East Africa – Climate Projected 2000-2050

Climate and Crop Modeling Methodology

Four AR4 general circulation models (GCMs) were downscaled to a high resolution, 6 km using SRES scenario A1B (moderately aggressive growth) (Moore at al. 2011). The models were chosen because they represent weather extremes (generally CCSM being very wet and HadCEM being very dry), and two widely used models that are in the middle (CSIRO and ECHAM5). The downscaling was conducting using thin plate smoothing splines via the ANUSPLIN V4.3 software (Hutchinson 2002) using latitude and longitude as independent variables for rainfall, and latitude, longitude, and elevation for temperature. Data were downscaled to the 6km surface. We splined 30-year averages for each month for each GCM dataset. Two datasets were prepared per model, one centering on 2000 and one on 2050. The difference between 2000 and 2050 in temperature, rainfall and other variables represent the perturbed climatology due to enhanced greenhouse gases.

To estimate the impact of climate change on crop growth and productivity, we use a deterministic, process-based simulation model, the CERES-Maize crop model (DSSAT; ICASA 2007). WorldClim (Hijmans et al. 2005) was used to represent current climatology, and perturbed climate data from the GCMs described above were superimposed on WorldClim to represent a future climatology. Soils from FAO soils map calibrated with soil profiles from WISE database (ISRIC) were used in the crop model, and local crop varieties were chosen and validated against observed data. In East Africa, two maize varieties were modeled–Katumani Composite, a  a short season, drought-resistent variety  developed by the Kenya Agricultural Research Institute (KARI) for medium to low potential zones, and a hypothetical hybrid similar to hybrids developed by KARI for higher potential zones. Agronomic practices were assumed to be similar to farmer practices identified  in surveys. Yields under low and moderate nitrogen fertilizer levels were compared.

Results: Future Climate Projections

The maps below in Figures 5 show projected changes in temperature between 2000 and 2050 for the four GCMs. The maps illustrate  widespread increases in temperature, a continuation of current trends. The difference between models is partly related to their relative cloudiness (CCSM being the most cloudy). The rise in maximum and minimum temperatures affect plant development and growth in various ways including phenology and demand for water. Maize in particular is already at the limit of its heat tolerance in many areas in East Africa, and the warming temperatures alone would lead to declining yields.


Figure 5. Changes in temperatures between 2000 and 2050 for four GCMs. Source: Moore et al. 2011; Alagarswamy et al. 2015 and 2013.

The next set of maps, Figure 6, show projected change from 2000 to 2050 of annual rainfall. The rainfall of 2000 was compared to projected 2050 rainfall, and the change was mapped onto our best estimate of current climate (WorldClim). The small map on the right shows current annual rainfall patterns. Rainfall varies from desert-like conditions in northeastern Kenya and Somalia, to rainforest conditions in eastern Congo.

The four GCMs project different changes in rainfall, with CCSM having the wettest projection particularly across Uganda and southern Sudan, with over 200 mm/year increases. ECHam is the next wettest projection. HadCEM also projects some increases in precipitation, in central and northern Kenya, but much of the domain shows declining precipitation, especially along the Indian Ocean coast and, unlike any of the other four models, declines across the southern part of the domain). CSIRO is the driest of the models for East Africa.

In general, there does not seem to be a “consensus” of how rainfall may change. CCSM and ECHam show similar geographical patterns, if differences in the degree of projected increases. They show the western and northern areas of the domain becoming wetter, with little change or slight declines elsewhere. CISRO and HadCEM both show large areas receiving less rainfall. Three of the four models agree on areas of declining precipitation: 1) the Indian Ocean coastal zone in Somalia and Kenya, and 2) the highlands in Rwanda, Burundi, Central Kenya, and Southwestern and northern Tanzania.

The variations between the models highlights the uncertainty related to projecting future climate change, especially at the local to regional level. Although temperatures are expected to continue to rise at the same rate or faster than they have been during the past few decade, changes in precipitation are much more difficult to forecast with certainty. The impact of local scale factors  complicates the effect of global climate change.


