Strong Agricultural Resilience to 2022 Southern China Drought

Meteorological drought, especially when influenced by human activities, significantly impacts agriculture. We assessed the Yangtze River Basin (YRB)’s crop ecosystem resilience during the 2022 southern China drought. Using the elasticity framework, we quantified crop Leaf Area Index (LAI) changes with meteorological factors and human activities (irrigation) in the YRB. Our study covered maize, wheat, early and late‐season rice, evaluating model accuracy with and without irrigation. Results indicate high accuracy (R2 > 0.8, p < 0.05) in simulating LAI changes, notably improved with irrigation. Irrigation emerged as the primary driver of LAI changes in the past two decades, except for solar radiation’s impact on maize. In 2022, irrigation crucially influenced LAI changes, particularly in rice varieties. Early‐season rice and late‐season rice saw over 40% contribution from irrigation, while maize and wheat responded mainly to precipitation and radiation. Overall, 2022 yields remained stable, with late‐season rice showing an apparent increase. Drought timing correlated with maize, wheat, and early‐season rice yields, consistent with natural patterns. Conversely, late‐season rice exhibited the opposite pattern due to artificial irrigation, impacting 58% of the growing areas in 2022. The study offers insights into investigating farmland ecosystems responding to extreme droughts.


Introduction
Drought, a longstanding natural disaster, continues to impact various aspects of agricultural systems (Barriopedro et al., 2012;Dai et al., 2020;Shen et al., 2019).In the era of global warming, the frequency and severity of droughts have notably increased (Kumar et al., 2020;Liu et al., 2023;Yao et al., 2022).In China, droughts have traditionally afflicted northern regions, but recent years have witnessed an escalating trend in their occurrence in the southern regions as well (Mokhtar et al., 2022;Zhang et al., 2020).Notably, the Southern Drought of 2022 is considered the worst drought in the Yangtze River Basin (YRB) since 1961 (Zhang, 2022).Amidst these changing conditions, the rising influence of human interventions, especially advancements in irrigation technology, has substantially shaped agroecosystems, with both natural factors and human activities now playing pivotal roles (Hou et al., 2022;Kath et al., 2019).This convergence has led to a complex interplay of factors influencing the resilience of agroecosystems (Li et al., 2023).Currently, the impact of increasing human interventions on agriculture systems remains unclear (Qu et al., 2020).Thus, the primary objective of this paper is to explore the characterization of agricultural system resilience in the face of severe drought occurrences, examining the roles played by both climatic factors and human activities in shaping their impact.
The YRB stands as a pivotal grain-producing region in China, covering a quarter of the nation's arable land (Guo et al., 2021).It significantly contributes to the country's agricultural output, representing 40% of the total agricultural value and 40% of the nation's overall grain output, with rice production playing a dominant role at 70% (Li et al., 2020;Qu et al., 2020).The 2022 drought, commencing in June, intensifying through July, and extending until November, coincided with a critical growth phase for double-season rice in the YRB.Early-season rice was in the fruiting and harvesting stage, while late-season rice was in its crucial reproductive phase, demanding ample water supply.As a result, the impact of this drought on southern rice production was particularly pronounced.In response to this challenge, various regional governments in the south implemented a series of measures to ensure a successful autumn grain harvest, including early warning and forecasting systems, strategies for water diversion and conservation, and emergency irrigation efforts (Zhang, 2022).
Presently, while the primary focus of ecosystem resilience research centers on natural vegetation ecosystems (Chen et al., 2020;Dubovyk et al., 2016;Liu et al., 2023;Qu et al., 2020), specific analyses of resilience within ecosystems such as grasslands (Hou et al., 2022;Kath et al., 2019;Liu et al., 2021;Zheng et al., 2019) and forests (Jiménez-Muñoz et al., 2016) are well-documented.In contrast, research on the resilience of farmland ecosystems remains relatively understudied, with inadequate consideration of the interplay between anthropogenic and climatic factors influencing farmland vegetation (Qu et al., 2020).Particularly in the scenario of extreme drought, it remains unclear to what extent anthropogenic factors contribute.
Hence, there is a compelling need to bolster research on agricultural resilience, deepening our understanding of the impacts of climate and anthropogenic factors on agriculture, especially in the context of severe droughts (Qu et al., 2020).With the wealth of advanced remote sensing data, remote-sensing vegetation products have gained prominence in extensive research concerning agriculture's response to climate (Allen et al., 2006;Habib-ur-Rahman et al., 2022;Hou et al., 2022;Liu et al., 2023;Wei et al., 2013).These products encompass the Leaf Area Index (LAI), Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index, and more.Among these, NDVI and LAI are widely utilized, though NDVI presents challenges related to saturation for which no ideal solution has been found.LAI stands out as an effective and crucial indicator for monitoring crop growth (Liang et al., 2015;Yao et al., 2008), assessing crop health, predicting potential yields, addressing agricultural disasters, and providing decision support, making it of significant value in agricultural research and analysis (Liu et al., 2018;Luo et al., 2020;Naeem et al., 2023).
In this study, we endeavor to enhance research on agricultural resilience by delving into the factors that influence crop LAI variations in the YRB in 2022.Our approach encompasses two decades of observed LAI data and the simulation of 20 years of irrigation data based on the water balance equation.We employ the elasticity coefficient model to account for both natural factors and anthropogenic irrigation in shaping farmland LAI changes in the YRB.Moreover, our analysis extends to the LAI resilience of different crops during the severe 2022 drought in the southern region.Combining this analysis with the PML_V2 biophysical model simulation, we derive crop yield results for the same time period in the YRB and elucidate the variations in crop yield for 2022.This is accomplished by exploring the relationship between yield levels and the timing of drought occurrence, considering the resilience attributes of farmland in the YRB.

