Quantitative analysis method for vegetation physiology and structure response under extreme dry-heat composite stress

Abstract

The invention provides a vegetation physiology and structure response quantitative analysis method under extreme dry-heat composite stress, and relates to the field of vegetation remote sensing, the method comprises the following steps: collecting hydro meteorological data, vegetation structure data, vegetation function data and soil moisture data; calculating a soil drought index of the soil moisture data, and determining a drought vertex; extracting an abnormal value form through a local weighted regression method; calculating an average change track of the hydro-meteorological data; establishing a random forest model; training the random forest model based on the time window with the drought vertex as the center to obtain a prediction result; according to the prediction result, obtaining a physiological signal proportion of the vegetation function data; according to the drought vertexes and the abnormal value forms, driving variables are extracted, attribution analysis is carried out, and the relative importance sequence of the driving variables to the three types of vegetation function variables is obtained. According to the invention, the physiological and structural response signals of vegetation can be accurately separated, and the monitoring precision and analysis capability of the dynamic change of vegetation are improved.

Description

Technical Field

[0001] The present application relates to the field of vegetation remote sensing, and in particular to a quantitative analysis method of the physiological and structural responses of vegetation under extreme dry-heat combined stress. Background Art

[0002] In recent years, global warming has exacerbated the occurrence of droughts and high-temperature heat waves, forming more destructive dry-heat compound extreme events, posing a serious threat to water resources, food security and ecosystems. Such events are caused by the combined effects of atmospheric high-temperature drought and soil drought, producing a synergistic effect. Dry-heat compound stress affects the physiological functions and structural characteristics of vegetation, resulting in reduced photosynthesis and transpiration efficiency, ultimately inhibiting vegetation productivity and even causing vegetation death. Although the structural changes of vegetation are more significant and the research is more in-depth, the physiological response is more hidden and rapid and has not been fully studied. In-depth exploration of the physiological response of vegetation will help understand its adaptation mechanism in extreme climate events and provide a scientific basis for clarifying the main driving factors of the decline in vegetation functions.

[0003] Under the combined stress of dryness and heat, the functional responses of vegetation mainly include three mechanisms: photosynthesis, transpiration, and water regulation. First, drought and high temperature conditions will cause stomatal closure and reduce carbon dioxide intake, which will in turn affect photosynthesis efficiency, and high temperature may damage the light system and weaken the efficiency of light energy utilization. Secondly, vegetation reduces water loss by reducing transpiration, which is closely related to stomatal closure. Finally, the water regulation mechanism helps vegetation regulate water behavior during droughts, maintain the water potential balance of leaves and roots, and reduce the risk of xylem cavitation. These three types of responses reflect the complex coping mechanisms of vegetation systems, including stomatal regulation, deepening of roots, leaf adjustment, etc., demonstrating the ability of vegetation to cope with extreme dry and hot disasters in multiple dimensions. If we only focus on a single physiological mechanism, it is difficult to fully capture the drought state of vegetation and subsequent changes in productivity. Therefore, a comprehensive assessment of vegetation photosynthesis, transpiration, and water regulation is crucial to a more accurate understanding of the impact of dry and hot combined events on vegetation.

[0004] Satellite remote sensing technology is an important tool for studying the physiological and structural responses of vegetation under combined dry-heat stress. The application of data such as sunlight-induced chlorophyll fluorescence, near-infrared reflectance index, evapotranspiration, vegetation optical thickness and leaf area index has strongly promoted the development of related research. However, current research faces the important challenge of mixing physiological and structural signals. Data such as fluorescence and evapotranspiration obtained by satellites simultaneously reflect the physiological and structural changes of vegetation, and it is difficult to separate the two signals for separate analysis. Therefore, to deeply understand the impact of combined dry-heat events on vegetation, it is necessary to improve the accurate separation of physiological and structural signals. Summary of the invention

[0005] The purpose of the present invention is to provide a quantitative analysis method for the physiological and structural responses of vegetation under extreme dry-heat combined stress in order to solve the technical problems that traditional remote sensing technology is difficult to separate physiological response signals from structural response signals in large-scale vegetation monitoring, and the model has limited ability to interpret changes in vegetation physiological functions.

