Climate change and crop production in North Africa: Insights from machine learning models.

Summary

This study applies multiple machine learning models (MLR, Random Forest, XGBoost) to examine the relationship between climate variables and crop yields across North Africa (Morocco, Algeria, Tunisia, Libya, Egypt). Using ERA5 climate reanalysis and FAO yield data, the authors project future crop yields under two emissions scenarios (SSP2-4.5, SSP5-8.5) to 2100, finding that rising temperatures and declining precipitation will reduce yields for several crops, though maize and sorghum demonstrate greater resilience under irrigation. The findings contribute evidence to support climate adaptation planning and identification of crop-switching strategies in semi-arid regions.

Regional applicability

The study focuses on North African semi-arid conditions, which differ significantly from United Kingdom climate and farming systems. However, the methodological approach of combining ERA5 reanalysis with FAO yield data and machine learning models could be adapted to UK contexts, and the findings on crop resilience and adaptation strategies may inform broader European agricultural policy discussions on climate adaptation, though direct yield projections would not transfer to temperate climates.

Key measures

Crop yield (FAO data), temperature, precipitation (ERA5 reanalysis), projected yields under SSP2-4.5 and SSP5-8.5 CMIP6 scenarios across three time horizons (2015-2050, 2051-2080, 2081-2100)

Outcomes reported

The study projected future crop yields under climate change scenarios across North Africa using machine learning models, identifying which crops are most vulnerable and which show greater resilience to temperature and precipitation changes. Results reveal projected yield reductions for several crops, with maize and sorghum showing greater resilience, particularly under irrigated conditions.

Topic tags