Rainfed agriculture in Ethiopia: climate vulnerability and adaptation

Dates: 2019-2023
Funding: ETH4D Doctoral Scholarship
PhD Student:
Mosisa Wakjira
Dissertation: research collection (2024)
Principal Investigators: Peter Molnar, Johan Six
Partner: external page Nadav Peleg

Agriculture in countries such as Ethiopia is predominantly a smallholder rainfed system that supports the food and incomes of the rural community and significantly contributes to the national economy. The productivity of the system remains among the lowest in the world due to multiple limiting factors, primarily policy and institutional capacity, agricultural inputs and technology, and agro-environmental challenges. This project aims to assess the climatic aspects of agro-environmental challenges and opportunities in rainfed agriculture (RFA), and their implications for future climate adaptation and resilience-building plans and decision-making in Ethiopia. To this end, three major topics have been focused on: a) the temporal attributes of rainfall and their influences on rainfed crop production, b) Green water availability and water-limited attainable yields under current and future climate conditions, and c) the current status and future changes in cropland suitability across the RFA region of Ethiopia. We combine state-of-the-art climate, soil, and crop yield datasets through statistical analyses and modelling approaches. The outcomes of this project are vital for informed planning, policy design and decision-making for in the nationally determined efforts of ensuring food security, poverty reduction, and building climate resilience, through sustainable intensification of crop production and climate risk management.

Our main findings are reported below.

RainFedAg-Fig2

High-quality gridded climate data is vital, among many other applications, for a better understanding of crop-climate relationships in a spatial setting. This is a major gap in data-scarce regions including the RFA region of Ethiopia. In the framework of this project, we have produced maximum and minimum 2-meter air temperatures at an enhanced quality and spatial resolution for Ethiopia. We utilized the statistical information contained in the in-situ observation (IOBS) network of 146 stations across the country to adjust the statistical biases in the ERA5-Land 2-meter air temperature dataset over the Ethiopian domain. We applied a hybrid spatial interpolation approach that combines linear regression and inverse generalized distance weighting (IGDW) interpolation techniques, along with quantile mapping bias-correction, to produce Bias-Corrected and Downscaled ERA5-Land (BCE5) datasets at a 0.05° x 0.05° grid resolution with significantly improved accuracy. The BCE5 datasets cover the period 1981-2010, which is considered the current climate normal in several climate change studies. It is compatible with widely used high-resolution precipitation datasets like CHIRPS and is suitable for a wide range of environmental applications, such as hydrology, agriculture, and ecology. The BCE5 dataset has been stored in a NetCDF file in an open-access format at the ETH Zurich research collection and can be accessed through a external page permanent link.

Wakjira, M. T., Peleg, N., Burlando,P., and Molnar, P. (2023), Gridded daily 2-m air temperature dataset for Ethiopia derived by debiasing and downscaling ERA5-Land for the period 1981-2010, Data in Brief, 46, 108844, external page https://doi.org/10.1016/j.dib.2022.108844.

RainFedAg-Fig3

In this climatic analysis, we looked into the variability and change in four temporal rainfall attributes, namely seasonality, the dates of onset, cessation, and duration of the rainy season – across the RFA areas of Ethiopia based on CHIRPS rainfall dataset. We investigated the responses of cereal crop production to the dynamics of the rainfall seasonality and timing. We employed an entropy-based numerical approach to quantify the rainfall seasonality, and cumulative rainfall anomaly curves to define the onset, cessation, and duration of the rainy season. We then performed correlation and regression analyses to assess the influences of changes in these attributes on crop production. We show that rainfall seasonality and timing have significant influences on crop production in the rainfed agricultural region of Ethiopia. However, these impacts vary depending on the rainfall pattern. For example, in regions with unimodal rainfall patterns, crop production is highly determined by the rainy season onset. On average, a 5-day delay in the onset of the rainy season results is an estimated 1.5% reduction in cereal production. This implies that farmers in these climatic regions should choose the right planting dates to reduce crop yield gaps associated with crop water stress. In regions with less seasonal and erratic rainfall regimes like in the eastern and southeastern part of the RFA region, crop production is largely influenced by rainfall seasonality (amount and temporal distribution). In such regions, on-farm water management that maximize infiltration and minimize evaporative losses should be prioritized.