Study Area
The YRB, depicted in Figure 1, plays a pivotal role in Chinese agriculture.This region boasts extensive fertile farmlands, making a substantial contribution to the nation's grain production, including key crops such as rice, wheat, and others.The YRB predominantly lies in a subtropical monsoon zone characterized by a warm and humid climate, well-defined seasons, a high annual accumulated temperature, and an extended growing period for crops.Many areas in this region experience synchronous rainfall and warmth throughout the growing season, creating optimal conditions for agricultural production.
The harvest area is extracted from the 2019 harvest area data set generated by Luo et al. (2020).The YRB encompasses over 24.6 million hectares of arable land, as illustrated in Figure 1b, constituting a quarter of the nation's total arable land area.This region's agricultural production value contributes significantly, accounting for 40% of the national total, with grain production also making up 40% of the national total.Notably, rice production alone represents 70% of the nation's total.
The extreme drought event that struck the YRB in June 2022 and persisted until November stands as the most severe drought since 1961 see Figure S3 in Supporting Information S1.This drought evolved rapidly, impacting 9.67 million acres of farmland across six provinces and direct-controlled municipalities, including Anhui, Jiangxi, Hubei, Hunan, Chongqing, and Sichuan.It also had repercussions on the water supply of 830,000 people due to the drought.

Crop Data
In this research, we utilized the LAI as the primary indicator for assessing crop growth and estimating the influence of both climatic factors and human activities on crop performance.The LAI data were sourced from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor, specifically the MOD15A2H.061product (Dokoohaki et al., 2022).This product compiles composite data at 8-day intervals and offers a spatial resolution of 500 m.
To ensure consistency in our analysis, we obtained annual average LAI data with 0.1°resolution spanning from 2003 to 2022.We acquired this data using the Google Earth Engine platform.Notably, the algorithm employed selects the most representative pixel from all Terra sensor acquisitions within each 8-day period.

