Food security in Southern Madagascar informed by satellite-based remote sensing data

Abstract Recurrent droughts and food shortages in southern Madagascar underscore the urgent need for real-time food security monitoring. Since 2018, monthly surveys of 602 households have documented dietary coping strategies in response to food shocks. Here, we evaluate whether remotely sensed Normalized Difference Vegetation Index (NDVI) observations over crop and pasturelands can track temporal variation in household food consumption. While standard food security indices show no discernible relationship with NDVI, a leading mode of dietary variability—termed Dietary Component 1 (DC1)—exhibits a strong correspondence. DC1 captures an anticorrelation between the consumption of maize and sorghum during food-secure periods and red cactus and lower-preference roots and tubers, such as cassava, during food-insecure periods. When aggregated to the commune level, DC1 explains an average of 43% of the variance in monthly dietary patterns. Linear models using commune-averaged NDVI explain 32–58% of DC1 variability, and models including monthly fixed effects explain 38–70%, with NDVI remaining a highly significant predictor throughout. Although not a replacement for on-the-ground data collection, these results demonstrate that satellite-derived NDVI can reliably capture key dietary shifts and serve as a valuable component of food security monitoring systems in southern Madagascar.