Combined agricultural and metrological seasonal drought assessment using geospatial technology in North Wollo Zone, Ethiopia

Abstract

Drought remains a persistent challenge affecting agricultural productivity and water resources globally. In Ethiopia, the North Wollo Zone known for frequent droughts and chronic food insecurity was purposively selected as the study area from the 11 Zones of the Amhara Region. This study presents a short-term seasonal drought assessment for the 2024 cropping season by integrating meteorological data with agricultural indicators reflecting crop health. A combination of multi-source remote sensing indices and explanatory variables was analysed using non-parametric machine learning techniques. Crop production data were collected from 104 kebeles using purposive sampling, enabling validation of satellite-derived data and supporting a comprehensive analysis through a mixed-method explanatory sequential research design. The findings show that the most drought-affected districts are Bugna, Mekete, and Dawunt. Results indicated that there are the strongest drought indicators like NDVI, NDRE, SAVI, and LST. Rainfall and temperature variability significantly influence drought severity and crop yield in the area. In addition, among the machine learning models tested Random Forest Regression, Linear Regression, and Support Vector Machines (SVM) Random Forest performed best, achieving an R 2 of 0.76 and a mean absolute error (MAE) of 819.09. Variables derived from satellite imagery, particularly NDVI, NDRE, SAVI, and LST, proved most effective in predicting drought conditions. The study concludes that integrating remote sensing, meteorological data, and ML models enhances seasonal drought monitoring and prediction accuracy. It recommends promoting rainwater harvesting, timely planting, and climate-resilient practices to mitigate drought impacts. The findings should be interpreted within the context of short-term drought monitoring, and long-term (≥ 30 years) climatic datasets would be required for robust climate trend analysis.