Taking China's 31 provincial-level administrative regions, excluding Hongkong, Macao and Taiwan as research objects, this paper systematically evaluates the phased characteristics, regional disparities and dynamic evolution mechanisms of the regional-scale decoupling relationship in China from 2003 to 2023 by employing methods including remapping of decoupling index, kernel density estimation, Dagum Gini coefficient decomposition and spatial Markov chain. The results indicate that China's overall decoupling level has improved steadily, among which eastern region has maintained a high-level decoupling state for a long time, and central and western regions present a pattern of coexisting phased improvement and structural fluctuation. Decoupling index shows an overall upward trend in each period with continuously enhanced regional coordination, but a certain degree of reversal has occurred from 2021 to 2023. Interregional disparity is dominant contributor to unbalanced decoupling development, and unbalanced regional development is core issue. Although the internal coordination of each region has gradually improved over time, polarization trend has intensified in the later period. Since 2021, provincial decoupling index has exhibited significant spatial agglomeration characteristics: eastern region has formed a high-high agglomeration area with high decoupling level, while some western provinces have formed a low-low agglomeration area with low decoupling level. The regional decoupling state presents obvious spatial dependence and neighborhood spillover effects. The high decoupling state of neighboring regions can significantly increase the probability of local upward transition, while the adverse state also poses the risk of spatial diffusion.

Water scarcity, food crisis, and ecological degradation have been the bottlenecks in a considerable percentage of the world under the dual influence of global climate and human activities. Trade-offs between agricultural development and ecological effects are of great importance, especially in arid areas with highly developed agriculture. Water resources can be allocated, according to the mutual feedback relation among water, agriculture, and ecology. The agricultural and ecological water supply can also be optimized in different units. Synergy effects of the system can be expected to compromise the conflicts for the water security, sustainable development, and ecological health. In this study, an optimal allocation model of multi-water resources was constructed in the Shiyang River Basin using water-agriculture-ecology synergistic regulation. The objectives were taken as the groundwater balance, agricultural benefits, and ecological water use. Besides, the coordinated development degree was deduced using collaborative optimization with the NSGA-II algorithm. Trade-off and collaborative relationships were then quantified among water, agriculture, and ecology. The allocation schemes of water resources were finally proposed for the suitable water use proportion between agricultural and ecological under the synergistic promotion of water-agriculture-ecology. The results showed that the net water demand of irrigation was 14.99×10⁸ m³ in the Shiyang River Basin, whereas, the upper and lower water demand of ecological vegetation were 0.91×10⁸ and 3.35×10⁸ m³, respectively. The current balance between the supply and demand of water resources was attributed to the crowding out of ecological water and overexploiting groundwater. The water demand was also exceeded the water supply, in order to fully meet the minimum demand of for ecological water and groundwater. There was a water shortage of 1.44×10⁸-2.93×10⁸ m³/a in the Liuhe subsystem alone. Furthermore, the agricultural benefits, ecological water use, and groundwater balance were 137.53×10⁸ Yuan, 0.60, and 0.59×10⁸ m³, respectively, in the optimal allocation scheme. The agricultural benefits increased by 1.9% than before. The positive balance of groundwater was achieved in 0.59×10⁸m³. The agricultural and ecological water was accounted by for 90% and 10%, respectively, in the whole watershed, while that in canals and wells was 67% and 33%, respectively. The economic benefits of schemes S2 and S4 were reduced by 5.4%, compared with the optimal decision scheme (S0), while the groundwater balance of schemes S1 and S3 was reduced by 92.9% and 95.2%, respectively. Compared with the baseline, there was a 6% reduction in the water diversion from the middle reaches under diverse water and soil resource management, in order to guarantee a water inflow of 3.48×10⁸m³/a from Caiqi. In this case, the agricultural and ecological water consumptions of the Liuhe subsystem were 13.87×10⁸ and 1.62×10⁸m³, respectively, while the groundwater was in a positive equilibrium of 0.70×10⁸m³. When the inflow in Caiqi was 3.48×10⁸m³/a in a normal year, the ecological water use and groundwater were achieved to balance 1.6% of the agricultural benefits in the middle reaches. The ecological water use and the balance of groundwater were elevated by 4.8% and 18.6%, respectively, compared with the baseline. There was a synergistic trade-off between groundwater balance and ecological water in agriculture. The water resources in arid areas can be managed to maintain an appropriate balance among water security, agricultural development, and ecological health. Failure to consider the mutual influence among different water users can result in the decision biases and suboptimal schemes. This finding can provide an effective way to analyze the complex relationship among water, agriculture, and ecology. A strong reference was also offered for water resource planning and management in arid basins.