Crop Phenology Data
This study incorporated phenology and harvest area data sets for three pivotal crops in China, specifically maize, wheat, and rice.These data sets were developed by a research team at Beijing Normal University and encompass the time frame from 2000 to 2019 (Luo et al., 2020).For the years spanning from 2020 to 2022, the harvest area data were extrapolated from the 2019 data set.Additionally, the phenology data for 2020-2022 years were derived using a method that involves smoothed long-term time series of LAI data, coupled with the application of the double logistic function and dynamic threshold method.All the processes are conducted on the software Timesat 3.3 based on Matlab, the details of the method are available in the publication describing TIMESAT (Eklundh & Jönsson, 2016).

Yield Data
In the discussion section of this article, we closely examined the spatial distribution characteristics of the 2022 yield results within the YRB.The yield calculations were carried out using the following formula: where k represents the quantity of Gross Primary Productivity products computed during the growing season, and the value of k depends on the length of the growing season.Since our model is set at an 8-day scale, for example, if the length of the crop growing season is 116 days, the value of k is calculated as 15 (rounded up from 116/8).This calculation is performed using the PML_V2 model (Gan et al., 2018;Zhang et al., 2019).HI represents the harvest index for maize, wheat, and rice, which is constructed using data extracted from Chinese Statistical Yearbook (https://data.stats.gov.cn/)The yields obtained by this method have been verified to be credible (Li et al., 2023).

Climate Data
The China Meteorological Forcing Dataset (CMFD) stands as the inaugural high-resolution meteorological driving data set specially designed for land process research within China, as detailed by He et al. (2020).The development of this gridded data set involved an amalgamation of remote sensing products, reanalysis data, and in-situ observations.Encompassing the land areas of China, it spans a comprehensive 40-year period from 1979 to 2019.This data set offers a spatial resolution of 0.1°and a temporal resolution of 3 hr, making it a valuable resource for research.It comprises seven essential meteorological variables, including near-surface temperature, atmospheric pressure, specific humidity, total wind speed, downward shortwave radiation flux, downward longwave radiation flux, and precipitation rate.
In this study, we used the Data Cube correction method, as outlined by He et al. (2022), and integrated data from the Global Land Data Assimilation System (GLDAS) to augment and extend the coverage of CMFD data, encompassing the years 2019-2022.This augmentation not only extends the temporal coverage but also enhances the data set's consistency, rendering CMFD one of the most dependable meteorological data products in China (He et al., 2022;Naeem et al., 2023).
For our analysis, we focused on five of the seven meteorological variables provided by CMFD, specifically, precipitation (Pr), specific humidity (Sh), solar radiation (Sa), and wind speed (Wi).

Root Zone Soil Moisture (GLDAS)
The GLDAS, as detailed in the project goals, represents a collaborative initiative involving the Goddard Space Flight Center of the National Aeronautics and Space Administration and the National Centers for Environmental Prediction of the National Oceanic and Atmospheric Administration.GLDAS leverages data assimilation techniques to amalgamate ground observations and satellite remote sensing data, enabling the driving of land surface models (Rodell et al., 2004).The data produced by this system has evolved into a fundamental resource for research in the domains of global change and the hydrological cycle (Wei et al., 2013).
In our study, we harnessed GLDAS data spanning from 2003 to 2022 to rectify CMFD data.Furthermore, we integrated soil root zone data into our calculations to determine irrigation amounts, enhancing the robustness of our analysis.According to the validation study conducted by Tian and Zhang (2023), root zone soil moisture from GLDAS-NOAH 2.1 was proved to have relatively good accuracy over China.

Data Processing
In this study, all the data were synthesized at 8-day intervals, by following MODIS LAI data products, and the spatial resolution of all the data was resampled to 0.1°to align with meteorological data.

