Research on the sustainability evolution of cultivated land ecosystems in the North China plain integrating policy scenario simulation and deep learning diagnosis

As the core region for China's national food security, the North China Plain faces severe human-land conflicts and cultivated land ecological degradation. Existing studies have critical limitations: disconnection between assessment, prediction and policy analysis, and insufficient uncertainty quantification. This work is highly relevant to researchers in agricultural sustainability, food security and land spatial governance.We hypothesize that differentiated policy scenarios can achieve targeted regulation of Carbon-Emission-Coupled Ecological Footprint of Cultivated Land (EF). This study aims to establish an integrated diagnostic framework, to quantify EF spatiotemporal dynamics, simulate multi-scenario evolution, and diagnose implementation difficulty and ecological risks.Using 2010–2020 EF data of the study area, we employ a novel WGAN-coupled LSTM-Attention model under three scenarios: business-as-usual (BAU), ecological security priority (ESP), and food security priority (FSS). This framework advances the research paradigm from separate assessment to coupled diagnosis, extending deterministic prediction to probabilistic intervals.Results show 2010–2020 total EF rose by 1.31% while per capita EF fell by 3.94%, with significant spatial heterogeneity. By 2035, EF will increase by 17.6% under BAU; ESP curbs EF growth but has high implementation difficulty, while FSS achieves optimal system stability. Ecological risk presents a core-periphery structure, with 14.6% high-risk areas overlapping core food security zones. This study realizes quantitative characterization of cultivated land ecological risk management.