In recent years, extreme precipitation events have become increasingly frequent and widespread largely due to global warming and the urban rain island effect. Intense rainstorms often trigger secondary hazards such as floods, urban waterlogging, debris flows and landslides, resulting in multi-hazard scenarios and the loss amplification effect. These rainstorm-induced cascading hazardsRainstorm-induced cascading hazards significantly snowball the challenges of hazard risk prevention and emergency response. This study focuses on the Daqing River basin in China and initially develops a historical hazard dataset of rainstorm and its secondary hazards across 47 county-level administrative units using internet data miningData mining and generative AIGenerative AI techniques. Subsequently, the spatial distribution of diverse types of rainstorm-induced cascading hazardsRainstorm-induced cascading hazards is analyzed. The loss amplification coefficient is finally calculated from six loss dimensions—affected population, fatalities and missing persons, emergency relocations, collapsed houses, affected crop area and direct economic losses—to quantitatively examine the loss amplification effect of rainstorm-induced cascading hazardsRainstorm-induced cascading hazards. Results show that from 2010 to 2024, a total of 404 rainstorm-induced cascading hazardRainstorm-induced cascading hazards events occurred in the Daqing River basin. Among them, the rainstorm-urban waterlogging cascade was the most frequent (123 times), followed by the rainstorm-landslide cascade (121 times). The frequency of rainstorm-induced cascading hazardsRainstorm-induced cascading hazards in the central and eastern parts of the basin was significantly higher than in the west, with a marked spatial concentration in Baoding City in the central basin. The loss amplification coefficients ranged from 1.67 to 4.62 for two-hazard cascades (e.g., rainstorm-urban waterlogging, rainstorm-flood, rainstorm-landslide and rainstorm-debris flow), and from 3.06 to 6.03 for three-hazard cascades (e.g., rainstorm-flood-landslide and rainstorm-flood-debris flow). The amplification effect was particularly pronounced in rainstorm-landslide among two-hazard cascades and rainstorm-flood-debris flow cascades among three-hazard cascades.

In recent decades, warming-induced advances in plant phenology have become widespread across the globe. However, whether early spring greening has beneficial or adverse effects on ecosystem responses to subsequent summer droughts remains unclear. To bridge this gap, this study employs vegetation remote sensing and meteorological data (1982-2022) to construct a Drought Propagation Index (DPI), analyzing the spatiotemporal patterns of how spring greening influences summer drought propagation risk from meteorological to ecological drought in Northern Hemisphere (NH) ecosystems. Research indicates that earlier spring greening exacerbates the overall risk of summer drought propagation in NH ecosystems by significantly increasing drought propagation probability (PP) and shortening propagation time (PT). This adverse effect amplifies synchronously with the intensification of meteorological drought stress. Under extreme drought conditions, greening leads to an average PP increase of 11.4%, resulting in heightened propagation risks across over 60% of the study area. The impacts of greening exhibit spatial heterogeneity. In high-latitude humid and sub-humid regions, the effects of greening are dominated by the Vegetation Structural Overshoot (VSO) pattern, leading to increased DPI in over 75% of grids in boreal forest and significantly heightened summer drought propagation risks. In contrast, grassland and shrub in arid and semi-arid experience relatively minor adverse effects from spring greening, with more than half of the grids showing alleviated summer drought propagation processes. Our results highlight that ecological drought risk planning under climate change, should consider the role of spring greening in promoting drought propagation, especially for ecosystems in humid regions with marked greening trends.

