PRISM: A methodological framework for reproducible data-to-model translation in spatial crop modeling

Process-based crop models are increasingly applied to forecast yields and assess climate risks in rainfed smallholder agriculture across sub-Saharan Africa. However, the prerequisite step of transforming heterogeneous agricultural data into standardized, model-ready inputs remains a major challenge for reproducible and comparable spatial crop modeling. Here we present PRISM (Platform-Ready Inputs for Spatial Modeling), a framework that formalizes data retrieval, quality control, gap-filling, spatial harmonization, and format conversion to produce standardized input packages compatible with multiple crop modeling frameworks. PRISM separates agronomic specifications from model-specific settings, with provenance records documenting preparation decisions. Case studies across four crop modeling frameworks in Mali, Niger, Nigeria, and Senegal demonstrated that all generated packages loaded without manual editing and that simulated outputs fell within published benchmark ranges. By automating the data-to-model translation process, PRISM enables systematic data preparation for spatial crop modeling across platforms in data-scarce regions, facilitating multi-model comparison studies for yield estimation, climate impact assessment, and agricultural planning.