Over the past few decades, occurrences of extreme climatic events have escalated significantly, with severe repercussions for global ecosystems and socio-economics. northern China (NC), characterized by its complex topography and diverse climatic conditions, represents a typical ecologically vulnerable region where vegetation is highly sensitive to climate change. Therefore, monitoring vegetation dynamics and analyzing the influence of extreme climatic events on vegetation are crucial for ecological conservation efforts in NC. Based on extreme climate indicators and the Normalized Difference Vegetation Index (NDVI), this study employed trend analysis, Ensemble Empirical Mode Decomposition, all subsets regression analysis, and random forest to quantitatively investigate the spatiotemporal variations in historical and projected future NDVI trends in NC, as well as their responses to extreme climatic conditions. The results indicate that from 1982 to 2018, the NDVI in NC generally exhibited a recovery trend, with an average growth rate of 0.003/a and a short-term variation cycle of approximately 3 years. Vegetation restoration across most areas was primarily driven by short-term high temperatures and long-term precipitation patterns. Future projections under different emission scenarios (SSP245 and SSP585) suggest that extreme climate change will continue to follow historical trends. However, increased radiative forcing is expected to constrain both the rate of vegetation growth and its spatial expansion. These findings provide a scientific basis for mitigating the impacts of climate anomalies and improving ecological quality in NC.

Accurate estimation of forage aboveground biomass (AGB) from unmanned aerial vehicle (UAV) imagery is essential for monitoring alpine grassland productivity and supporting sustainable grassland management. However, the biomass of non-palatable species may be included in remote sensing-based estimates when their spatial distribution is not explicitly considered, potentially affecting spatial assessments of forage biomass. This study developed a UAV RGB-based framework for spatially constrained grassland AGB assessment by integrating the extraction of the growing-season non-palatable species Achnatherum splendens, mask-based spatial exclusion, feature optimization, and AGB inversion using field measurements from the northwestern shore of Qinghai Lake. Among tested semantic segmentation models, the Attention U-Net achieved the highest segmentation accuracy and was selected to identify A. splendens, while vegetation indices (VIs) and gray-level co-occurrence matrix (GLCM) texture features were optimized using the minimum redundancy maximum relevance (mRMR) algorithm. Random forest regression (RFR), support vector regression (SVR), and partial least squares regression (PLSR) models were subsequently evaluated. The Attention U-Net achieved high segmentation accuracy (PA = 89.98%, mIoU = 88.89%, and F1 = 93.88%). Feature optimization improved the performance of all regression models, with SVR providing the highest estimation accuracy (R² = 0.72; RMSE = 23.73 g m⁻²). The estimated AGB ranged from 67.88 to 205.85 g m⁻², revealing pronounced spatial heterogeneity. In a typical sample area with high-density A. splendens, excluding pixels classified as A. splendens reduced the estimated AGB by 15.45%, demonstrating the influence of dense A. splendens patches on spatial AGB assessment. These results demonstrate that integrating deep learning-based species masking with feature optimization provides a spatially explicit approach for grassland AGB assessment after excluding areas classified as A. splendens and provides useful spatial information for fine-scale grassland monitoring and management.