Frequent droughts have posed a serious threat to food security in the agricultural production bases of the Ningmeng Hetao Plain, FenWei Basin, and Yellow River irrigation area within the Yellow River basin. Accurately predicting drought is crucial for ensuring regional food production and water allocation. Among them, atmospheric circulation factors can significantly influence the occurrence, development, and transmission of agricultural drought via remote correlation. It is unclear whether introducing atmospheric circulation factors to the prediction model will improve the prediction performance of agricultural drought. Furthermore, the Meta-Gaussian (MG) model has been applied to drought prediction and achieved great prediction results. This study aims to determine the predictors of agricultural drought using correlation analysis, which includes agricultural drought, high temperature, and atmospheric circulation factors. Three groups of predictors were set: 1) agricultural drought (MG2 model), 2) agricultural drought and atmospheric circulation factors (MG3 model), and 3) agricultural drought, high temperature, and atmospheric circulation factors (MG4 model). The MG2, MG3 and MG4 models were used to predict agricultural drought in the Yellow River basin in typical years under the different forecast periods (1, 12, 24, 36, and 48 months). Then, a systematic investigation was made to explore the influence of atmospheric circulation factors on the prediction performance of agricultural drought in the MG model using the Nash-Sutcliffe efficiency coefficient (NSE) and root mean square error (RMSE). The results showed that the agricultural drought and high temperature were suitable for the predictors to predict the later agricultural drought. The standardized western Pacific subtropical high intensity index (SWPSHI) on 12-month time scale shared the most significant correlation with agricultural drought compared to other standardized atmospheric circulation indexes. Furthermore, 2014 was selected as a typical year because of the severe drought event in the Yellow River basin of 2014. Predictions of typical agricultural drought events in 2014 showed that the MG models with the different predictors in the long forecast period were different from the actual observed drought in the spatial scope and severity of the drought. The difference of NSE and RMSE indexes between different MG models were compared. It was found that the atmospheric circulation factor spatially varied in the prediction performance of the MG model. Compared to the one-factor prediction, the MG3 model with atmospheric circulation factor could improve the prediction performance of agricultural drought in the Yellow River basin by up to 46%, and the MG4 model with atmospheric circulation factor and high temperature had the largest improvement grid by 50%. Moreover, the prediction performance of the MG3 model with atmospheric circulation factor was improved in the Inner Mongolia Autonomous Region, Ningxia, Gansu, and Shaanxi provinces in the forecast period of more than one year. However, the prediction performance declined slightly in the rest region. Compared with the MG3 model, the MG4 model further improved the overall prediction accuracy of agricultural drought in the larger spatial range. Thus, it is recommended to take the atmospheric circulation factor into account when predicting droughts. The findings can provide guidance to predict the agricultural drought in the Yellow River basin.

