Under climate change and increasing water demands, groundwater depletion has become regional and global threats for water security, which is an indispensable target to achieving sustainable developments of human society and ecosystems, especially in arid and semiarid regions where groundwater is a major water source. In this study, groundwater depletion of 2003–2016 over Xinjiang in China, a typical arid region of Central Asia, is assessed using the gravity recovery and climate experiment (GRACE) satellite and the global land data assimilation system (GLDAS) datasets. In the transition of a warm-dry to a warm-wet climate in Xinjiang, increases in precipitation, soil moisture and snow water equivalent are detected, while GRACE-based groundwater storage anomalies (GWSA) exhibit significant decreasing trends with rates between-3.61 ± 0.85 mm/a of CSR-GWSA and −3.10 ± 0.91 mm/a of JPL-GWSA. Groundwater depletion is more severe in autumn and winter. The decreases in GRACE-based GWSA are in a good agreement with the groundwater statistics collected from local authorities. However, at the same time, groundwater abstraction in Xinjiang doubled, and the water supplies get more dependent on groundwater. The magnitude of groundwater depletion is about that of annual groundwater abstraction, suggesting that scientific exploitation of groundwater is the key to ensure the sustainability of freshwater withdrawals and supplies. Furthermore, GWSA changes can be well estimated by the partial least square regression (PLSR) method based on inputs of climate data. Therefore, GRACE observations provide a feasible approach for local policy makers to monitor and forecast groundwater changes to control groundwater depletion.

Urban particulate air pollution is a known cause of adverse human health effects worldwide. Urumqi is a large oasis city in which rapid urbanization has caused a series of eco-environmental problems including serious air pollution, water shortage, dense population, excess energy consumption, and the creation of an urban heat island, among others. Coal is the most important source of energy and air pollutants that are poorly dispersed into the natural surroundings are the main reasons for serious pollution in the Urumqi urban area. Using differential optical absorption spectroscopy (DOAS), aerosol levels were determined using the double optical path method. We found that aerosol concentrations in Urumqi increased rapidly in winter, and that the concentration of fine particles was much higher than that of coarse particles. The background aerosol concentration was highest in winter in the research area, and the air-flow speed had a significant impact on this because high speed surface winds that correspond to high air flows can transport the aerosol to other places. Some of the observed day-to-night differences may be caused by differing wind directions that transport air masses from different emission sources during the day and the night. Daily and seasonal differences in PM_1.0 concentrations of different grades of polluted air were statistically analyzed using average daily concentration data for particles smaller than 10, 2.5 and 1.0 microns (PM₁₀, PM_2.5 and PM_1.0), and meteorological observations for Urumqi, Tianshan District in 2010.

Near surface wind speed has significant impacts on ecological environment change and climate change. Based on the CN05.1 observation data (a gridded monthly dataset with the resolution of 0.25 latitude by 0.25 longitude over China), this study evaluated the ability of 25 Global Climate Models (GCMs) from Coupled Model Intercomparison Project phase 6 (CMIP6) in simulating the wind speed in the Arid Region of Northwest China (ARNC) during 1971–2014. Then, the temporal and spatial variations in the surface wind speed of ARNC in the 21st century were projected under four Shared Socioeconomic Pathways (SSPs), SSP1-2.6, SSP2-4.5, SSP3-7.0, and SP5-8.5. The results reveal that the preferred-model ensemble (PME) can fairly evaluate the temporal and spatial distribution of surface wind speed with the temporal and spatial correlation coefficients exceeding 0.5 at the significance level of p = 0.05 when compared to the 25 single models and their ensemble mean. After deviation correction, the PME can reproduce the distribution characteristics of high wind speed in the east and low in the west, high in mountainous areas, and low in basins. Unfortunately, no models or model ensemble can accurately reproduce the decreasing magnitude of observed wind speed. In the 21st century, the surface wind speed in the ARNC is projected to increase under SSP1-2.6 scenario but will decrease remarkably under the other three scenarios. Moreover, the higher the emission scenarios, the more significant the surface wind speed decreases. Spatially, the wind speed will increase significantly in the west and southeast of Xinjiang, decrease in the north of Xinjiang and the south of Tarim Basin. What’s more, under the four scenarios, the surface wind speed will decrease in spring, summer and autumn, especially in summer, and increase in winter. The wind speed will decrease significantly in the north of Tianshan Mountains in summer, decrease significantly in the north of Xinjiang and the southern edge of Tarim Basin in spring and autumn, and increase in fluctuation with high values in Tianshan Mountains in winter.