Study region The Dongjiang River Basin in southeast China, a critical water source region for the Pearl River Delta with a subtropical monsoon climate and frequent dry–wet transitions. Study focus Flash droughts threaten water and food security due to rapid onset, yet operational warnings remain largely single-indicator based. This study developed a progressive early warning system using pentad-scale ERA5-Land root-zone soil moisture (1950–2025). Precursors were screened through collinearity diagnosis, time-lag correlation, and random forest importance, followed by Receiver Operating Characteristic (ROC) analysis, with predictive skill quantified by the area under the ROC curve (AUC) to determine warning thresholds within a two-stage regression-to-classification framework. New hydrological insights for the region Five precursors spanning three pentads prior to onset were identified. A two-pentad-ahead precipitation anomaly (Plag2, AUC = 0.74) and a one-pentad-ahead anomaly (Plag1, AUC = 0.92) formed the main warning chain, issuing L1 and L2 alerts 2 and 1 pentads ahead, respectively. Three weaker precursors (Plag3, ET/PETlag3, Tlag2; all AUC ≈ 0.5) were integrated into a three-category supporting-condition score (LOW/MEDIUM/HIGH). The system achieved a false positive rate of 6.3% (training) and 8.5% (validation), substantially lower than single-indicator (16.5%–18.6%) and logistic regression (16.3%–22.3%) baselines, with a higher Heidke Skill Score (0.370 vs. 0.262–0.269). Results were robust across 10 flash drought definition parameter combinations, and application to the 2019 event demonstrated timely, coherent warnings with no false alarms.

Percentile-based thresholds widely used in flash drought monitoring lack a clear physical basis, as the same percentile may correspond to substantially different soil moisture states under different soil and climatic conditions. This study constucted a framework based on the soil moisture loss rate function L(SM) to identify root-zone loss stages, estimate physically informed thresholds, and compare them with statistical percentile thresholds. Using ERA5-Land reanalysis data from 1950 to 2024, the framework combines non-parametric identification of L(SM) with piecewise linear fitting to characterize root-zone soil moisture loss stages and estimate physically informed thresholds in the Dongjiang River Basin, southeastern China. Two physically informed critical thresholds were identified: SMWT, representing the wet–transitional boundary, and SMTD, representing the transitional–dry boundary. Their relationships with statistical percentile thresholds and reference soil hydraulic parameters were then evaluated through spatial comparison and regression analysis. Results showed that root-zone L(SM) curves were overwhelmingly dominated by a wet–transitional–dry three-stage structure, accounting for 92.5% of all grid cells, while the gravitational drainage stage was absent at the root-zone scale. The basin-mean SMWT and SMTD were 0.36 and 0.27 m³/m⁻³, respectively, with mean bootstrap 95% confidence interval widths of 0.03 and 0.02 m³/m⁻³. The 40th and 20th percentile thresholds were fully nested between SMWT and SMTD, and their correlations with the physically informed thresholds reached 0.98–0.99. However, the statistical threshold window covered only approximately 55.6% of the physically informed transition zone, indicating weaker ability to characterize terminal drought severity than flash drought onset. The proposed framework provides physically informed reference thresholds for flash drought monitoring and supports regional adaptation of percentile-based methods.

The dampening effect of antecedent soil moisture is crucial for terrestrial ecosystems to withstand the impacts of meteorological drought. However, the capacity of soil moisture to mitigate the propagation of meteorological drought to ecosystems has not been quantified at large scales. In this study, we focused on Northern Hemisphere ecosystems during the peak summer growing season. Using soil moisture and remote sensing data (1982-2022), we quantified the Dampening Rate (DR) of soil moisture on drought based on the proposed Drought Propagation Index (DPI). Our results reveal that antecedent soil moisture exerts significant dampening effects on drought propagation, with an average DR of 54.4% across all drought severity levels. Spatially, the dampening capacity of soil moisture was substantially stronger in Humid and sub-Humid regions than in Arid and semi-Arid areas. Vertically, middle and surface soil layers showed the most pronounced buffering against extreme drought, particularly in Humid zones, where surface and middle-layers maintained average DR values of 59.2% and 61.4% respectively. Over the past four decades, the dampening capacity remained generally stable across most of the study region. However, significant declines in DR (p < 0.05) were detected in several critical areas including the Midwestern US, southwestern Europe, southwestern Russia, and the Mongolian Plateau, where rising temperatures combined with decreasing soil moisture have substantially weakened this buffering capacity. Our findings underscore the critical role of antecedent soil moisture in mitigating meteorological droughts, and quantification method of the dampening provide valuable insights for ecological drought monitoring and prediction under climate change.

