The aboveground biomass (AGB) of withered grass is an important early-warning indicator for grassland fire risk. Most grassland fires occur during the dry-grass season. In order to improve the fire-warning efficiency of withered AGB, it is essential to rapidly acquire the amount of withered-grass biomass. Remote-sensing data has been widely used in monitoring and estimating grassland yields during the growing season. However, applying remote sensing to the estimation of withered grass is still in need of exploration. The aim of this work was to try to establish a remote-sensing estimation model for withered AGB in the dry-grass season. The estimation of aboveground biomass can effectively prevent the occurrence of fire, protect the environment, facilitate local management and reduce economic losses. Our approach was to, first, calculate a dry-grass index based on Sentinel-2 image data and using ENVI, SNAP, and ArcGIS software. Second, a model to estimate the fuel quantity during the dry-grass season was established by regression analysis combined with field-measured data. Finally, the estimation model was used to predict the amount of fuel in different months of the dry-grass season, followed by the fire-defense elements, which were quantified and mapped in the Longzhao Marsh wetlands. It was found that: 1) the two indices were significantly correlated (0.678) with the amount of fuel; 2) the established model could accurately estimate the amount of fuel in the study area during the dry season, and accurate test results demonstrated that the correlation between the estimated results of the best model and the measured values was 0.863, indicating high accuracy; 3) the spatiotemporal variation of withered grass in the study area was obviously different, and the quantities of fuel predicted for the other months were more accurate, which may reflect monthly dynamic changes in actual fuel quantities; and 4) the establishment of a remote-sensing estimation model for fuel quantity in the Longzhao Marsh during the dry-grass season could provide important parameters for fire-risk warning in the western grassland of Jilin Province and Northeast China.

Urban form profoundly influences the effectiveness of coordinated management of carbon emissions and air pollution, which is of great significance for sustainable urban development. This study took 336 Chinese cities as research samples and used carbon emissions, PM_2.5, and nighttime light datasets from 2000 to 2020 to investigate the contribution of urban form to the degree of coordination between urban carbon emissions and PM_2.5. We constructed a multi-dimensional indicator system for urban form and adopted the coupling coordination degree (CCD) model to quantify the coordinated development of these two variables. A novel analytical framework was established using the stacking ensemble learning model. Combined with the SHAP method, this framework provided nonlinear associations of various urban morphological factors with the CCD. The results are as follows: (1) The stacking ensemble model performs best in capturing high-dimensional nonlinear relationships, with test set R², RMSE, and RPD values of 0.75, 0.082, and 1.98, respectively. (2) Urban form exhibits nonlinear threshold effects on the carbon-pollution coordination degree. The top predictors contributing to the model output are the class area (CA), the landscape shape index (LSI), the percentage of like adjacencies (PLADJ), and the number of patches (NP). Among them, CA exhibits a monotonically increasing asymptotic saturation response, while the LSI presents an inverted U-shaped response. (3) The contribution of urban form to the CCD exhibits significant spatial heterogeneity. The developed regions of eastern China constitute the high-sensitivity zone of the CCD, and are mainly constrained by scale expansion, land fragmentation, and excessive agglomeration, while the central and western regions show the low-sensitivity zone of the CCD. These results offer a scientific reference for designing pollution control and carbon mitigation strategies.