Drought is one of the key factors to constrain the stability of the agricultural yield. The loss of crop yield can also influence food security. Therefore, it is particularly important to determine the yield loss under drought stress. This study aims to explore the yield response of its main crops (summer maize and winter wheat) to drought stress. The research area was selected as the drought-prone Jinghui Canal Irrigation District in the Weihe River Basin. Firstly, the one-factor-at-a-time approach was used to screen out the parameters with higher sensitivity in the AquaCrop-OSPy model. Subsequently, the sensitive parameters were used as the decision variables in order to minimize the difference between simulated and statistical yields for each crop. The particle swarm algorithm was used to calibrate the sensitive parameters of the AquaCrop-OSPy model. Then, the calibrated model was used to simulate the yield and irrigation process of two crops under ideal meteorological conditions without drought stress. After that, the Latin Hypercube Sampling was employed to randomly generate multiple sets of the irrigation discount coefficients within the interval [0,1]. These coefficients were then used to proportionally scale the drought-free irrigation. Thereby, the irrigation scenarios were constructed with the varying levels of drought stress. The crop yield response was also simulated under each scenario. Finally, an S-shaped growth curve was fitted to the relationship between water deficit and yield reduction. An analysis of variance was applied to quantify the monthly variations in the water deficit effects on the crop yield. The results showed that the high value was found in the relative sensitivities of the maximum canopy cover (CCx), crop coefficients after canopy formation and before senescence (Kcb), normalized water productivity (WP), the reference harvest index (HI0), and base temperature (Tbase). The relative errors between the simulated and statistical yields of the summer maize and winter wheat in the calibration and validation periods were calculated to be within reasonable ranges, with the root mean squared error of 0.34 and 0.56 t/hm², and the normalized root mean square error of 4.91% and 8.82%, respectively. The calibration was qualified to verify the model. The CCx, Kcb, WP, HI0, and Tbase of the summer maize were 0.99, 0.87, 30.00, 0.30, and 5.00, respectively, while those of winter wheat were 0.70, 1.20, 18.97, 0.50, and 0.13, respectively. The simulation analysis showed that the S-shaped curve effectively represented the response characteristics of the agricultural yields to the drought stress in three stages and multiple inflection points. When the total water deficit rates during the reproductive period of summer maize and winter wheat were 23% and 25%, respectively, the yield reduction rates of the two crops began to increase; When the water deficit rates were 32% and 35%, respectively, the increase rates of yield reduction reached the maximum; And when the water deficit rates were 41% and 45%, respectively, the yield reduction rates tended to the extreme value. In terms of the individual months, the water shortage rates in June and November had the most significant effects on the summer maize and winter wheat, respectively. Once the water shortage rates in June and November were greater than 75%, there was an extremely high possibility in the yield reduction rates of the two crops. The most significant cumulative effect of the water shortage rate in June and July was on the yield reduction rate of summer maize, whereas November and April were on the winter wheat. Therefore, there was a nonlinear response of the crop yield to the drought stress. The key water-sensitive period was quantified to provide the theoretical support for the dynamic water allocation and precise drought in irrigation areas.

Frequent and severe drought events have posed the serious threat to vegetation growth and ecosystem stability. The underlying dynamics of vegetation response to droughts can be expected to predict and mitigate the impacts of these environmental stresses on ecosystems. However, a single ecological index cannot fully meet the requirements of the vegetation responses to droughts. In this study, a robust conditional probability model was constructed to combine the normalized difference vegetation index (NDVI), gross primary productivity (GPP), and the standardized precipitation and evapotranspiration index (SPEI). The Copula framework was designed to clarify the relationships between meteorological drought stress and the probabilities of vegetation productivity and canopy structure loss. Additionally, a systematic investigation was also made to explore the interactions between different vegetation growth statuses and environmental factors, as well as the driving forces behind drought trigger thresholds using partial least squares path modeling (PLS-PM). A case study was also carried out in the Yangtze River Basin (YRB) from 1982 to 2018. The temporal and spatial disparities were observed in the vegetation productivity and canopy structure response to the meteorological droughts. The results indicate that: 1) The proportion of areas where vegetation productivity and canopy structure in the YRB are positively correlated with SPEI is 50.61% and 63.04%, respectively.. Notably, the canopy structure exhibited a closer association with the SPEI, compared with the vegetation productivity. There was a shorter response time to the meteorological drought stress. The canopy structure shared the higher vulnerability under drought duress, compared with the vegetation productivity. 2) Spatially, the high vulnerability was found in regions, such as the Hanjiang Basin and the middle and lower reaches of the Yangtze River, in terms of vegetation productivity. While the increasing susceptibility of canopy structure was displayed in the Jinsha River Basin and the mainstream of the Yangtze River. 3) The average probabilities of loss were 20.04%, 23.63%, and 28.99%, respectively, for vegetation productivity under mild, moderate, and severe drought stress. The heightened vulnerability was predominantly concentrated in the Hanjiang Basin and the middle and lower reaches of the Yangtze River. Canopy structures shared the average probabilities of loss of 23.26%, 24.77%, and 26.86% under similar drought stress levels. Among them, the regions of elevated vulnerability were located primarily in the Jinsha River Basin. 4) There were starkly pronounced disparities of vulnerability among various vegetation types over the basin. Irrigated cropland and evergreen broadleaf forests showed higher-than-average probabilities of productivity loss (21.05% and 17.26%, respectively). While the grasslands and evergreen needleleaf forests demonstrated elevated probabilities of canopy structure loss (36.35% and 35.73%, respectively). Furthermore, the wetter regions exhibited higher average probabilities of vegetation productivity loss under varying intensities of drought stress. While the drier regions showed the heightened average probabilities of canopy structure loss. 5) Canopy structure was more susceptible to external disturbances. While the vegetation productivity was dominated by the soil moisture and its own canopy structure. There was a complex interplay between vegetation responses to drought and environmental factors in the YRB. The findings can provide invaluable insights for the effective management and conservation of ecosystems against droughts.