The interactions between atmospheric and soil dryness are complex. Nevertheless, their influence on vegetation vulnerability and its evolutionary patterns has not been fully elucidated. To bridge this gap, we developed a novel probabilistic framework based on set intersection and union theory to disentangle the individual and interactive effects of soil moisture (SM) and vapor pressure deficit (VPD) on the Normalized Difference Vegetation Index (NDVI). Using this method, we quantified the strength, relative contribution, and temporal dynamics of SM-VPD interactions. The results revealed that: (1) Across Mainland China, excluding the permafrost regions of the northwest, the interaction between VPD and SM primarily exerted an inhibitory effect on NDVI, and this interaction intensified with increasing dryness severity. (2) Moreover, the relative contribution of this interaction to vegetation changes increased under more severe dryness conditions. (3) Spatially, the strength and contribution of the interaction showed significant increasing trends predominantly in northeastern China, while decreasing trends dominated in central and southern regions. Conversely, the individual effects of SM and VPD predominantly exhibited significant decreasing trends in the northeast and increasing trends in central and southern China. (4) In regions characterized by intensified land-atmosphere coupling, both the interaction strength and its contribution generally increased, whereas the individual effects of SM and VPD decreased. Overall, the dynamic changes in the interaction were found to be primarily associated with land-atmosphere coupling.

Study region The Daqing River basin in China has been frequently affected by floods. Flood risk exhibits both temporal and spatial interdependence, driven by hydrological memory effects and cascading flood processes within spatially connected river systems. Study focus The study elucidated spatiotemporal propagation patterns of flood risk and developed a novel spatiotemporal propagation function for risk prediction. Initially, flood risk in twenty-two subbasins of the Daqing River was evaluated by integrating hazard, exposure, vulnerability and mitigation capacity. Markov-based state transition diagrams were subsequently employed to calculate flood risk propagation probabilities, distinguishing temporal and spatial risk propagation patterns. Moreover, a spatiotemporal risk propagation function incorporating risk propagation effect and flood control infrastructure interference was proposed to predict flood risk. New hydrological insights Results indicate pronounced spatial heterogeneity in flood risk across the Daqing River basin during 2000–2020, with high-risk areas concentrated in northern region and eastern edge. Temporally, interannual flood risk variation in each subbasin was driven by mid-high and high, and low and medium grade transitions. Spatial propagation between adjacent subbasins was dominated by risk attenuation across 44.8% of adjacent subbasins. The spatiotemporal propagation function achieved an average prediction accuracy of 90.5% and disentangled temporal and spatial propagation contributions to flood risk. Additionally, five reservoirs and flood detention areas in the central-eastern basin reduced risk propagation by 26.4% and 30.3% via their buffering capacities.

Under global climate change, frequent droughts threaten ecosystem functions, but how drought characteristics affect ecosystem resilience remains unclear. Focusing on the Dongjiang River Basin, China, we identified drought events at an 8-day scale from 2000-2024 using multi-source remote sensing and reanalysis data. The water use efficiency-based resilience index (R-de) was calculated, and a random forest model quantified the contributions of 21 potential driving factors. The model explained 68% of R-de variance (R-2 = 0.68, RMSE = 0.12). Downward shortwave radiation was the primary factor, followed by antecedent water use efficiency and soil moisture anomaly, with drought intensity and air temperature ranking fourth and fifth. All dominant factors exhibited nonlinear threshold effects: R-de decreased significantly after radiation exceeded similar to 110 W.m(-2).(8d)(-1); R-de declined when standardized soil moisture anomaly fell below -2.0; and R-de increased sharply when drought intensity exceeded 12%. Drought intensity far outweighed duration and severity, establishing it as the key drought attribute. This study reveals the dominant drivers and their thresholds governing ecosystem resilience in the Dongjiang River Basin, providing quantifiable indicators for ecological drought early warning.