[0006] The above-mentioned purpose of the present application is achieved through the following technical solutions: S1: Collect hydrological and meteorological data, vegetation structure data and vegetation function data of the study area and pre-process them; S2: Obtain soil moisture data of the study area; determine the drought peak by calculating the soil drought index of the soil moisture data; S3: Extract the outlier form of hydrometeorological data by combining hydrometeorological data with local weighted regression method; S4: Set a time window centered on the drought apex and calculate the average change trajectory of each variable of the hydrological and meteorological data during the drought period; S5: Taking hydrological and meteorological data and vegetation structure data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, a random forest model is established; based on the time window, the random forest model is trained and the prediction results of extremely drought areas are obtained; S6: According to the prediction results, the physiological signal ratios of the three types of vegetation function variables of the vegetation function data are obtained; The three types of vegetation functional variables include: photosynthesis functional variables, transpiration functional variables and water regulation functional variables; S7: Extract driving variables of drought development and drought recovery periods according to drought apex and outlier forms; driving variables include: hydrological and meteorological data, vegetation characteristics and drought duration; S8: Taking the three types of vegetation functional variables as the objects and combining them with the proportion of physiological signals, the driving variables were attribution analyzed to obtain the relative importance ranking of the driving variables to the three types of vegetation functional variables, and to complete the quantitative analysis of the physiological response and structural response of vegetation under the combined stress of extreme dryness and heat.

[0007] Optionally, step S1 includes: The hydrological and meteorological data include: temperature, rainfall, soil moisture and solar radiation data; The vegetation structure data include: leaf area index and near-infrared vegetation reflectance; The vegetation function data include: solar induced chlorophyll fluorescence, vegetation optical thickness and evapotranspiration; Solar induced chlorophyll fluorescence includes: clear sky instantaneous data and daily average data; Vegetation optical depth includes: daytime vegetation optical depth data and nighttime vegetation optical depth data.

[0008] Optionally, the specific steps of the preprocessing include: Perform quality screening and coordinate conversion on hydro-meteorological data, vegetation structure data and vegetation function data; The hydrological and meteorological data, vegetation structure data and vegetation function data are mean synthesized and resampled to unify the temporal and spatial resolutions.

[0009] Optionally, step S2 includes: S21: Using the soil moisture data of the historical preset years in the study area, the soil drought index is calculated as follows:

[0010] in Indicates Month and Annual soil moisture, i.e. soil drought index; and They are Mean and standard deviation across all years; Filter out Pixels with ,are considered as extremely drought areas; S22: The time when the minimum value of soil moisture in the extremely drought area occurs is taken as the drought peak, and the time when the drought peak occurs is obtained.

[0011] Optionally, step S3 includes: S31: Extracting long-term trends of hydrological and meteorological data using a local weighted regression method based on LOWESS; S32: Calculate the monthly average of hydrological and meteorological data to obtain the average seasonal cycle; S33: Based on the hydrological and meteorological data, the long-term change trend and the average seasonal cycle are removed to obtain the outlier form of the hydrological and meteorological data.

[0012] Optionally, step S5 includes: S51: Taking hydrological and meteorological data and vegetation structure data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, the first random forest model was established; S52: Taking hydrological and meteorological data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, the second random forest model was established; S53: training the first random forest model and the second random forest model by using the hydrological and meteorological data, the vegetation structure data, and the vegetation function data outside the time window; S54: Predicting the hydrological and meteorological data and vegetation structure data outside the time window through the trained first random forest model and the second random forest model to obtain prediction results.

[0013] Optionally, step S6 includes: Calculating the residual between the prediction results of the first random forest model and the prediction results of the second random forest model to separate the physiological signals of the vegetation function data; According to different drought development stages and wetness levels, combined with physiological signals, the physiological signal ratios of the three types of vegetation function variables in vegetation function data were statistically analyzed.

