Evaluating the Resilience of Cold-Region Rice Varieties to Climate Variability and Change Using the BDH-Rice Model: A Case Study in the Sanjiang Plain

The Sanjiang Plain is a key cold-region rice production area in Northeast China, where climate variability and warming may alter the adaptability and yield stability of rice varieties. Identifying varieties with both high yield potential and stable performance under current and future climates is therefore essential for climate-resilient rice production in this region. This study developed and evaluated a site-specific BDH-Rice-based framework for screening cold-region japonica rice varieties at the 856 Farm, a single representative production site in Heilongjiang Province. Multi-year field observations of 37 rice varieties from 2014 to 2023, together with local meteorological, soil, and management data, were used to localize variety-specific parameters. Seven parameters governing phenological development and yield formation were calibrated through a stepwise procedure. The localized model accurately simulated rice phenology, with RMSEs of 2.46 and 2.96 days for flowering and maturity, respectively (NRMSE < 3%), and reproduced grain yield with reasonable accuracy (NRMSE = 10.43%; index of agreement, D = 0.78). Simulated and observed yield rankings were significantly correlated in four of the seven testable year–dataset combinations (Spearman’s ρ = 0.782–0.933; Kendall’s τb = 0.600–0.833, p < 0.05). The high stability coefficient (HSC), which integrates yield level and interannual stability, was evaluated against field observations before its scenario application: simulated and observed HSC rankings were significantly correlated (ρ = 0.696, p < 0.001), indicating overall consistency in the stability ordering of the variety population, although agreement among the top-ranked varieties was only partial. Historical (1985–2020) and future (2030–2049) simulations identified twelve varieties with HSC > 1.0 in both periods. The future simulations were driven by a single global climate model (ACCESS-CM2) under the SSP2-4.5 scenario and are therefore conditional on this specific climate realization; the twelve candidates were proposed as site-specific, climate-conditional priority candidates requiring further field validation, with Longjing 63 leading in overall performance and Longjing 39 leading in future adaptability. These findings demonstrate that the localized BDH-Rice model can effectively support site-specific climate-adaptive variety screening at the 856 Farm and provides, in principle, a transferable methodological framework for prioritizing candidate varieties in comparable cold-region production systems of Northeast China. The cultivar rankings reported here were derived from a single-site calibration and have not yet been independently validated across multiple locations; multi-site evaluation is therefore the necessary next step before these single-site cultivar rankings can be generalized beyond the 856 Farm.