In the past few decades, a warming-wetting trend has been observed in the arid and semiarid Northwest China, especially in Xinjiang. According to the Clausius-Clapeyron (C-C) relationship, the atmospheric moisture-holding capacity increases with rising temperature, which may lead to an intensification of extreme precipitation. In this study, we examined the spatial distribution of scaling relationships between precipitation and dew point temperature in Xinjiang, using intra-annual binning scaling and inter-annual trend scaling based on station observations and ERA5. The results show that the predominant binning scaling in Xinjiang is sub C-C scaling (0%/°C-5%/°C), transitioning from northern Xinjiang to C-C like scaling (5%/°C-9%/°C) or super C-C scaling (>9%/°C) near the Tianshan Mountains and southern Xinjiang. The observed trend scaling exhibited a similar spatial pattern to that of binning scaling, but with more super C-C scaling. Furthermore, the reduction in vertical velocity at the southern foothills of the Tianshan Mountains may lead to a decrease in precipitation intensity. Overall, these findings provide a solid basis for enhancing our understanding of how climate change influences precipitation events in Xinjiang.

In the context of climate change, severe soil wind erosion has caused a series of negative impacts on the ecosystems in China's arid Northwest region. With the increasing frequency and intensity of extreme weather and climate events, the vulnerability of arid ecosystems has been further exacerbated. However, the mechanisms through which different extreme events affect soil wind erosion remain unclear. Thoroughly investigating the occurrence patterns of extreme events and revealing the response processes of soil wind erosion to these events hold significant importance for future climate risk mitigation initiatives. Based on this understanding, this study focused on China's northwestern arid region as the research area. Using the relative threshold method and the revised wind erosion equation model (RWEQ), we explored the spatiotemporal characteristics of extreme events—droughts, strong winds, and high temperatures—as well as soil wind erosion. The relationship between the intensity of extreme events and soil wind erosion during their occurrence was also analyzed. The main findings are as follows: 1) Areas with high drought intensity were primarily concentrated in the southwestern part of the study region, showing an initial decline followed by an increasing trend over time. Regions with high strong wind intensity were clustered in the central and northern areas, where strong wind intensity decreased significantly during the study period and stabilized after 1990. High-temperature intensity exhibited minimal spatial variation across the region and showed a significant upward trend over time. 2) Soil wind erosion across the study area remained at a relatively high level overall, with high-erosion zones mainly distributed in the Tarim Basin and the central part of the Hexi Corridor. Over the four decades, soil wind erosion displayed a pronounced decreasing trend throughout the region, particularly in areas near the Tianshan Mountains. 3) Soil wind erosion reached its peak during drought events, followed by high-temperature events. A positive correlation existed between the intensity of extreme events and soil wind erosion during such events. Additionally, extreme events incurred higher risks of soil wind erosion in barren lands and desert areas. This study provides a reference for understanding the impacts of extreme weather and climate events on soil wind erosion under global change and serves as a decision-making basis for regional desertification prevention and control.

The northern slope of the Tianshan Mountains is an important water source conservation area and a key grassland animal husbandry base in Northwest China. The meltwater from snow is crucial for maintaining ecosystems, agricultural irrigation, and urban water supply. In order to solve the problem of missing data caused by cloud interference in MODIS snow products, this study extended the MODIS data input and used existing snow data to identify pixels as either snow or non-snow as "true values". Machine learning algorithms such as random forest, support vector machine, and BP neural networks were applied to determine the best approach for snow identification. By combining various cloud removal methods with the hidden Markov random field (HMRF) algorithm, a comparative analysis of cloud removal effects was conducted, and the accuracy of the experimental results was verified using high-resolution Landsat data. The results show that: 1) The random forest model performed best in the binary snow classification task, with an accuracy of 90.15% and a precision of 91.95%. 2) Collaborative cloud removal using multiple data sources yielded good results, with a Kappa coefficient of 0.729, but combining the HMRF method achieved the best cloud removal effect, with an overall accuracy of 82.84%, producer accuracy of 88.46%, and Kappa coefficient of 0.795. 3) The trends of annual average snow days, snow-covered days in relation to altitude, monthly average snow coverage rate, and annual average snow-covered area show high consistency with existing data.

The northern slope area of the Tianshan Mountains, as an important energy base in China, is focusing on the development of renewable energy sources such as photovoltaic power. To support the industrial layout, there is an urgent need for high-quality, long-term historical radiation data. This study proposed a solar radiation data fusion method that combines subspace and boosting ensemble learning methods to construct a regression model, fully leveraging the spatiotemporal representativeness of the data sources. Based on this framework, three gridded radiation datasets—ERA5, FLDAS, and TERRA—were fused to generate the 0.1° solar radiation data for the northern slope area of the Tianshan Mountains from 1990 to 2020. Compared to the original datasets, the newly generated dataset shows significant improvements in validation metrics such as correlation coefficient, root mean square error (RMSE), and mean absolute error (MAE). Among the 12 seasonal indicators, the new dataset ranks the highest in 10, providing a more accurate reflection of radiation variations over the 30 years. The results indicate that the average radiation across the northern slope area of the Tianshan Mountains showed an increasing trend until 2008, after which it began to decrease. In the eastern region, radiation increased most significantly in the spring (with a rate of 3.6 W/(m²·10 a)), followed by the winter, while radiation in the western region showed a notable decrease in the autumn. Comparing the spatial distribution of solar radiation with existing photovoltaic power stations, the utilization of solar energy resources in the east of Urumqi and Hami can be further optimized. The research results can provide a reference for the high quality development of the regional photovoltaic industry.

A detailed analysis of the spatiotemporal dynamics of different land use types and their impacts on carbon storage in arid regions is of great significance for regional ecological protection and sustainable development. Taking the northern slope of the Tianshan Mountains in northwestern China's arid region as a case study, this study systematically analyzed the spatiotemporal change of land use and carbon storage, along with their driving mechanisms, from 2000 to 2030 based on the SD-PLUS-InVEST model and the Geodetector model. The results are summarized as follows: 1) From 2000 to 2020, land use transitions in the study area were frequent, with cultivated land and construction land expanding, while grassland and woodland declined. Under the 2030 land use projection scenarios (SSP119, SSP245, and SSP585), grassland and unused land are dominant land use types. Cultivated land, woodland, and construction land are projected to expand, while grassland and water bodies are expected to decline further. 2) Carbon stocks in the study area exhibited a general increasing trend over the past two decades, with a total increase of 4.70×10⁷ t. The most significant increase in carbon stocks occurred in cultivated land, followed by construction land, whereas grassland consistently experienced decline in carbon storage but remained the region's primary carbon sink. Compared to 2020, carbon stocks are projected to increase across all scenarios by 2030, with increases of 5.90×10⁷ t, 5.28×10⁷ t, and 3.08×10⁷ t respectively. 3) NDVI, soil type, and population density are the main driving factors of spatial variation in carbon storage in the study area, and their interactions show stronger explanatory power than individual factors, with the interaction between NDVI and soil type being the most significant. The results can provide a scientific basis for land spatial resource planning and carbon sink enhancement in the study area under the framework of the "dual carbon" goals.