Elasticity Model
In the construction of our model, we drew inspiration from the concept of elasticity coefficients commonly used in economics.These coefficients serve to quantify the elastic response of LAI to various influencing factors.Essentially, they indicate the extent of LAI variation when a specific driving factor undergoes a 1% change (Zhang et al., 2023).This approach enables us to evaluate the individual contributions of different driving factors to LAI.The model framework can be succinctly represented as follows: where N represents the number of driving factors, and x N represents a specific driving factor, and r is the model residual.In this study, the differencing was performed over time window (8-day by 8-day).At the 8-day scale, the differences between LAI and other meteorological factors introduced by human activities are small.Equation 2 can be further expressed as: When we replace the absolute change with the deviation from the mean value (∆x n ) in Equation 3, we derive: where the annual average of each variable is the average for the period of 2003-2022.Similarly, by calculating the annual change rates for all driving factors, Equation 4 is further deduced as: The elasticity coefficient is expressed as Formula 6: εx n represents the elasticity coefficients for x n , ∆LAI represents the change in LAI, and ∆x n represents the change in variable x n .Combining Equations 5 and 6, we derive: Earth's Future To calculate the contribution of different driving factors to LAI changes dLAI x i , the variation in LAI caused by different driving factors was computed by where dLAI x i is the amount of LAI change induced by a specific driving factor, calculated using Equation 9: where x n is the average value of the driving factor from 2003 to 2022.

Irrigation Calculation
Irrigation data used in the elasticity model is calculated using the water balance equation, with the formula is expressed as where S i is the soil water content in i th 8-day, likewise, the P i , ET i , R i , and IRRI i denotes the precipitation, the evapotranspiration, the runoff and the irrigation (conceptualized) in the same time period, where ET is derived from the PML_V2 model.Considering that farmland terrain is generally relatively flat, the influence of runoff was neglected in the irrigation calculations.Therefore, the irrigation calculation can be reformulated as: Additionally, to make mathematical balance, Equation 11 is further partitioned into: We conducted a comparison between the computed irrigation results and the data from the water resources bulletin of YRB (Table S1).The findings reveal that the simulated irrigation amounts exhibit differing degrees of deviation in various years.Importantly, these errors remain within an acceptable range, rendering them suitable for modeling purposes.Notably, the simulated irrigation amounts for the year 2022 closely match the records found in the Water Resources Bulletin (http://www.stats.gov.cn/sj/).Additionally, we calculated the model' R 2 for all grid points in the YRB.It can be observed that the R 2 of the model with irrigation added is higher than that of the model without irrigation (Figure S1 in Supporting Information S1).Specifically, for maize, the R 2 values are mostly above 0.6, with a median value around 0.7; for wheat, the R 2 values are mostly above 0.4, with a median value around 0.55; for early season rice, the R 2 values are all above 0.4, with a median value around 0.60; and for late-season rice, the R 2 values are all above 0.5, with a median value around 0.65.

Model Performance
Furthermore, we analyzed the spatial distribution of model-simulated and MODIS-observed changes in crop LAI in 2022 (Figure 3).For different crops, the spatial distribution characteristics of the simulated LAI changes were found to be quite consistent with the spatial distribution characteristics of the observed LAI changes.In certain regions, it was observed that the model-simulated LAI changes were smaller than the observed LAI changes.In detail, the observed variations in LAI within the lower reaches of the early season rice planting region exceed the model-estimated changes in LAI.Conversely, the spatial distribution of late-season rice exhibits minimal differences, with only a minor portion of the model underestimating LAI changes in the southern part of the planting area.When considering maize, LAI changes are more pronounced compared to the other three crops, and the results estimated by the model is closely with the observed results in terms of spatial distribution.In fact, the model's estimates and the observed outcomes exhibit a higher level of spatial consistency.Wheat cultivation areas are relatively sparse, primarily concentrated in a small northern region within the middle and lower reaches of the YRB, and the model's estimates also closely correspond with the observed results.

The Relative Contribution Rates of Different Factors for Crop LAI Changes
The amount of contribution of several factors to the change in LAI of different crops was calculated according to Equations 7 and 8, which in turn calculated the relative contribution rates of different factors to crop LAI changes from 2003 to 2022.For rice, it can be observed that the relative contribution rate of irrigation shows an increasing trend.For rain-fed crops such as maize and wheat, the contribution of irrigation exhibits significant fluctuations between years (Figure S2 in Supporting Information S1).

