust indicator of an enhanced La Niña-like amplification of the Walker Circulation. Over East Africa, we see a general tendency to slightly above normal rainfall.  An examination of historical correlation between the NMME and OND rainfall in our target region, will question and revise this assessment (below).

Figure 3. NMME NINO3.4 forecasts, from
Figure 3. NMME NINO3.4 forecasts, from
Figure 4.Standardized OND NMME ensemble forecasts of precipitation.
Figure 4.Standardized OND NMME ensemble forecasts of precipitation.

Here we explore the application of ‘adaptive regression’, a new type of statistical reformulation being developed by FEWS NET. AdReg filters potential global climate predictors by three criteria: correlation, activity, and dynamic significance. Correlation is based on the historical correlation between our target and each locations historical SST or precipitation forecast time series. Our main target is 1997-2016 OND rainfall in our Eastern East Africa study site (Figure 1). We begin by using a threshold of |0.5| as our criteria for correlation.

Figure 5 shows the correlation between OND EEA precipitation and the NMME forecast SST and Precipitation. Correlation values less than 0.5 and greater than -0.5 have been set to 0. For OND we see a positive-negative-positive relationship to SST in the Western Indian, Western Pacific and Eastern Pacific Ocean. The precipitation correlation pattern looks similar to that shown in diagnostic analyses of the short rains. Note, however, the details of the correlation structure over Eastern Africa and the Indian Ocean. During dry EEA OND years, the model produces strong subsidence and drying to the east of East Africa, primarily over the Indian Ocean. In the actual NMME forecast (Figure 5), conditions are fairly neutral in this area, and fairly active over the ENSO-related regions of the western and eastern Pacific.

To generate AdReg forecasts, we follow two steps: selection and estimation. AdReg employs three selection criteria: correlation, activity, and dynamic significance. Here we use correlation threshold of 0.5, a standardized anomaly threshold of 0.5Z, and a mean precipitation threshold of 30 mm for these criteria. NMME locations a historical correlation of less than -0.5 or greater than +0.5 are retained.

We next screen for activity. Areas that have a standardized anomaly, in this season, of less than -0.5Z or greater than +0.5Z are retained. We then screen for ‘dynamic significance’; areas with long term average NMME precipitation of less than 30 mm are not used. Figure 6 shows an example of masking, based on the 2017 OND forecast process.

Figure 5. Correlation between observed 1982-2016 EEA OND precipitation and 1982-2016 NMME Precipitation forecast made in early October.
Figure 5. Correlation between observed 1982-2016 EEA OND precipitation and 1982-2016 NMME Precipitation forecast made in early October.
Figure 6. Mask used to predict the OND 2017 rains. Orange areas are positively correlated with observed EEA OND rainfall. Blue areas are negatively correlated.
Figure 6. Mask used to predict the OND 2017 rains. Orange areas are positively correlated with observed EEA OND rainfall. Blue areas are negatively correlated.

The next step in the AdReg process calculates the first and second principal components (PCs) of the NMME forecasts. As an example, Figure 7 shows the correlation between the first AdReg principal component and the NMME precipitation predictions. This is an ENSO signal. The final step uses regression to translate the PCs into forecasts of the predicted target.

Take-one-away cross-validation is used to assess the out-of-sample skill of the AdReg procedure. For each year, that year’s data is withheld, the AdReg process calculated and then forecast accuracy is evaluated.

It should be noted that selection processes can create artificial skill, even when using cross-validation. The AdReg process attempts to minimize this effect by working with the low order principal components of the NMME forecasts. In the NMME models, this often corresponds with ENSO-like patterns of variability. The results presented here tend to focus on ENSO-like patterns similar to those studied in FEWS NET diagnostic analyses (see papers below).

Figure 7. Correlation between AdReg PC1 and the OND NMME precipitation forecasts.
Figure 7. Correlation between AdReg PC1 and the OND NMME precipitation forecasts.


Results 1. The OND Outlook

We begin by presenting (Figure 8) results for OND 2017 based on a model with masking based on a historical correlation of |0.5|, as standard anomaly of at least ±0.5Z and a mean precipitation rate of 30 mm. This model performs fairly well (cross-validated correlation of 0.68 with standard error of 0.8Z). While some droughts are missed, most are captured fairly well, including droughts in 2010 and 2016. A brief sensitivity analysis (Table 1) indicates fairly robust performance across our model criteria, except for when we set the correlation value to a high value (0.7). These strict criteria likely induces over-selection which performs poorly under cross-validation. Also shown in table 1 are cross-validated correlation and standard error values based on the raw NMME OND forecasts over the study area. While the models have some skill (r=0.45), AdReg skills are higher. A correlation of 0.45 implies explaining ~20% of the variance, AdReg correlations (~0.68%) imply explaining 46% of the variance. Raw NMME forecasts, regressed against the target, are shown in Figure 8 (triangles). These results are based on the bias corrected NMME forecasts, more experimentation with the underlying NMME precipitation fields will be examined in later analyses.