Figure 6. Projected Precipitation (2000 to 2050). Source: Moore et al. 2011; Alagarswamy et al. 2015 and 2013.

The East African region, in general, shows large geographical differences in projected rainfall changes. The downscaled models project that several Highland zones will be especially affected by declining precipitation. Currently, these are highly productive agricultural areas producing high value commodities such as tea, coffee, wheat, dairy and horticultural crops. With rapidly warming temperatures and declining in precipitation, their high productivity is threatened.

It should be noted that the agreement of GCMs and observations for East Africa has generally been found to be low compared to other regions of the world (Williams and Funk 2011) and may be symptomatic of poorly resolved (or unresolved) ocean circulations. A second reminder about GCMs is that the maps illustrate long term trends that mask the expected increase in irregular and unpredictable rainfall. High variability between years, and from season to season have an enormous impact on farming. Indeed farmers in the region report that this is already occurring.

Results: Impact of Climate Change on Crop Yield

In East Africa where rainfall is a limiting factor across much of the region, even no change in rainfall amounts is expected to lead to more water stress and more rapid plant phenology because of the warming temperatures. As seen in Figures 7 below, this is the case across much of Tanzania and Uganda. Where yields were already low due to limited rainfall, such as in central Tanzania, maize yields may fall below economic viability. In Uganda, the projected declining yields may lead to maize being somewhat less profitable, but it could still be preferred depending on alternative crops.


Figure 7. Simulated change in Katumani composite maize yield (35 kg N) between 2000 and 2050. Source: Moore et al. 2011; Alagarswamy et al. 2015 and 2013.

The impact of climate change on both Katumani composite and the hybrid cultivars is generally a decline in yield across most of the region, except in the Highland zones where precipitation will still be sufficient and the warming temperatures will be beneficial to maize. The increase in yield in northern Kenya and Uganda is due to several GCMs projecting an increase in precipitation there, but most of those areas will still be too dry to support maize.

This high resolution analysis revealed unexpected localized effects of climate change on yields. Local topography and soils play an important role determining how climate change is expected to affect crop growth and development. Each country has zones that will need individualized responses to climate change, from fertilizer programs, to development of heat resistant, longer season maize cultivars, to consideration of alternate crops.

Adaptation will require a combination of practices depending on location – reducing water stress (e.g., irrigation or other water management practices) in some areas, selecting drought resistant varieties in some areas, focusing on highly productive varieties in the maize producing zones, and applying sufficient fertilizer almost everywhere.

Summary of the Impact of Climate Change on Crops

  1. Precipitation extremes and variability are already increasing affecting crop production, and are expected to increase.
  2. Warmer temperatures reduce plant productivity across wide areas.
  3. Warmer temperature, combined with low precipitation results in significant yield decline.
  4. Warmer temperatures accelerate plant development (phenology) and reduce growth duration and yield.
  5. The influence of future climate on maize yield is highly variable over space.
  6. Adaptive strategies and technological innovations for adaptation need to be geographically and socially specific.


Alagarswamy, Gopal, Jennifer Olson, Jeff Andresen, Nathan Moore, Philip Thornton, Pius Yanda, Joseph Maitima, Jenni Gronseth, David Campbell (2015). The Highly Variable Response of Maize Yield to Climate Change across East Africa. DOI: 10.13140/RG.2.1.4240.5921  Poster presented at the: 5th AgMIP Global Workshop, Gainesville FL, February 25, 2015. http://www.agmip.org/wp-content/uploads/2015/03/Alagarswamy-Gopal.pdf

Alagarswamy, G. JA Andresen, JM Olson, PK Thornton, NJ. Moore. 2013. Climate Change Impacts on Maize Production in East Africa. In proceeding of: The Association of American Geographers Annual Meeting.

Moore, N., Alagarswamy, G., Pijanowski, B., Thornton, P., Lofgren, B., Olson, J., Andresen, J., Yanda, P. and Qi, J. 2011. East African food security as influenced by future climate change and land use change at local to regional scales. Climatic Change June 10, 2011. Vol. 107. DOI: 10.1007/s10584-011-0116.