Wakjira, M., Peleg, N., Anghileri, D., Molnar, D., Alamirew, T., Six, J., and Molnar, P. (2021), Rainfall seasonality and timing: implications for cereal crop production in Ethiopia, Agric. For. Meteorol., 310, 108633, external page https://doi.org/10.1016/j.agrformet.2021.108633.

RainFedAg-Fig4

As a major part of data preparation for the climate change impact assessments in this project, we have downscaled future precipitation and temperature projections from multiple GCMs (25 CMIP6 models for precipitation and 21 for temperature) over Ethiopia using the change factor (delta) method. Three Shared Socioeconomic Pathways (SSPs) were considered: the sustainable development (low emission) pathway (SSP1-2.6), the middle of the road (medium emission) pathway (SSP2-4.5), and the fossil fuel energy (high emission) pathway (SSP5-8.5). The period 1981-2010 was used as a reference climate normal, with CHIRPS and BCE5 datasets serving as reference rainfall and temperature climates to produce the downscaled future changes during three future periods: 2020-2049 (2030s), 2045-2074 (2060s), and 2070-2099 (2080s). The multi-model median projection indicates that the two growing seasons -- meher (May-September) and belg (February-May) are likely to become warmer and wetter in the respective producing regions during each season in the future and under all scenarios. The highest increase in precipitation is evident under the high-emission scenario and towards the end of the century. Dry seasons are expected to remain dry or become even drier.

RainFedAg-Fig5

The intensifying atmospheric evaporative demand driven by the increases in temperature and increasing rainfall uncertainty are expected to affect green water (soil moisture from rainfall input) and its availability for rainfed crop production and broader agroecosystem productivity. We examined the spatiotemporal changes in green water availability in various climatic zones of the RFA region under the current climate and how this change under the warming and wetting future climate. To this end, we built a Climate-Hydrological-Crop yield (CHC) modelling framework to assess the effects of climatic and hydrological conditions on crop yield at climatological timescales. The climate module computes the reference evapotranspiration accounting for several climatic factors using the FAO penman-Monteith algorithm, the hydrology module simulates root zone hydrological balance based on curve number-based bucket soil water balance model, and the crop yield module estimates water-limited attainable yield (AY) as the percentage of energy limited potential yield based on the FAO water production function. We show that green water a highly reliable resource for agricultural evapotranspiration during the meher (May–September) season across most of the RFA region, AY exceeding 80% of potential yields in over half the area. From 1981-2010, median AY ranged from 46% in dry semi-arid areas to 93% in humid areas during meher. Variations in AY are due to rainfall distribution, soil properties, and surface conditions. We identified huge crop yield gaps through comparison of AY with observed yields, emphasizing the need for comprehensive agro-environmental resource management. Under future climates, AY is expected to experience slight or no change especially in humid and sub-humid areas for meher, and up to a 20% increase in belg (February - May) towards the end of the century.

Wakjira, M. T., Peleg, N., Six, J., and Molnar, P. (2024), Green water availability and water-limited crop yields under a changing climate in Ethiopia, Hydrol. Earth Syst. Sci. Discuss. [preprint], external page https://doi.org/10.5194/hess-2024-37, in review.  

RainFedAg-Fig6

In this analysis, we assessed the potential crop cropland suitability and investigated their climate-driven changes in the future. We used a numerical modelling approach to construct cropland suitability indices from crop yield data combined with climatic and soil factors for four major cereal crops that are grown in Ethiopia. We forced the suitability models with the present and future climates at every grid to map the potential suitability and analyze the changes. The results reveal significant agroecological shifts from lowlands toward highlands in the future, eventually leading to complete suitability losses in the low-altitude regions. The changes are massive, particularly for crops like tef and wheat, with more suitability losses than gains. The projections suggest that the areas suitable for teff and wheat will decrease, for example, by 18% and 27%, respectively, under SSP5-8.5 by the end of the century. Maize and sorghum croplands will experience altitudinal shifts with relatively less areal losses compared to teff and wheat croplands.

Paper under review in Agricultural and Forest Meteorology.
 

JavaScript has been disabled in your browser