To understand the spatial distribution and evolution trend of ecological drought vulnerability in Northwest China, the concept of ecological drought vulnerability and its quantitative characterization methods were proposed in terms of exposure, sensitivity and impact. Entropy weight method is adopted to develop the ecological drought vulnerability indices. Moran index is adopted to evaluate the correlation of drought vulnerability indices in spatial distribution and accumulation of vulnerability in local space. The regional distribution, vulnerability degree and trend of ecological drought vulnerability in Northwest China during 1985-2014 were analyzed. The results showed that ecological drought highly vulnerable areas in Northwest China were mainly distributed in Tianshan Mountains, southern Qinghai, the area near Qinghai Lake, central Gansu, northern Shaanxi and central Shaanxi, while the rest of the areas are moderate or low vulnerable. The spatial distribution pattern of ecological drought vulnerability index is relatively concentrated, showing a “large aggregation and small dispersion” pattern. Ecological drought vulnerability shows a trend of increasing and then decreasing during the studied period. The ecological drought vulnerability in Yili Basin, Turpan Basin and Hexi Corridor has improved significantly, while that in areas including southern Qinghai and northern Shaanxi still show a trend of increase. The ecological protection works in these areas need to be strengthen in the future.

In order to explore the changes of terrestrial water storage (TWS) and its components in China from April 2002 to March 2021, the interval data between GRACE gravity satellite and its follow-on satellite GRACE-FO were interpolated by using a hybrid VMD-LSTM model based on the decomposition-integration idea. The Theil-Sen slope analysis method and the Mann-Kendall trend test method were used to study the spatiotemporal evolution of TWS and its components in nine major river basins in China. The random forest method was used for analyzing the relative contribution rate of each component to TWS. The results showed that the VMD-LSTM model can effectively interpolate the GRACE sequence in China, with the Nash-Sutcliffe efficiency coefficient greater than 0.6, correlation coefficient greater than 0.9, and root mean square error less than 2cm in most regions, which significantly improves the interpolation accuracy. There is a spatial consistency between the change trends of TWS and groundwater storage. Except for the Southeast River Basin, Pearl River Basin, and Yangtze River Basin, TWS in other basins shows a downward trend, which is mainly caused by groundwater deficit. TWS in western arid and semi-arid regions and North China is mainly affected by groundwater storage, and TWS is greatly affected by canopy water storage and soil water storage in humid and semi-humid regions, compared to arid and semi-arid regions.

In arid areas, the rapid expansion of agricultural activities has caused a series of ecological and environmental issues such as vegetation degradation, declining groundwater levels, and land desertification, posing significant challenges to the stability and development of inland river basins. The development of agricultural water suitability is the key to ensure the water and land resources efficiency and ecological health in arid areas. The core is how to implement the scientific control of irrigation scale and agricultural irrigation water consumption. Taking the Shiyang River Basin as the study area, this paper analyzed the variations of high and low flow for surface runoff in the basin based on the anomaly percentage and the runoff modulus ratio coefficient. After estimating the non-agricultural irrigation water demand, the water balance model was employed to calculate the available surface water and groundwater resources specifically for agricultural irrigation, while also considering water transformation process. Based on this, a multi-objective planting structure optimization model was established to determine the water-saving and efficient crop planting structure. To determine the appropriate irrigation scale in arid areas, a comprehensive model was constructed by improving the water-heat balance model, which incorporates various factors such as the water transformation process, crop coefficient in different growth stages, agricultural water saving level, and crop planting structure. Lastly, the suitable irrigation scales of the Shiyang River Basin under different scenarios were explored by using this model. The results showed that the runoff in Shiyang River Basin presented obvious high and low flow variations during the studied period. Specifically, the basin’s total surface water resources amounted to 17.47×10⁸, 14.19 ×10⁸, and 12.25×10⁸ m³ during high, normal and low flow years, respectively. Regarding non-agricultural irrigation water demands, the critical and suitable ecology water demand scenarios were 3.43×10⁸ and 5.57×10⁸ m³, respectively. Following the fulfillment of water resources demand for non-irrigation sectors, and under the joint regulation of surface water and groundwater resources, the available water consumption for agricultural irrigation were 18.97×10⁸-21.75×10⁸, 14.75×10⁸-17.51 ×10⁸, and 12.31×10⁸-14.95×10⁸ m³ in high, normal and low flow years, respectively. With the improvement of water saving level, the decrease of groundwater recharge led to the decrease of irrigation water consumption in the basin. After optimizing the crop planting structure, 14.13% of irrigation water saving can be achieved by reducing the economic benefit of 2.94%. The suitable irrigation scale determined by the suitable irrigation scale calculation model in arid areas considering the water transformation process was consistent with the actual situation, which is more reasonable than the conventional method in the inland river basin. Based on this method, the suitable irrigation scales of Shiyang River Basin under the current condition were 27.73×10⁴-31.66×10⁴, 21.55×10⁴-25.76×10⁴, and 18.01×10⁴-22.03×10⁴ hm² in high, normal and low flow years, respectively. After improving the water-saving level and adjusting the planting structure, the suitable irrigation scale of the basin has increased. In 2020 (normal flow year), the actual irrigation area exceeded the critical suitable irrigation scale, necessitating reducing the irrigation area of 2.13×10⁴-6.34×10⁴ hm² to balance the water and soil resources in the basin. The research results provide a theoretical foundation for decision makers to formulate water-appropriate agricultural development plans at the macro level.