The outlook for the OND EEA rains based on AdReg is -0.8Z ± 1Z, based on 80% confidence intervals. Below normal appears to be the most likely outcome.

Figure 8. AdReg forecasts for EEA OND rains. Cross-validated results for 1982-2016, and the 2017 forecast. Bars denote 80% confidence intervals. Also shown are estimated based on regression with the raw EEA OND NMME (triangles).
Figure 8. AdReg forecasts for EEA OND rains. Cross-validated results for 1982-2016, and the 2017 forecast. Bars denote 80% confidence intervals. Also shown are estimated based on regression with the raw EEA OND NMME (triangles).
Correlation Activity Mean Rainfall CV R CV Standard Err
|0.5| >±0.5Z > 30 mm 0.68 0.8Z
|0.7| >±0.5Z > 30 mm -0.4 1.4Z
|0.3| >±0.5Z > 30 mm 0.63 0.8Z
|0.5| >±0.5Z > 0 mm 0.70 0.7Z
|0.5| >±0.5Z > 100 mm 0.70 0.7Z
|0.5| >±0.0Z > 30 mm 0.69 0.7Z
|0.5| >±0.8Z > 30 mm 0.67 0.8Z
Raw EEA OND NMME Precipitation 0.45 0.9Z

Table 1. Sensitivity Analysis for the OND Rains.

Results 2. The MAM Outlook

We next examine the relationship between the NMME OND precipitation forecasts and historical MAM EEA precipitation. We use the past 20 years (1998-2017), because FEWS NET research has shown a much stronger teleconnection with La Niña following the 1997/98 El Niño event, when the western Pacific SSTs increased substantially. Interestingly, we see quite a strong correlation between the OND NMME precipitation forecasts and the observed MAM EEA rainfall in the following spring (Figure 9). This ENSO-like pattern likely indicates the likelihood of a persistent overturning circulation that continues to disrupt rainfall across the tropics. The magnitude of these correlations is quite high (~±0.7). The cross-validated AdReg correlation is high (+0.7), with a fairly low standard error (0.69Z). The associated 2018 forecast would be -0.6Z ± 0.9Z; a below normal outlook with considerable uncertainty. Figure 10 shows the scatterplot of cross-validated forecast results, along with the 2017 and 2018 MAM forecasts. Figure 10 shows the corresponding NMME correlation pattern for the MAM AdReg forecast; again a strong ENSO-like pattern (the PC1 time series is dominated by strong ENSO-like inter-annual variations).


These results suggest, in line with numerous studies, that La Niña-like conditions are associated with dry conditions over the easternmost section of equatorial East Africa. The NMME can successfully anticipate such conditions, and provides a very useful guide to anticipating East African hydrological extremes. By focusing on the models’ large scale representation of ENSO, AdReg can be used to produce skillful automated forecasts, in early October, for both the short and long rains. For the short rains, our level of confidence might be quite high, since we have already begun the season, rainfall to date is slow to start, we can see evidence of a La Niña-like transition in the observed circulation, and near-term weather predictions tend to be pessimistic, at least over large parts of East Africa. If October is dry, the overall season will likely be poor (Figure 11). Our level of confidence in the MAM 2018 outlook is lower, given the large span of time, and the good possibility that La Niña-like conditions may fade. On the other hand, our forecast of -0.6Z is also quite similar to the ‘New Normal’ implied in the long term decline of the EEA long rains (Figure 2). Given the combination of current La Niña-like conditions, the pessimistic but skillful AdReg results shown here, and general tendency for EEA MAM rains to be below normal in the absence of strong forcing from El Niño or Indian Ocean Dipole-like conditions, assumptions of below normal rains for both the short and long rains may be warranted. The MAM outlook value would hover at the edge of the normal and below normal tercile boundary.