Spatial Distribution Patterns of Main Drivers Contributing Crop LAI Changes in 2022
Based on the relative contribution of climate factors and irrigation factors to the LAI changes of different crops, the main contributing factors to the LAI changes of different crops in the YRB in 2022 were calculated (Figure 5).Overall, the main driver of LAI change was wind speed with the least share, but the main drivers of LAI change with the largest share varied.While the main driver of LAI change for rice in the middle and upper reaches of the YRB was irrigation And the main driver of LAI change was solar radiation and precipitation for maize and wheat respectively, In addition, in terms of spatial distribution characteristics, the area where irrigation was the main driver for rice was mainly concentrated in the middle and lower reaches of the YRB, while the spatial distribution of the main driver of LAI change for dry crops did not differ much.
Specifically, for early-season rice, the main driver of LAI change was irrigation, accounting for 52%, followed by specific humidity and radiation accounting for 20% and 17%, respectively, with precipitation and wind speed accounting for less, and wind speed accounting for a minimum of 2%.Similarly, for late-season rice, irrigation was also the main driver of LAI variation with the largest share of 58%, while the main drivers were radiation, specific humidity, and precipitation with the shares of 16%, 12%, and 10%, respectively.The largest proportion of the main drivers of LAI change for maize is solar radiation, with a proportion of 42%-56%, followed by irrigation, precipitation, and wetness, with a proportion of 28%, 20%, and 9%, respectively.The main driver of wheat LAI change was precipitation with the highest percentage of 56%, followed by radiation, irrigation, and wetness with 17%, 14%, and 13%, respectively.

Climatic Factors and Human Activities Driving Crop LAI Changes in YRB (Climatic Factors and Human Activities Driving Agricultural Resilience in YRB)
As shown in Result section, climatic factors and human activities together drove crop agricultural resilience in YRB.First, climate factors and irrigation (human factors) jointly influence the variation in crop LAI in the YRB, with irrigation making a significant contribution.This conclusion aligns with previous research findings (Li & Troy, 2018;Qu et al., 2018;Troy et al., 2015).
To prove the point, due to the highly significant positive correlation between crop LAI and yield (Zhu et al., 2021), variations in LAI can to some extent be reflected in yield anomalies.Therefore, the yield results from 2003 to 2022 were simulated using the PML_V2 model, it was found that in 2022, the extreme-drought scenario, the yield in the YRB did not decrease (Figure S2 in Supporting Information S1), but rather increased, compared to previous years.This is a side effect of the fact that changes in yields are not only influenced by strong climatic conditions but also by strong human activities.Therefore, we calculated the disparity between the 2022 yields of different crops and the multi-year average yields spanning from 2003 to 2022, subsequently generating a spatial distribution plot to illustrate the findings (Figure 6), and it was found that maize, wheat and early season rice all had different degrees of yield reduction and increase, and overall there was a presentation of yield increase in the middle and upper reaches, and yield reduction in the middle and lower reaches, This spatial characterization is largely related to natural climatic conditions, on the one hand, due to the fact that the development of the drought in 2022 was spreading from the middle and upper reaches to the lower reaches and broke out mainly in the middle and lower reaches of the river, and on the other hand, the timing of the planting of the crop and the concentration of the drought (the middle of the year,and in which the maize and wheat probably has been harvested,and is the timing of the alternation between early and late season rice) may also be the reason for this spatial variability in the yield of the crop.But the late-season rice, however, showed an overall increase in the performance of the production.This may be related to the fact that the main driver of LAI change in 2022 for late-season rice is irrigation, as we mentioned earlier.That is, in the extreme drought of 2022, both climate and human activities jointly drove changes in crop growth.

