<p>With ongoing global climate change, drought has become the primary threat constraining food security in China. Traditional assessment frameworks based on administrative boundaries or macro-climatic zoning overlook variation in vulnerability affected by key agronomic practices, such as crop phenology and cropping systems, thereby limiting their accuracy. To address this research gap, this study developed and validated a novel drought risk assessment framework based on agricultural cropping zones (single-, double-, and triple-cropping zones). The framework coupled a Geographical and Temporal Neural Network Weighted Regression (GTNNWR) model for forecasting future crop vegetation dynamics with the Standardized Precipitation Evapotranspiration Index (SPEI) to assess drought risk under historical (2001–2020) and projected future (2021–2100) scenarios. The GTNNWR model achieved <i>R</i>² values ranging from 0.72 to 0.82 and RMSE values between 0.11 and 0.14 for NDVI prediction, significantly outperforming conventional models. Historical drought risk assessment revealed that drought events were most frequent during summer and concentrated in single-cropping and double-cropping zones. Future projections indicate a substantial intensification of drought risk. Under the Shared Socioeconomic Pathway (SSP)126 scenario, drought risk is projected to increase in the triple-cropping zones of the middle and lower reaches of the Yangtze River Plain. Under the SSP245 scenario, the frequency of spring and winter droughts is anticipated to rise markedly. Under the SSP585 scenario, drought intensity is projected to intensify in central–eastern single-cropping zones and southwestern double-cropping zones. This assessment framework based on agricultural cropping zones can precisely identify drought risks and facilitate adaptation in agricultural management, such as optimizing irrigation systems and adjusting crop structures.</p>