[0014] Optionally, step S7 includes: The time window before the drought peak is defined as the drought development period, and the time window after the drought peak is defined as the drought recovery period; Hydrometeorological data include: temperature, rainfall, solar radiation and soil moisture; Vegetation characteristics include: vegetation cover fraction and wetness; The drought duration was obtained by identifying the time when soil moisture recovered from abnormal values ​​to normal levels before and after the peak of drought.

[0015] Optionally, step S8 includes: The SHAP method was used to conduct attribution analysis on the driving variables, quantify the characteristic contributions of hydrological and meteorological data, vegetation characteristics and drought duration to the three types of vegetation functional variables, and obtain the relative importance ranking of the driving variables to the three types of vegetation functional variables.

[0016] An electronic device comprises a processor, a memory, a user interface and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device performs a quantitative analysis method for the physiological and structural responses of vegetation under extreme dry-heat combined stress.

[0017] A computer-readable storage medium stores instructions. When the instructions are executed, a quantitative analysis method for physiological and structural responses of vegetation under extreme dry-heat combined stress is performed.

[0018] The beneficial effects of the technical solution provided by this application are: The local weighted regression method is used to extract abnormal changes in hydrometeorological data, vegetation structure data, and vegetation function data, and the change trajectory of each index during drought is analyzed by setting a time window; random forest modeling is performed on hydrometeorological driving variables, vegetation structure variables, and vegetation function variables, and the physiological signals and structural signals of the three types of functional responses of vegetation in extreme dry and hot events are separated based on the model residual idea; the model-independent machine learning interpretation method is used to quantify the characteristic contribution of each driving variable to the photosynthesis function, transpiration function, and water regulation function of vegetation. Clarifying the physiological and structural signals of the three types of functional responses of vegetation under dry and hot stress, overcoming the difficulties of mixed physiological response signals in large-scale vegetation remote sensing monitoring, provides methodological support for improving the ability to accurately monitor and analyze dynamic changes in vegetation, and is of great significance for accurately evaluating changes in vegetation productivity and ecosystem carbon sequestration functions. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present application will be further described below with reference to the accompanying drawings and embodiments, in which: Figure 1 is a step diagram in an embodiment of the present application; Figure 2 is a map of the locations where drought events occurred in the study area in the embodiments of the present application; Figure 3 It is a time series change diagram of various hydrological and meteorological data and vegetation index data before and after extreme drought in the embodiment of the present application; Figure 4 This is a graph showing the separation results of the physiological components of the functional response of vegetation during drought in the embodiment of the present application; Figure 5 It is a zoning statistical result diagram of the overall and physiological parts of the functional response of vegetation during drought in the embodiment of the present application; Figure 6 is an attribution analysis result diagram of driving factors of physiological response of vegetation during drought in an embodiment of the present application; Figure 7 It is a schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0020] In order to have a clearer understanding of the technical features, purposes and effects of the present application, the specific implementation methods of the present application are now described in detail with reference to the accompanying drawings.

[0021] The embodiments of the present application provide a method for quantitatively analyzing the physiological and structural responses of vegetation under extreme dry-heat combined stress.

[0022] Please refer to Figure 1 , Figure 1 This is a step diagram of a quantitative analysis method of vegetation physiological and structural response under extreme dry-heat combined stress in an embodiment of the present application, comprising: S1: Collect hydrological and meteorological data, vegetation structure data and vegetation function data of the study area and pre-process them; S2: Obtain soil moisture data of the study area; determine the drought peak by calculating the soil drought index of the soil moisture data; S3: Extract the outlier form of hydrometeorological data by combining hydrometeorological data with local weighted regression method; S4: Set a time window centered on the drought apex and calculate the average change trajectory of each variable of the hydrological and meteorological data during the drought period; Specifically, the time window is set to be three months before and after the drought peak, and the drought peak is the time when the minimum soil moisture record of the pixel occurs in 4 days. By extracting extreme drought events pixel by pixel and counting the average change trend of various hydrological and meteorological data and vegetation index data during the drought, the average change trajectory of each variable during the drought is finally obtained; the vegetation index data includes: vegetation structure data and vegetation function data and preprocessing.