Transitioning from passive drought control to active risk management requires a systematic understanding of drought propagation beyond conventional short chains. This study established a novel framework to characterize high-resolution, cascading drought dynamics across meteorological, hydrological, agricultural, ecological, and socioeconomic dimensions in the drought-prone Yellow River Basin. By integrating SWAT and AquaCrop simulations with electrical network-inspired “series–parallel–hybrid” theory, we identified dominant propagation pathways and thresholds across 403 sub-basins. Results revealed distinct spatiotemporal groupings where meteorological–ecological and meteorological–agricultural droughts shared similar propagation times, as did meteorological–socioeconomic and meteorological–hydrological droughts. Meanwhile, when ecological drought of varying severity occurred in phase 1, the cascade consistently progressed from meteorological to ecological, then to agricultural, further to hydrological, and ultimately to socioeconomic drought. Series and hybrid modes dominated drought propagation pathways, with series propagation increasing under intensified meteorological drought. As drought severity rose, propagation thresholds declined while exhibiting clear spatial clustering patterns. Driving force analysis indicated that dynamic cumulative processes rely more on environmental drivers than static threshold conditions. This study fills a critical gap in long-chain drought propagation research and supports the development of cascade-based early warning and targeted regulation systems.

Reservoir operations alter natural flow regimes by redistributing water spatially and temporally. Understanding and addressing the resulting impacts on natural flow regimes are crucial for effective river management to maintain the health and resilience of riverine ecosystems. This study contributes a comprehensive assessment regarding the impact of reservoir operations on flow regimes across the Contiguous United States (CONUS). We first identified admissible flow variation range based on reservoirs’ inflows and calculated the High and Low Flow Impacts (HFIs and LFIs) of reservoir operations on flow regimes. The dynamic time warping algorithm was implemented to cluster HFIs and LFIs, enabling detailed analysis of their intra-annual characteristics. Furthermore, we conducted a comprehensive trend analysis of HFIs and LFIs from 1980 to 2020 across CONUS. Our analysis reveals notable seasonal variations in the impacts of reservoir operations on high and low flows throughout the year, with positive correlations between the annual mean HFIs and LFIs. We also identify three distinct patterns for both intra-annual LFIs and HFIs, which show minimal correlation with geographical location or the primary purposes of reservoir use. The trends of the annual HFIs and LFIs highlight diverse patterns, indicating substantial reductions in HFIs and simultaneous increases in LFIs in numerous reservoirs over recent decades. These trends are not associated with reservoir capacities, discharge levels, or the degree of regulation but are correlated with trends in annual mean flows. Our results highlight the complex, evolving nature of flow regime alterations and emphasize the need for adaptive management strategies that account for both seasonal variations and long-term trends in reservoir operations. These findings provide critical insights for developing sustainable river management practices that balance human water needs with ecosystem conservation requirements.

Climate change has intensified the global water cycle, leading to more extensive, concurrent, and frequent occurrences of drought. Water serves as a critical driver of soil carbon dynamics, which in turn significantly influences soil quality. However, the coupled effects of drought and land degradation on vegetation resilience remain poorly understood, particularly in ecologically vulnerable and strategically important regions like Northeast China’s black soil zone. Here, we conducted a comprehensive regional assessment to quantify the vulnerability, resistance, and recovery of three major vegetation types (forest, crop, and grassland) under varying drought intensities and land degradation levels in Northeast China from 2001 to 2020. Our approach leveraged multi-source remote sensing data and meteorological reanalysis to derive the Standardized Precipitation Evapotranspiration Index (SPEI), a Land Quality Index (LQI), and Solar-Induced Chlorophyll Fluorescence (SIF) as a proxy for vegetation productivity. Drought events and degradation states were identified based on their natural occurrence and statistical thresholds within the study period. Our findings reveal that land degradation markedly amplifies drought impacts, increasing vegetation vulnerability and reducing resistance. Notably, forests, while displaying the highest resistance, were the most dependent on land quality and became the most vulnerable under coupled stresses. We also observed a trade-off between resistance and recovery across ecosystems. These results provide mechanistic insights into vegetation dynamics under coupled stresses and highlight the necessity of incorporating land quality feedbacks into vegetation models for improved predictive capability under climate change.

