A hybrid machine-learning and process-based regional crop modelling framework to improve rice production strategies under data constraints

• Coupled PSO and DSSAT to model Lao rice systems under severe data scarcity. • Simulated rice yield and CH 4 , N 2 O, and NH 3 emission in Laos. • Identified a national economic N guidance band of 25–30 kg N ha −1 . • Province-specific N tuning improved profit while maintaining high NUE. Rice production in the Lao People’s Democratic Republic (Lao PDR) is fundamental to national food security, yet its productivity and environmental impacts remain poorly quantified due to limited monitoring and data availability. To address this challenge, we developed a spatially explicit and cost-effective regional rice crop modelling framework by coupling the Decision Support System for Agro-technology Transfer (DSSAT) model with a particle swarm optimization (PSO) algorithm. This hybrid framework was applied to simulate rice growth and associated environmental impacts across Laos from 2012 to 2022. Using available provincial yield statistics for model calibration, we quantified rice yields and greenhouse gas emissions in wet-season rain-fed rice systems. At the province scale, fertilizer response curves show that high fertilizer prices relative to rice prices substantially reduce the economically optimum N application rate. We identified a national guidance range of 25–30 kg N ha −1 with province-specific adjustments, capable of improving farm profitability (70–110%) while sustaining high N use efficiency (NUE) (90–110 kg kg −1 ). Our results highlight the potential for Laos to advance environmentally sustainable (green) rice production. The proposed PSO + DSSAT framework offers a transferable tool for supporting sustainable rice management in regions with limited data coverage.