Vegetation is one of the key components in terrestrial ecosystems, particularly in climate, carbon balance, and water cycling. The healthy growth of vegetation can depend mainly on the context of climate change, such as more frequent occurrences of droughts and heat waves. Moreover, the degradation of vegetation can cause irreversible damage to the structure and functionality of terrestrial ecosystems. Compound hot and dry events can further exacerbate their impact on the vegetation, due to the positive feedback between droughts and heatwaves. Therefore, it is very necessary to evaluate the response relationship between vegetation and compound hot and dry events, in order to clarify the ecosystem response to climate change for effective mitigation and adaptation strategies. The correlation between vegetation and drought or hot indices has been commonly used to determine their relationship in recent years. However, the overall correlation coefficient is often insufficient to capture the tail dependence between extreme events and vegetation conditions. In this study, a Vine Copula-based model was constructed to assess the vulnerability of vegetation using the detrended and standardized normalized vegetation index (SNDVI), standardized precipitation and evapotranspiration index (SPEI), and standardized temperature index (STI). The probability of vegetation loss was determined for the different land use and climatic regions in the Loess Plateau from 1982 to 2015. The results show: 1) There was a positive correlation between SNDVI and SPEI in most areas of the Loess Plateau, whereas, a negative correlation was found between SNDVI and STI. Grasslands shared the highest correlation between SNDVI and SPEI, followed by cropland, and the forests with the lowest correlation. There was no significant difference in the correlation distribution between grassland and cropland SNDVI with SPEI from June to August. However, there was a gradual decrease in the correlation between forest and shrub SNDVI with SPEI. The correlation distribution between grassland and cropland SNDVI with STI remained relatively unchanged, while the correlation between forest and shrub SNDVI with STI gradually increased. 2) The vulnerability values of vegetation to extreme compound dry and hot conditions were 0.51, 0.57, and 0.55 respectively, in June, July, and August, which was significantly higher than those to single drought or high-temperature events. Regions with higher vulnerability were concentrated in the northern part of Shaanxi, Ningxia, eastern Gansu, and Inner Mongolia. 3) The vegetation exhibited a significantly higher vulnerability to extreme compound dry and hot conditions in arid and semi-arid regions, compared with the single extreme drought or high-temperature events. In humid and semi-humid regions, vegetation vulnerability to extreme compound dry and hot conditions was generally comparable or slightly lower than that to single drought or high-temperature events. 4) There was a great variation in the vulnerability of different vegetation types. The vulnerability was ranked in the descending order of grassland, cropland, shrub, forest. The vulnerability of grassland to extreme compound dry and hot conditions increased by 26% to 56% during summer, compared with single drought or high-temperature events, while the cropland vulnerability increased by 11% to 24% and 19% to 48%, respectively. Since the SNDVI can only represent the vegetation cover, more vegetation indices (such as gross primary productivity (GPP) and net primary productivity (NPP)) can be expected to assess the vegetation dynamics and their response to extreme events, in order to reduce some uncertainties in the future research.