Irrigation Is the Primary Driver for the Increase in Rice Production in 2022
To further investigate the evidence of irrigation playing a dominant role in the increase in late-season rice production in 2022, we explored the relationship between the timing of the 2022 drought event and yield anomalies in different crops, conducting an in-depth analysis of the key factors driving these outcomes.PDSI (Palmer Drought Severity Index) is widely used in monitoring agricultural droughts and can be used to indicate the occurrence of specific drought events (Palmer, 1965;Zhao et al., 2017).Therefore, we calculated the timing of extreme drought (PDSI < 4) occurrence using the PDSI drought index, and the specific calculation method is shown in the figure (Figure S3 in Supporting Information S1).In addition, due to the absence of phenological data for the years 2019-2022, we utilized and analyzed the time series of LAI to determine the planting and harvesting periods of crops in 2022 (Figure S4 in Supporting Information S1).We calculated the differences between these periods and the timing of drought occurrence and examined their relationship with yield anomalies (Figure 7).
Results from Figure 7a revealed a consistent conclusion for early season rice, maize, and wheat, that is, earlier planting before the drought onset led to higher yields, aligning with natural patterns.However, the results for lateseason rice were the opposite.From Figure 7b, we observed that the farther the harvesting period from the timing of drought occurrence for early season rice and wheat, the higher the yields, which is consistent with natural patterns.Yet, late-season rice and maize exhibited contrasting results.Upon analyzing these findings, it becomes evident that the results for late-season rice contradict natural patterns from two perspectives (differences between the planting, harvesting periods and the timing of drought occurrence).This is likely due to human intervention (irrigation for crop planting) occurring promptly after the onset of drought.Supported by the results of 2022, lateseason rice in over 58% of areas experienced changes in LAI dominated by irrigation, indirectly affirming that the increased yield of late-season rice in 2022 was the result of human intervention.Similarly, these studies have also highlighted the crucial role of irrigation in crop growth, with crop growth depending to some extent on artificial irrigation (Liu et al., 2009;Shen et al., 2013).Specifically, for maize and wheat, irrigation has a greater impact on wheat compared to maize.However, the contribution of irrigation to crop yield varies significantly across different regions, which is consistent with the finding in Wang et al. (2021).On the other hand, for rice, irrigation is the primary driver of rice growth (Li et al., 2015).Against the backdrop of large-scale climate change, human activities such as irrigation, vegetation conservation and restoration, land development, and grazing can influence vegetation growth within relatively small spatial scales (Feng & Fu, 2013;Hou et al., 2022;Hussain et al., 2020;Wang et al., 2021).

Limitations and Uncertainties
In this study, we established a responsive relationship between various crops and both climate and irrigation factors utilizing the elasticity coefficient model.The model's R 2 , along with the goodness of fit demonstrated by the estimated and observed data on scatter plots, attests to a certain degree of reliability which is sufficient to conduct an analysis of the relative contributions of different driving factors.However, there are still some limitations and uncertainties in this study, for instance, although this study has categorized crops into different types and considered specific crop growing seasons, it does not fully account for human activity factors.Only irrigation factors were considered, while factors such as cropping systems and fertilization practices in field management, which can significantly impact crop growth (Hussain et al., 2020;Jiang et al., 2017;Li et al., 2015), were not fully considered.In addition, in the scenario analysis depicted in Figure 7, we investigated the correlation between crop yield and the temporal gap between sowing and harvesting, along with the timing of drought occurrence.The objective was to illustrate how the perceived factor of irrigation plays a pivotal role in counteracting the impacts of natural factors, thereby ensuring stable or potentially increased yields.However, a degree of uncertainty exists in this context.Beyond the human-induced irrigation factor, it is possible that the presence of drought-resistant crop varieties might also contribute to mitigating the adverse effects of drought on yields.This additional variable introduces a layer of ambiguity into the results, as the interplay between irrigation practices and crop resilience may vary, adding complexity to our understanding of the relationship.It is not perfect to simply attribute yield spacing to drought, but it is feasible to utilize the correlation that yield exhibits with drought to side-step the main drivers of crop variability in this study.
Despite these limitations and uncertainties, this study provides valuable insights into the contributions of environmental and human factors on crop growth.The findings underscore the importance of considering both natural and human-induced influences when assessing agricultural outcomes.These results can inform more effective strategies for managing crops in the face of changing climatic conditions and evolving agricultural practices.Furthermore, they emphasize the need for further research to refine our understanding of these intricate relationships and reduce uncertainties, ultimately aiding in the development of more robust and resilient agricultural systems.By establishing an elasticity coefficient model to analyze the contributions of climate factors and irrigation to crop LAI, it is possible to understand the primary influences of both natural and human factors on crop growth.This can assist farmers and ecological managers in more effectively addressing key factors, adapting to climate change, and improving agricultural production by adjusting planting strategies.