[0023] S5: Taking hydrological and meteorological data and vegetation structure data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, a random forest model is established; based on the time window, the random forest model is trained and the prediction results of extremely drought areas are obtained; S6: According to the prediction results, the physiological signal ratios of the three types of vegetation function variables of the vegetation function data are obtained; The three types of vegetation functional variables include: photosynthesis functional variables, transpiration functional variables and water regulation functional variables; S7: Extract driving variables of drought development and drought recovery periods according to drought apex and outlier forms; driving variables include: hydrological and meteorological data, vegetation characteristics and drought duration; S8: Taking the three types of vegetation functional variables as the objects and combining them with the proportion of physiological signals, the driving variables were attribution analyzed to obtain the relative importance ranking of the driving variables to the three types of vegetation functional variables, and to complete the quantitative analysis of the physiological response and structural response of vegetation under the combined stress of extreme dryness and heat.

[0024] Specifically, Figure 1 It is a flow chart for the implementation of quantitative analysis methods of physiological and structural responses of vegetation under extreme dry-heat combined stress; Figure 2 This is a map of the locations of drought events and the corresponding time of drought events in the study area in the Amazon region from 2016 to 2020; Figure 3 It is a time series change chart of hydrological and meteorological data such as soil moisture, vegetation index data such as LAI, and vegetation function index data such as SIF before and after extreme drought; Figure 4 It is the result of the separation of the physiological components of the three types of vegetation functional variables, namely photosynthesis, transpiration and water regulation, in response to drought; Figure 5It is a zoning statistical result map of the overall and physiological parts of the functional response of vegetation during drought; Figure 6 It is the result of attribution analysis of the driving factors of the physiological responses of three types of functional variables of vegetation during the drought development and drought recovery periods. Step S1 includes: The hydrological and meteorological data include: temperature, rainfall, soil moisture and solar radiation data; Specifically, the hydrological and meteorological data are derived from ERA5-Land reanalysis data, with a temporal resolution of daily and a spatial resolution of 0.1°; ERA5-Land reanalysis data are calculated and exported on the Google Earth Engine (GEE) platform, where the temperature is calculated using 2m surface temperature data, and the soil moisture is calculated using the weighted average of soil moisture data at 7cm, 21cm, and 72cm. The meteorological data used are synthetic reanalysis data on a daily scale, which are synthesized and cropped over 4 days on the GEE platform and then exported.

[0025] The vegetation structure data include: leaf area index and near-infrared vegetation reflectance; Specifically, vegetation structure data include leaf area index (LAI) and near-infrared vegetation reflectance (NIRv). Vegetation structure data comes from MODIS MCD15A3H dataset and MCD43A4 dataset, with a temporal resolution of 4 days and a spatial resolution of 500 meters. MODIS MCD15A3H dataset is on a 4-day time scale, and MCD43A4 is on a daily scale. Both datasets are calculated, synthesized, and exported in the GEE platform.

[0026] The vegetation function data include: solar induced chlorophyll fluorescence, vegetation optical thickness and evapotranspiration; Specifically, vegetation function data include solar induced chlorophyll fluorescence (SIF), vegetation optical depth (VOD) and evapotranspiration (ET); vegetation function data are derived from the global space continuous solar induced fluorescence (CSIF) dataset and LPDR v3 vegetation optical depth dataset, with temporal resolutions of 4 days and daily, and spatial resolutions of 0.05° and 25km, respectively; Solar induced chlorophyll fluorescence includes: clear sky instantaneous data and daily average data; Vegetation optical depth includes: daytime vegetation optical depth data and nighttime vegetation optical depth data.

[0027] In one embodiment of the present application, the daily average data is used for screening calculations during the growth period because it can reflect the long-term trend of photosynthesis function; the clear sky instantaneous data is used for subsequent separation calculations of physiological components; the vegetation optical thickness includes: daytime and nighttime vegetation optical depth data, the ratio of which is used here to represent the water regulation function of vegetation, which is hereinafter represented by VOD ratio. In addition, vegetation coverage score and irrigation score data are also included, and the data sources are the ESACCI 2020 global vegetation coverage score dataset and the FAO 2005 global irrigation score dataset.