Multiple deep learning methods were used to interpolate gravity recovery and climate experiment (GRACE) data, and the random forest algorithm was used to spatially downscale GRACE data. The actual evapotranspiration in the Yellow River Basin was calculated based on the water balance equation. And the data were verified using four evapotranspiration products to analyze the spatio-temporal evolution of actual evapotranspiration in the Yellow River Basin. The results indicate that the overall interpolation accuracy of the long short-term memory neural network is superior to that of deep neural network and convolutional long short-term memory neural network. The average correlation coefficient between the actual evapotranspiration estimated based on GRACE data and four evapotranspiration products is 0.903, indicating that the applicability of the actual evapotranspiration results estimated based on GRACE data is good. The average annual actual evapotranspiration in the Yellow River Basin from 2003 to 2021 was 144.38 to 775.62 mm, with a spatial distribution pattern of more in the south and less in the north, and a seasonal variation pattern of more in summer and less in winter. From 2003 to 2016, it increased at a rate of 2.51 mm/a, and showed a downward trend after 2017.

To study the spatiotemporal response of different types of droughts in the Yangtze River Basin in 2022, a three-dimensional spatiotemporal cluster identification method was used to extract meteorological drought, agricultural drought, and hydrological drought events on monthly scale in 2022, and the matching rule for three types of drought events was employed to identify the meteorological-agricultural-hydrological drought event pair with spatiotemporal connections on pentad scale (i. e. , five-day) in 2022, so as to quantitatively reveal the spatiotemporal response characteristics among various drought types in the Yangtze River Basin. The results indicated that three meteorological drought events, two agricultural drought events, and two hydrological drought events lasting longer than two months were extracted on monthly scale in the Yangtze River Basin in 2022, and the three-dimensional spatiotemporal cluster identification method can clearly depict the occurrence, development, and extinction processes of various drought types in time and space. Furthermore, a meteorological-agricultural-hydrological drought event pair, in which a hydrological drought event that was jointly triggered by meteorological drought and agricultural drought events in 2022, was identified based on the matching rule. The migration paths of the three drought types were all from the upper reaches toward the middle and lower reaches, and finally terminated in Hubei, Chongqing, and Hubei, respectively.

A water resources carrying capacity evaluation index system consisting of five subsystems, including input, consumption, vitality, regulation, and output, was constructed based on the theories of water metabolism and water cycle. The weights of different indicators were determined using the least squares method, together with the analytic hierarchy process and entropy weight method, and the water resources carrying capacity of the Shiyang River Basin was comprehensively evaluated using the variable fuzzy set model. The Gaussian mixture regression model was coupled with three interpretable machine learning methods to quantify the impact of each evaluation indicator on the water resources carrying capacity, and explore their relationships on both global and local scales. The results indicate that from 2011 to 2020, the water resources carrying capacity of the river basin showed an improving trend with fluctuation, but it was still on the verge of overload. The score of water resources carrying capacity increased from 3.79 in 2011 to 4.18 in 2013, and then decreased to 3.23 in 2020. The Gaussian mixture regression model performed well in handling high-dimensional and small-sample water resources carrying capacity indicator data. Indicators, including per-unit water consumption for agricultural irrigation, reuse rate of sewage treatment, water consumption rate of ecological environment, water resources development and utilization rate, water production modulus, and groundwater exploitation rate are the dominant factors restricting the water resource carrying capacity of the river basin. From a global perspective, there was a non-linear relationship between water resources carrying capacity and the dominant factors, and water resources carrying capacity changed non-monotonically with those indicators. From a local perspective, the dominant factors tended to inhibit water resource carrying capacity from 2011 to 2015, while they gradually shifted to promotion factors from 2016 to 2020. Although the water resource carrying capacity in the river basin has been improved, it is necessary to strengthen the management in water resources development and utilization and reduce groundwater extraction rate in future.