Drought is one of the extreme natural disasters threatening global economic development and ecosystem health. In some regions, the occurrence of droughts (or floods) may be associated with El Nino-Southern Oscillation events. Nevertheless, the propagation characteristics of El Nino-induced droughts, such as propagation probability and propagation time, remain largely unknown. To fill this knowledge gap, a framework based on copula and conditional probabilities was constructed to explore the propagation characteristics dynamics of El Nino-induced droughts in China. Specifically, propagation probabilities and propagation time of drought triggered by El Nino were identified based on conditional probability. Then, drought propagation dynamics were analyzed through multiscale sliding windows and Mann-Kendall and other trend testing methods. Finally, main drivers of propagation time dynamics were attributed based on random forest method. The results indicated that propagation probability in northwest Inland River basin (ILR), Pearl River basin, southeast Yangtze River basin, and north Yellow River basin are high, reaching up to 0.82. The propagation probability under winter El Nino stress is ranked in descending order as mild drought, severe and extreme drought, and moderate drought. Propagation time is characterized by a regional distribution. The winter El Nino signal propagates to most of northern China the following year and causing drought. The regions with an increasing trend in propagation probability are larger than those with a decreasing trend, and they are mainly distributed in northern China. This reveals that El Nino triggers an increasing trend of drought in China. Propagation time dynamics were dominated by large-scale sea-atmospheric circulation factors (i.e. Atlantic Multidecadal Oscillation, Indian Ocean Dipole, and Pacific Decadal Oscillation), especially in northwest China. In general, this study sheds new insights into large-scale drought propagation from ocean to land, which is helpful for regional drought forecasting and early warning.

Drought is considered a major contributor to carbon sink fluctuations in terrestrial ecosystems and is expected to lead to more frequent carbon sink-source transitions under future climate change. The drought threshold for carbon sink-source transition reflects the critical inflection point at which the carbon sequestration capacity of vegetation is affected by water deficit. However, the spatiotemporal patterns of the global drought threshold and their underlying mechanisms remain poorly understood. Here, we use three independent datasets from vegetation dynamics models, inversion modeling, and observational data to map and explore the drought thresholds expressed by the standardized precipitation evapotranspiration index (SPEI) during the growing season over the past four decades. Sink-source transition is indicated by changes of sign for net ecosystem productivity (NEP). The drought thresholds were identified across 66.3% of global land, with an average threshold of -1.08 ± 0.68. Regions with lower thresholds are primarily located in the Northern Hemisphere at middle and high latitudes, whereas Australia, Africa, western South America, and southern North America exhibit higher thresholds. The dominant factor influencing the spatial pattern of drought thresholds is potential evapotranspiration. Our dynamic results show that 36.4% of the thresholds increased, while 55.8% decreased. We found that disproportionate decreases in photosynthesis and respiration caused by drought in South America led to decreased thresholds and increased drought resilience in this region. Under conditions of reduced soil moisture, lower radiation, increased vapor pressure deficit, and enhanced heatwave intensity, drought in North America had a greater effect on reducing photosynthesis than it did on respiration. This resulted in an increasing threshold trend, where even relatively low levels of drought can induce a carbon sink-source transition. In addition, CO2 fertilization plays a major role in reducing thresholds and mitigating climate change. Our findings emphasize that the risk of carbon sink-source transition is more acute in regions with rising thresholds. This implies that the stability of ecosystem carbon sequestration in these regions may decrease under persistent water stress.

