Author(s)

EMMANUEL IGWE, NAZIH YACER REBOUH, DARIA OLEGOVNA KAPRALOVA, SHAIBU OCHOCHE, JEREMIAH AKOMAYE IKWEN, ONYINYECHI GRACE OPARA, NIAMBE OBED KOHOL, DOOSHIMA RITA DUGERI, MOHAMED ALPHA JALLOH, SAVIOUR DOTSEY KWAKU

Abstract

Sub-Saharan Africa (SSA) is facing escalating food insecurity due to climate change, which intensifies droughts, floods and desertification, degrading cropland productivity and disrupting smallholder livelihoods. Although climate-smart agriculture (CSA) frameworks have proliferated, significant adaptation deficits persist, driven by top-down interventions, weak institutional capacity and limited integration of traditional knowledge (TK), which is essential for local resilience. This study employs an integrated, geo-spatial and AI-assisted framework, combined with locally led adaptation (LLA) methodologies, to assess climate-driven cropland vulnerability across SSA. We analysed multi-source satellite data (Landsat, Sentinel and MODIS) from 2000 to 2024 using machine learning classifiers (Random Forest and XGBoost) to map land-use change, climatic exposure and soil degradation. A weighted index integrating exposure, sensitivity and adaptive capacity quantified vulnerability, while Getis-Ord Gi* statistics identified hotspot clusters. AI-driven predictive models (LSTM, CNN-RF hybrids, and gradient boosting) forecast future cropland vulnerability under the SSP2-4.5 and SSP5-8.5 scenarios, indicating a persistent intensification of drought and flood hotspots, particularly in West and East Africa. Mixed-methods assessments – including focus groups, surveys and participatory workshops – identified barriers to CSA adoption and documented the efficacy of TK practices, such as drought-resilient cropping calendars and community-based water harvesting. The results demonstrate that integrating TK with AI-informed CSA interventions enhances adaptation by improving the accuracy of climate risk detection, aligning strategies with local priorities and strengthening early warning systems. Empirical data from 2017 to 2024 reveal a doubling of flood magnitude, with affected populations increasing from 3.0 to 8.5 million and associated economic losses arising from USD 1.5 billion to USD 15 billion, highlighting severe adaptation gaps. Evaluations of the joint CSA+TK pilots confirmed improvements in soil health, yield stability and household resilience. This study concludes that a synergistic approach – combining AI-enabled hotspot mapping, LLA and TK – provides a scalable pathway for designing context-specific, equitable, and climate-resilient agricultural strategies across SSA.

Keywords

water risk, climate resilience, climate-smart agriculture, traditional knowledge, adaptation deficit