[0028] The specific steps of the pretreatment include: Perform quality screening and coordinate conversion on hydro-meteorological data, vegetation structure data and vegetation function data; The hydrological and meteorological data, vegetation structure data and vegetation function data are mean synthesized and resampled to unify the temporal and spatial resolutions.

[0029] In one embodiment of the present application, the preprocessing process includes quality screening, coordinate transformation, resampling and synthesis of the above data. The time resolution of all data is 4 days and the spatial resolution is 0.1 degree. Then, uniform cropping and masking are performed to obtain data in a specific study area and study time period. After processing, all data are calculated in the Python environment in the format of Numpy data.

[0030] Step S2 includes: S21: Using the soil moisture data of the historical preset years in the study area, the soil drought index is calculated as follows:

[0031] in Indicates Month and annual soil moisture, i.e. soil drought index; and They are Mean and standard deviation across all years; Filter out Pixels with ,are considered as extremely drought areas; S22: The time when the minimum value of soil moisture in the extremely drought area occurs is taken as the drought peak, and the time when the drought peak occurs is obtained.

[0032] Specifically, the time when the minimum soil moisture occurred in the area where extreme drought occurred was used as a mask, and finally the time step and pixel location map of the extreme drought event in the study area was obtained, such as Figure 2 shown.

[0033] Step S3 includes: S31: Extracting long-term trends of hydrological and meteorological data using a local weighted regression method based on LOWESS; S32: Calculate the monthly average of hydrological and meteorological data to obtain the average seasonal cycle; S33: Based on the hydrological and meteorological data, the long-term change trend and the average seasonal cycle are removed to obtain the outlier form of the hydrological and meteorological data.

[0034] Specifically, the long-term trend is extracted by calling the LOWESS package in the Python environment. This method extracts the long-term seasonal trend of the time series by means of local weighted regression, and then subtracts the long-term seasonal trend from the original time series to obtain the variable outlier form without seasonal trend. The hydrological and meteorological data from 2016 to 2020 are selected, the monthly average of all data is calculated first, the average seasonal cycle is extracted, and then the average seasonal cycle is subtracted from the original data to further eliminate the influence of seasonal fluctuations, so as to obtain the outlier form of all data.

[0035] Step S5 includes: S51: Taking hydrological and meteorological data and vegetation structure data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, the first random forest model was established; S52: Taking hydrological and meteorological data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, the second random forest model was established; S53: training the first random forest model and the second random forest model by using the hydrological and meteorological data, the vegetation structure data, and the vegetation function data outside the time window; S54: Predicting the hydrological and meteorological data and vegetation structure data outside the time window through the trained first random forest model and the second random forest model to obtain prediction results.

[0036] Specifically, data within the time window are selected for model prediction, while data outside the time window are used for model training. Random forest modeling and prediction are performed pixel by pixel, and finally the modeling results within the pixels in extremely arid areas are obtained.

[0037] Step S6 includes: Calculating the residual between the prediction results of the first random forest model and the prediction results of the second random forest model to separate the physiological signals of the vegetation function data; According to different drought development stages and wetness levels, combined with physiological signals, the physiological signal ratios of the three types of vegetation function variables in vegetation function data were statistically analyzed.

[0038] Specifically, both random forest models use vegetation function data as output variables, but use different input variables for modeling. In the final random forest prediction results, the difference in the model is the change in the form of abnormal values ​​caused by the physiological components of vegetation. The physiological response of vegetation is thus separated through the method of model residuals.

[0039] Step S7 includes: The time window before the drought peak is defined as the drought development period, and the time window after the drought peak is defined as the drought recovery period; Hydrometeorological data include: temperature, rainfall, solar radiation and soil moisture; Vegetation characteristics include: vegetation cover fraction and wetness; The drought duration was obtained by identifying the time when soil moisture recovered from abnormal values ​​to normal levels before and after the peak of drought.

[0040] Specifically, in the set time window, the time window before the peak is defined as the development period, and the window after the peak is defined as the recovery period; the hydrological and meteorological data include temperature, rainfall, radiation and soil moisture, and the vegetation characteristics include vegetation cover fraction and wetness; and the drought duration is calculated by identifying the time it takes for the soil moisture to recover from the abnormal value form to the normal level before and after the drought peak.