In order to study the influence of water network construction and water network optimization and allocation technology on the evolution of water resources system resilience in Guanzhong area of Shaanxi Province,an evaluation index system for water resources system resilience was constructed based on the complex network theory.The water resources system resilience before and after the construction of the water network in the base year and planning year of 2020 and 2035 is calculated by using the optimized allocation technology.Optimized dispatch model and optimized configuration model of the water network of the Guanzhong reservoir group were constructed combining with the Hanjiang-to-Weihe River Water Diversion Project,the Dongzhuang and Guxian reservoir water supply projects,eight major medium-to-large-scale regulating reservoirs in Guanzhong area,groundwater resources,and unconventional water sources.The model is solved by dynamic programming and genetic algorithm,and the balance between water supply and demand in the base year and planning year is analyzed.The results show that after the joint optimization and allocation of the reservoir group,the water resources system resilience in Guanzhong area has been improved except for Yangling,with the overall resilience improving from 31.11 to 32.22,an increase of 3.57％;the resilience of the water resources system in the planning year is increased from 44.05 to 44.51,a 1.04％improvement compared to the pre-optimization allocation scenario;the resilience of the water resources system increased from 31.11 to 44.05,an increase 41.59％,after the construction of the water network.

Drought significantly threatens terrestrial ecosystems health, through influencing both photosynthesis and respiratory processes. However, whether these processes have changed in response to intensified drought and the driving mechanisms remain unclear, even though the inconsistent responses may indicate an increased potential for unstable carbon sinks. The knowledge gap would hinder accurate prediction of the size of China's future terrestrial ecosystems carbon sink under increasing extreme droughts, thus impacting the realization of China's carbon neutrality goal. Here, we combined observational-based data and dynamic global vegetation model data to explore the response time (RT) of Gross Primary Productivity (GPP) and Ecosystem Respiration (TER) to meteorological drought in China and their dynamics over the past 40 years. Results reveal consistent spatial distribution patterns in GPP and TER responses to drought. During 1982-2021, widespread declines in the RT of both GPP and TER to drought were observed, indicating an increased likelihood of vegetation converting from a carbon sink into a carbon source under droughts. GPP responds slightly faster than TER, notably in arid regions influenced by land cover change and climate change. Hotspots of decreasing RT trends, such as the Tibetan Plateau and Yellow River Basin, underscore the diverse impacts of climate and land cover changes. Our findings shed new insights into ecosystem carbon fluxes mechanisms, thus providing accurate carbon budget for China's carbon neutrality goal.

Based on 0.1° × 0.1° soil moisture reanalysis data from 1950 to 2024, combined with remote sensing ecological products such as Enhanced Vegetation Index (EVI) and gross primary productivity (GPP), this study systematically investigates the spatiotemporal evolution, transition process, and ecological responses of flash droughts and slowly evolving droughts (including seasonal and cross-seasonal droughts) in the Dongjiang River Basin of China. The results indicate the following: (1) The average occurrence frequencies of flash droughts, seasonal droughts, and cross-seasonal droughts within the basin were 4.1%, 7.8%, and 8.4%, respectively. (2) The vast majority of flash droughts (approximately 90.1%) further developed into longer-lasting, slowly evolving droughts, indicating that flash droughts serve as a critical precursor to persistent drought events. Moreover, winter was identified as the key season for the occurrence of flash droughts and their transition to slowly evolving droughts. (3) In terms of ecological response, droughts significantly suppressed vegetation growth, but ecosystem resilience exhibited notable differences: although flash droughts caused relatively mild initial suppression, they were accompanied by a severe lack of ecosystem resilience; in contrast, cross-seasonal droughts, despite inducing stronger suppression, were met with higher ecosystem resilience. This study underscores the importance of the early monitoring and warning of flash droughts, and the findings provide a scientific basis for drought risk management in humid basins.