Figure 9. Correlation between observed EEA MAM rains and NMME OND precipitation from the preceding October.
Figure 9. Correlation between observed EEA MAM rains and NMME OND precipitation from the preceding October.
Figure 10. AdReg forecasts for EEA MAM rains. Cross-validated results for 1982-2016, and the 2017 forecast. Bars denote 80% confidence intervals.
Figure 10. AdReg forecasts for EEA MAM rains. Cross-validated results for 1982-2016, and the 2017 forecast. Bars denote 80% confidence intervals.
Figure 11. Correlation between NMME OND Precipitation and PC1 for MAM EEA forecast.
Figure 11. Correlation between NMME OND Precipitation and PC1 for MAM EEA forecast.
Figure 12. EEA October and OND rainfall, 1981-2016.
Figure 12. EEA October and OND rainfall, 1981-2016.
Prior research

While not detailed here, the reader should be aware of numerous studies supported by USAID, the USGS and NASA, to inform decision making in context like this. This work seeks to better understand and anticipate EEA droughts.*

Williams P. and Funk C. (2010) A Westward Extension of the Tropical Pacific Warm Pool Leads to March through June Drying in Kenya and Ethiopia, USGS Openfile Report 1199,

Williams P. and C. Funk (2011) A westward extension of the warm pool leads to a westward extension of the Walker circulation, drying eastern Africa, Climate Dynamics, V37.11-12, p. 2417-2435.

Liebmann, B., Bladé I., Kiladis G. N. , Carvalho L. M. V., Senay, G., Allured D., Leroux S., Funk C (2011) African Precipitation Seasonality Based on Daily Satellite Data from 1996-2009, J. of Climate, Journal of Climate 25, no. 12 (2012): 4304-4322.

Hoell, A. and C. Funk (2013): The ENSO-related West Pacific Sea Surface Temperature Gradient, Journal of Climate. J. Climate, 26, 9545-9562. doi:

Hoell A. and C. Funk (2013) Indo-Pacific Sea Surface Temperature Influences on Failed Consecutive Rainy Seasons over Eastern Africa, Climate Dynamics, DOI: 10.1007/s00382-013-1991-6. p1-16.

Hoell, A., Funk, C., & Barlow, M. (2014). La Niña diversity and Northwest Indian Ocean Rim teleconnections. Climate Dynamics, 1-18.

Liebmann, B., Hoerling M., Funk C., Dole, R. M., Allured A., Pegion, P., Blade I., Eischeid, J.K, (2014) Understanding Eastern Africa Rainfall Variability and Change, Journal of Climate.

Funk, C., Hoell, A., Shukla, S., Bladé, I., Liebmann, B., Roberts, J. B., and Husak, G. (2014). Predicting East African spring droughts using Pacific and Indian Ocean sea surface temperature indices. Hydrology and Earth System Sciences Discussions, 11(3), 3111-3136.

Shukla, S., McNally, A., Husak, G., & Funk, C. (2014). A seasonal agricultural drought forecast system for food-insecure regions of East Africa. Hydrology and Earth System Sciences Discussions, 11(3), 3049-3081.

Shukla S., Funk C. and Hoell A. (2014) Using constructed analogs to improve the skill of March-April-May precipitation forecasts in equatorial East Africa , Env. Res. Letters, Environmental Research Letters 9.9 (2014): 094009.

Funk, C. and Hoell A. (2015) The leading mode of observed and CMIP5 ENSO-residual sea surface temperatures and associated changes in Indo-Pacific climate, J. Climate, 28, 4309-4329,

Shukla S., Roberts, J., Hoell A., Funk C., Robertson F. and Kirtmann, B. (2016) Assessing North American Multimodel Ensemble (NMME) Seasonal Forecasts, Climate Dynamics. 1-17.

Funk, C and Hoell A. (2017) Recent Climate Extremes Associated with the West Pacific Warming Mode, AGU Monograph-Climate Extremes: Patterns and Mechanisms. Ed. by Simon Wang,  Jin-Ho Yoon, Chris Funk and Robert Gillies, published by Wiley Press 226 (2017): 165.

Liebmann B., Bladé I., Funk C., Allured D., Hoerling, M., Hoell, A., Peterson, P., Thiaw, M.T. (2017) Climatology and Interannual Variability of Boreal Spring Wet Season  Precipitation in the Eastern Horn of Africa and Implications for its Recent Decline,J. Climate, Journal of Climate 30.10 (2017): 3867-3886.

Hoell A, Hoerling M, Eischeid J, Quan W-X, Liebmann B (2017) Reconciling theories for human and natural attribution of recent East Africa drying. J Clim 30:1939-1957