Agricultural drought is characterized by prolonged soil moisture deficits caused by low precipitation, high evapotranspiration, and temperature extremes. Unlike meteorological drought, which is solely defined by precipitation anomalies, agricultural drought directly impacts vegetation health, crop productivity, and water resources. However, agricultural drought monitoring is still challenging due to imprecise index and low-resolution remote sensing observations, particularly in Africa. This paper proposes a novel Self-Organizing Agricultural Drought Index (SOADI) for agricultural drought monitoring in Africa through a machine learning using the Self-Organizing Map (SOM) technique. Drought is detected using the integrated multiple remote sensing indices NDVI, EVI, NDII, SAVI, and LST for a more comprehensive assessment. SOADI was evaluated across two contrasting agro-climatic zones: El Faiyum, Egypt (hyper-arid), and Jowhar, Somalia (semi-arid), from 2000 to 2023. The results indicated a high frequency of drought events during this period. SOADI correlated strongly with VHI (Egypt: r = 0.83-0.93; Somalia: r = 0.81-0.93) and SPI (Egypt: r = 0.75-0.89; Somalia: r = 0.71-0.81), demonstrating robustness across climates. In addition, SOADI successfully detected notable historical drought events, such as the severe droughts of Somalia in 2011 and 2022. SOADI represents a promising tool for accurate drought monitoring with offering deeper insight into spatiotemporal drought variability within agricultural regions and significantly improving upon traditional indices for more effective drought management. (c) 2025 COSPAR. Published by Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.