Short-term precipitation forecasting methods are mainly divided into statistical forecasting, numerical model-based forecasting, and radar image extrapolation techniques. The two methods based on statistical prediction and numerical model have the disadvantages of being unstable and generating large errors. Therefore, this study proposes the use of deep learning for radar image extrapolation for precipitation forecasting, in particular by developing algorithms for ConvLSTM and SmaAT-UNet. The ConvLSTM model is a fusion of a CNN (Convolutional Neural Network) and LSTM (Long Short-Term Memory network), which solves the challenge of processing spatial sequence data, which is a task that traditional LSTM models cannot accomplish. At the same time, SmaAT-UNet enhances the traditional UNet structure by incorporating the CBAM (Convolutional Block Attention Module) attention mechanism and replacing the standard convolutional layer with depthwise separable convolution. This innovative approach aims to improve the efficiency and accuracy of short-term precipitation forecasting by improving feature extraction and data processing techniques. Evaluation and analysis of experimental data show that both models exhibit good predictive ability, with the SmaAT-UNet model outperforming ConvLSTM in terms of accuracy. The results show that the performance indicators of precipitation prediction, especially detection probability (POD) and the Critical Success index (CSI), show a downward trend with the extension of the prediction time. This trend highlights the inherent challenges of maintaining predictive accuracy over longer periods of time and highlights the superior performance and resilience of the SmaAT-UNet model under these conditions. Compared with the statistical forecasting method and numerical model forecasting method, its accuracy in short-term rainfall forecasting is improved.

The arid region of northwest China (ARNC) is one of the most sensitive areas to global warming. However, the performance of new Global Climate Models (GCMs) from phase 6 of the Coupled Model Intercomparison Project (CMIP6) in simulating climate in this region, especially in the subregions, is not clear yet. Based on the temperature dataset from historical runs of CMIP6, this paper analyzed and evaluated the simulation ability of 29 GCMs in reproducing the annual mean temperature (tas), annual mean maximum temperature (tasmax) and annual mean minimum temperature (tasmin) in the ARNC and subregions from 1961 to 2014. The results show that (1) the correlation coefficients (CCs) between simulation and observation time series for the mean of two model ensembles (MME for equal-weight multi-model ensemble and PME for preferred-model ensemble) are generally better than those of 29 individual GCMs, with CCs ranging from 0.38 to 0.87 (p < 0.01). (2) All the models can simulate the significant warming trend of the three temperature elements in the study area well. However, the warming magnitude simulated by most of the models (41%) is smaller than the observations except for tasmax, which is also shown in the MME. (3) The spatial pattern of the three temperature elements can be better reflected by most models. Model simulation ability for the ARNC is better compared to that of the four subregions, with a spatial CC greater than 0.7 (p < 0.01). Among the subregions, the simulation performance of the north of Xinjiang for spatial pattern is superior to that of the other regions. (4) The preferred models for each subregion are various and should be treated differently when used. Overall, the PME outperforms both the MME and the individual models; it can not only simulate the linear trend accurately but also reduce the deviation effectively.

Agricultural oasis expansion and intensive management practices have occurred in arid and semiarid regions of China during the last few decades. Accordingly, regional carbon and water budgets have been profoundly impacted by agroecosystems in these regions. Therefore, study on the methods used to accurately estimate energy, water, and carbon exchanges is becoming increasingly important. Process-based models can represent the complex processes between land and atmosphere among agricultural ecosystems. However, before the models can be applied they must be validated under different environmental and climatic conditions. In this study, a process-based agricultural ecosystem model (Agro-IBIS) was validated for maize crops using 3 years of soil and biometric measurements at Wulanwusu agrometeorological site (WAS) located in the Shihezi oasis in Xinjiang, northwest China. The model satisfactorily represented leaf area index (LAI) during the growing season, simulating its peak values within the magnitude of 0–10%. The total biomass carbon was overestimated by 15%, 8%, and 16% in 2004, 2005, and 2006, respectively. The model satisfactorily simulated the soil temperature (0–10 cm) and volumetric water content (VWC) (0–25 cm) of farmland during the growing season. However, it overestimated soil temperature approximately by 4 °C and VWC by 15–30% during the winter, coinciding with the period of no vegetation cover in Xinjiang. Overall, the results indicate that the model could represent crop growth, and seems to be applicable in multiple sites in arid oases agroecosystems of Xinjiang. Future application of the model will impose more comprehensive validation using eddy covariance flux data, and consider including dynamics of crop residue and improving characterization of the final stage of leaf development.