Cross-border cropland indicators and field-scale rice system mapping from multi-sensor Earth observation in the Senegal River Valley

Kurzfassung

Food security in West Africa is increasingly challenged by rapid population growth, climate variability, and structural constraints that limit agricultural productivity. In Senegal and Mauritania, irrigated rice systems in the Senegal River Valley (SRV) are central to food-security strategies, yet cultivated area, cropping intensity, and management practices vary strongly across space and time. Effective planning therefore requires timely, spatially explicit information on where land is actively cultivated, how stable cultivation is across years, and the cropping intensity if rice systems — information that remains scarce in routine reporting. We present a multi-sensor Earth observation framework that combines (i) annual cropland indicators for the entire SRV and (ii) field-scale mapping of rice cropping systems. For cropland indicators, we apply a Random Forest classifier to yearly Sentinel-1 and Sentinel-2 time series from 2016–2024, trained with samples derived from high-resolution imagery. From annual maps we derive decision-relevant indicators including cropped area, first-appearance, cumulative extent, interannual stability statistics. Results show contrasting trajectories: Senegal remains relatively stable at ~120,000 ha (consistent with 2022/23 official figures), while Mauritania increases from ~60,000 ha (2016) to >100,000 ha (2024), with larger and more intensive used field structures. For rice systems, we complement these indicators with high-resolution field delineation and cropping-pattern classification across 2017–2025, integrating Sentinel-1 and Sentinel-2 phenological features and machine-learning classification. The resulting products reveal spatial gradients in cropping intensity and management regimes along the valley. Together, the framework supports evidence-based agricultural planning and monitoring for public authorities and basin managers and provides actionable information for financial institutions (credit and insurance) by distinguishing stable, expanding, and intermittently cultivated areas.