[0041] Step S8 includes: The SHAP method was used to conduct attribution analysis on the driving variables, quantify the characteristic contributions of hydrological and meteorological data, vegetation characteristics and drought duration to the three types of vegetation functional variables, and obtain the relative importance ranking of the driving variables to the three types of vegetation functional variables.

[0042] Specifically, the SHAP (SHapley Additive exPlanations) method was used. The long-term variables were reduced in dimension by calculating the average value of the stage, and all driving variables were finally quantified by calculating their spatial correlation with the three types of vegetation functional variables, and finally the relative importance ranking of all driving variables to different functional variables was obtained.

[0043] The present invention provides the following embodiments: Embodiment (1): Collection and processing of basic data; In this embodiment, four types of hydrological and meteorological data are collected, namely, temperature, rainfall, soil moisture and solar radiation. The hydrological and meteorological data are processed in the GEE platform by 4-day synthesis to obtain uniform temporal resolution and spatial resolution. The weighted average of soil moisture data at different depths can better reflect the overall water content of the soil.

[0044] Example (2): Extracting the extremely drought areas in the study area and the time when the drought peak occurs; In this embodiment, the SSMI index is calculated to extract the areas where droughts have occurred worldwide and the corresponding drought event times. First, the monthly SSMI index (soil drought index) is calculated using soil moisture data from the past 40 years, which can standardize the time series of soil moisture in each month to reflect the severity of the drought. In order to screen out extremely droughty areas, pixels with SSMI values ​​less than -2 are selected and marked as areas that have experienced extreme droughts. Then, the time of the drought event is further determined by detecting soil moisture records. Specifically, 4 days of soil moisture records are used to find the minimum value in these records as the drought apex, and the time of its occurrence is recorded. In this way, the specific time step of each drought event can be determined. Finally, by masking the minimum soil moisture value with the extremely drought areas that have been screened out, the occurrence time of extreme drought events in the study area and the corresponding pixel position distribution (such as Figure 2 The generated time step and location data can be used to further analyze the spatial and temporal characteristics of drought and help identify and respond to extreme drought events.

[0045] Embodiment (3): removing the average seasonal cycle and long-term seasonal trend of each variable of the hydrometeorological data to extract outliers; In this embodiment, hydrometeorological data from 2016 to 2020 were selected. First, the monthly averages of all data were calculated to extract the average seasonal cycle of each variable. This step averages the data of multiple years in the same month to obtain seasonal fluctuation characteristics. The long-term seasonal trend of each variable is extracted using the LOWESS method. In the Python environment, this operation is performed by calling the LOWESS package, and the extracted long-term seasonal trend is deducted from the original time series to eliminate the impact of the long-term seasonal trend on the data. Finally, the average seasonal cycle is subtracted from the original data to further eliminate the impact of seasonal fluctuations, thereby obtaining the outlier form of each hydrometeorological data and vegetation index data.

[0046] Embodiment (4): Setting a time window to extract the average change trajectory of each variable; In this embodiment, by setting a time window of three months before and after the drought apex as the center, the average change trajectory of each variable during the drought period is calculated. The specific steps are to first determine the drought apex, that is, the time point when the minimum soil moisture occurs in each pixel. Then, within the range of three months before and after this time point, the hydrological and meteorological data and vegetation index data (such as LAI, NIRv, SIF, etc.) are extracted pixel by pixel. By statistically analyzing the average change trend of these data, the average change trajectory of each variable during the drought period is finally obtained, which clearly shows the dynamic change characteristics from the drought development period, the drought apex to the drought recovery period.

[0047] During the drought development period, vegetation index data such as LAI, NIRv and SIF all showed a trend of first increasing and then decreasing, indicating the response of vegetation in the early stage of drought and the decline process as the drought intensified (e.g. Figure 3 Similar trends also appear in the trajectories of VOD ratio, ET, and soil moisture. Soil moisture drops significantly before the peak of drought, and then gradually recovers during the recovery period, reflecting the direct impact of drought on soil moisture storage. These change trajectories reveal the dynamic response mechanism between meteorological conditions and vegetation functions during drought.