The efficient coupling of a water allocation model with a reservoir operation model is key to maximizing the benefits of inter-basin water transfer (IBWT) projects. Most existing studies of IBWT operations and management do not account for matching water transfer and demand, and instead seek to optimize either the water allocation model or the reservoir scheduling model. In this paper, we propose a multi-stage joint optimization framework for reservoir scheduling and water allocation models that considers the key operation indicators of the project in the water source area and the multi-dimensional objectives in the receiving area. We derived a functional relationship that balances considerations regarding water, the economy, and ecology (WEE) in the receiving area and applied it to the water allocation model. The reservoir operation model was developed by considering the multiple benefits of water transfer and power generation in the water resource area. To reduce the complexity of the model and the tightness of its coupling, the model was divided into multiple stages of optimization and solved using the Multi-objective Evolutionary Algorithm Based on Decomposition (MOEA/D) algorithm. The results showed the allocation schemes of water resources influenced the competition between the economic and ecological benefits. Using the WEE model, a more reasonable water transfer and allocation method can be selected according to different development needs. As such, the water shortage rate of each user in the receiving area is kept below 8%, and there is significant improvement in all indicators compared with the conventional scheme. In addition, this model enables the reservoir group to achieve a power generation capacity of more than 5.64 kWh with the goal of meeting the water transfer demand, maintaining the reservoir empty rate at about 10%, and reducing the risk of water shortages. Our study can serve as a reference for effectively addressing the challenges associated with IBWT projects, and it can inform socioeconomic development in receiving areas while optimizing water transmission and allocation.

Extreme drought and flood events, as well as their combined events, pose significant challenges to global sustainable socio-economic development and ecological health. However, the impact of dry–wet combination events (DWCEs) on vegetation vulnerability remains to be investigated. The Loess Plateau (LP) was selected as the study area to explore the response time of vegetation to precipitation index changes by optimal correlation coefficient; then, the impact of different DWCEs on vegetation vulnerability under moderate and severe scenarios was analyzed; finally, a vegetation loss probability model was constructed based on the copula function and Bayesian framework, to quantify the vegetation loss probability under DWCEs stress. The results indicate that: (1) normalized difference vegetation index (NDVI) shows an upward trend in spring, summer, and autumn, with the proportion of areas are 90.5%, 86.2%, and 95.4%, respectively, and show an insignificant trend in winter; (2) the response time of vegetation to precipitation index changes tends to be one or two seasons; (3) moderate scenarios have more influence than severe scenarios, dry-to-wet events (DWEs), wet-to-dry events (WDE) and continuous dry events (CDE) in spring-summer have a significant impact on summer vegetation of Ningxia and Shanxi, and WDE and CDE have a higher impact on autumn vegetation. (4) in terms of the probability of vegetation loss, DWE, and CDE cause higher losses to summer vegetation, while WDE and CDE cause higher losses to autumn vegetation. This study quantifies the impact of adjacent seasonal DWCE stress on future vegetation vulnerability.

Soil temperature directly affects the germination of seeds and the growth of crops. In order to accurately predict soil temperature, this study used RF and MLP to simulate shallow soil temperature, and then the shallow soil temperature with the best simulation effect will be used to predict the deep soil temperature. The models were forced by combinations of environmental factors, including daily air temperature (Tair), water vapor pressure (Pw), net radiation (Rn), and soil moisture (VWC), which were observed in the Hejiashan watershed on the Loess Plateau in China. The results showed that the accuracy of the model for predicting deep soil temperature proposed in this paper is higher than that of directly using environmental factors to predict deep soil temperature. In testing data, the range of MAE was 1.158–1.610 °C, the range of RMSE was 1.449–2.088 °C, the range of R2 was 0.665–0.928, and the range of KGE was 0.708–0.885 at different depths. The study not only provides a critical reference for predicting soil temperature but also helps people to better carry out agricultural production activities.