Harnessing Earth Observation and Process-Based Models for Crop Yield Estimation in Ethiopia
Accurate and timely crop yield estimation is a cornerstone of food security planning, particularly in regions heavily dependent on agriculture. In Ethiopia, FAO has been at the forefront of integrating Earth Observation (EO) data and process-based models to enhance national agricultural statistics. Through collaboration with the Ethiopian Statistics Service (ESS) and key stakeholders, FAO has spearheaded a capacity-building initiative aimed at equipping national experts with advanced tools for crop yield prediction. This initiative, which is part of the EOSTAT project, focuses on leveraging the Decision Support System for Agrotechnology Transfer (DSSAT) to simulate crop growth at a regional scale.
Why Process-Based Models?
Traditional yield estimation methods often rely on statistical approaches that correlate historical yield data with climatic and environmental variables. While effective in stable conditions, these methods struggle with accuracy in the face of climate variability. Process-based models, such as DSSAT, offer a robust alternative by simulating the physiological processes of crops, including photosynthesis, nutrient uptake, and water balance. By integrating soil properties, climate data, and management practices, these models enable:
Prediction of yield under different scenarios (e.g., varying rainfall or fertilizer application levels).
Identification of limiting factors affecting crop growth.
Simulation of interventions for precision agriculture and improved decision-making.
Building National Capacity: A Hands-On Training Approach
To ensure sustainability and national ownership of these advanced methodologies, FAO organized a three-day intensive hands-on training on gridded DSSAT simulations in February 2025 in Adama, Ethiopia. The training, led by FAO experts and consultants, brought together participants from ESS and other key stakeholders involved in agricultural monitoring. The agenda covered:
Introduction to DSSAT and its applications in regional-scale crop yield estimation.
Installation and setup of DSSAT, R, and Google Earth Engine (GEE) for data preprocessing.
Downloading and preparing soil and climate datasets using EO data sources such as OpenLandMap and NASA POWER.
Simulation runs and result interpretation, allowing participants to analyze wheat yield across Ethiopia’s Arsi Zone, a key wheat-producing region.
Troubleshooting and discussion sessions, ensuring that participants gained practical experience in resolving technical challenges.
This training was not merely theoretical but deeply practical, enabling national experts to independently conduct DSSAT-based yield simulations for future agricultural monitoring.
Case Study: Yield Estimation in Ethiopia’s Arsi Zone
The Arsi Zone, located in the Oromia region, was selected as the pilot area due to its significant contribution to Ethiopia’s wheat production. The FAO-supported study utilized DSSAT to estimate wheat yield by integrating:
Climate data (2018-2024) from ERA5 and processed via GEE.
Soil properties extracted from OpenLandMap.
Crop management practices provided by ESS.
Wheat variety parameters calibrated using regional datasets.
Simulation results demonstrated how DSSAT can provide valuable insights into yield gaps under different management practices. The study also highlighted the potential of EO-based approaches, such as the Green Chlorophyll Vegetation Index (GCVI), to refine nitrogen input estimations, making yield predictions more reflective of actual field conditions.
Figure 1: Annual wheat yield (kg/ha) in Arsi Zone from 2019 to 2024 under current farming practices
Figure 2: Annual leaching of Nitrogen (kg/ha) in Arsi Zone from 2019 to 2024 under current farming practices
Beyond Training: Strengthening Institutional Capacity
In addition to hands-on training, FAO has implemented a broader strategy to institutionalize the use of EO and process-based models within ESS:
- Webinars and knowledge-sharing sessions to introduce key concepts and methodologies.
- Biweekly follow-up meetings to address challenges and ensure continuous learning.
- Development of standardized workflows and R scripts to facilitate reproducibility in future analyses.
These efforts are positioning ESS as a leader in advanced agricultural statistics, paving the way for Ethiopia to scale up the use of EO and crop modeling for national yield forecasting.
A Call for Sustained Investment
The success of this initiative underscores the need for sustained investment in EO-based capacity building. As Ethiopia seeks to modernize its agricultural statistics, further support from development partners and donors will be critical to:
- Expand DSSAT-based simulations to other key agricultural regions.
- Enhance integration with remote sensing data for real-time yield monitoring.
- Develop operational frameworks to embed these methodologies into official statistics.
The EOSTAT project provides a model for how EO and process-based models can transform agricultural monitoring in developing countries. With continued support, Ethiopia can build a resilient, data-driven agricultural sector capable of addressing future food security challenges.