Conclusions
In this study, we analyzed the resilience of agriculture in the YRB during the 2022 megadrought based on the elasticity coefficient model and crop yield under different scenarios.The results showed that various crops exhibited strong resilience characteristics in different scenarios of the 2022 megadrought, with overall human activity interference being the primary contributing factor to this resilience.Specifically, we utilized an elasticity model to identify the factors influencing crop LAI changes by modeling the elasticity of various climate factors, irrigation, and observed LAI variations in the YRB.Notably, the dominant drivers of LAI changes differ among various crops, highlighting the unique dynamics of each.Nevertheless, our multi-year analysis underscores the growing significance of irrigation as a pivotal factor driving crop LAI changes, particularly for rice.A noteworthy revelation emerged when examining crop yield data for the extraordinary drought year of 2022.Despite the YRB experiencing its most severe drought since 1961, there was no substantial decline in crop yields.Remarkably, many regions cultivating late-season rice even saw an increase in their yields.These intriguing results, when considered alongside our investigations into the primary drivers of late-season rice LAI changes in 2022, strongly suggest that the upturn in late-season rice yields that year can largely be attributed to deliberate human irrigation interventions.
By scrutinizing the attribution of crop LAI changes and discussing crop yields in 2022, our study underscores the evolving and increasingly crucial role of human intervention in the agricultural production process.This finding holds significant implications for the sustainable development of agriculture in the face of future challenges.
Due to the complex influences of both natural and anthropogenic factors on agricultural resilience and the limitations of this study, future research endeavors should center on a deeper exploration of how major crops adapt to climate change is crucial.This involves studying the responsiveness of different crop varieties to temperature fluctuations, altered precipitation patterns, and overall climate variability.A holistic understanding of the interaction between natural climate conditions, anthropogenic influences, and agricultural resilience is essential for shaping the future of sustainable agriculture.

Figure 1 .
Figure 1.Study area; (a) Location of study area; (b) Harvest area in Yangtze River Basin.

Figure 2
Figure2shows a comparison between the LAI changes estimated by the model and the observed LAI changes by MODIS.Within this visual representation, the model's consideration of irrigation is denoted by the blue color, whereas the yellow color signifies the model's calculations without irrigation while keeping all other factors consistent.The results distinctly indicate that the inclusion of irrigation in the elasticity coefficient model significantly enhances the goodness of fit (R 2 ) between the simulated LAI changes and the observed LAI changes across the four different crops.Moreover, the model exhibits impressive performance in aligning with the

Figure 2 .
Figure 2. Scatterplot between observed Leaf Area Index change and the estimated by the elasticity model with (without) irrigation.

Figure 3 .
Figure 3. Characterization of the spatial distribution of model-simulated and MODIS-observed changes in crop Leaf Area Index in 2022, (a) (b) for Early season rice; (c) (d) for Late season rice; (e) (f) for Maize; (g) (k) for Wheat.

Figure 4 .
Figure 4. Mean relative contribution rate for four main crops during 2003-2022 and 2022, respectively.

Figure 5 .
Figure 5. Spatial distribution of the main drivers of Leaf Area Index change for four major crops in 2022, (a) for Early season rice; (b) for Late season rice; (c) for Maize; (d) for Wheat.

Figure 6 .
Figure 6.Spatial yield anomaly of four main crops in 2022, (a) for Early season rice; (b) for Late season rice; (c) for Maize; (d) for Wheat.

Figure 7 .
Figure 7.The relationship between yield anomaly and the timing of drought onset.