[0048] Example (5): Establishing a random forest model of vegetation functional variables to separate physiological response signals In this embodiment, two random forest models are constructed based on hydrometeorological data and vegetation structure variables to predict changes in vegetation function variables such as photosynthesis function (SIF), transpiration function (ET) and water regulation function (VOD ratio). In the model construction process, a combination of hydrometeorological data + vegetation structure variables and separate hydrometeorological data are used as independent variables for pixel-by-pixel random forest modeling and prediction. Through pixel-by-pixel random forest model analysis, the model can accurately capture the changing trends of various vegetation function variables during drought, and finally obtain the vegetation function modeling results in pixels where drought has occurred, thereby revealing the different impacts of drought on vegetation functions (such as Figure 4 as shown).

[0049] Embodiment (6): Based on the model residual, the vegetation physiological response signal is separated and its distribution pattern is statistically analyzed; In this embodiment, two random forest models are used to separate the physiological signals and structural signals of three types of vegetation function variables, with the purpose of deeply studying the impact of drought events on vegetation functions. By taking hydrometeorological data + vegetation structure variables and hydrometeorological data alone as independent variables, two random forest models are established, and three types of vegetation function variables, photosynthesis function (SIF), transpiration function (ET) and water regulation function (VOD ratio), are used as dependent variables. Finally, the changes in vegetation physiological components are separated by calculating the model residuals. The residuals of the model reflect the abnormal changes in vegetation functions caused by physiological factors, and this method provides a more intuitive path for distinguishing vegetation structure and physiological signals. With the drought apex as the center, the time window covers the entire development stage of the drought event. The results of the random forest model show that during drought, the functional performance of vegetation is highly correlated with hydrometeorological data. By analyzing the residuals, the anomalies caused by physiological responses can be accurately separated. By modeling SIF, ET and VOD ratio separately, the results show that the responses of different functional variables during drought are significantly different.

[0050] The photosynthesis function showed negative anomalies throughout the drought period, especially at the peak of drought, which indicated that the drought event had a strong inhibitory effect on the photosynthesis function of vegetation (e.g. Figure 5 ET mainly showed positive anomalies during drought, indicating that plants may relieve drought stress by increasing water transpiration during drought. As a representative of water regulation function, the abnormal performance of VOD ratio varies with the different drought levels. The VOD ratio in some areas showed significant fluctuations during drought, reflecting the differences in the response of vegetation water conditions to drought.

[0051] In the specific drought grouping, by calculating the physiological anomaly ratio of SIF, ET and VOD ratio in pixels with different drought levels, the results show that the proportion of physiological anomalies is significantly higher than that of structural anomalies. For example, in the physiological response of SIF, the proportion of physiological responses under different drought levels is 0.97, 0.91, 0.97 and 0.92 respectively, indicating that the photosynthesis function of vegetation under different drought intensities is mainly dominated by physiological factors. For ET and VOD ratio, the proportion of physiological anomalies to the total anomaly is also concentrated between 0.92 and 0.97, indicating that physiological anomalies occupy a major position in the abnormal changes of the overall function of vegetation, especially during extreme drought, this proportion is even more significant. This high proportion of physiological anomalies indicates that the impact of drought events on vegetation functions is mainly reflected through its physiological functions, while structural anomalies account for a smaller proportion. This finding reveals the dynamic response mechanism of vegetation under drought stress: when soil moisture decreases, vegetation responds to environmental pressure through a series of physiological adjustments (such as photosynthesis and transpiration), while structural changes are relatively slow.

[0052] In addition, the abnormal performance patterns of vegetation functional variables also showed similar trends at different stages of drought development. Especially before and after the peak of drought, the physiological anomalies of SIF, ET and VOD ratio were consistent with the change trend of the overall anomaly, indicating that plants showed strong physiological adjustment capabilities in the process of drought response. In some extreme drought stages, the positive anomaly of ET and the fluctuation characteristics of VOD ratio were particularly significant, further reflecting the importance of vegetation in water regulation. This dynamic change characteristic provides an important basis for us to evaluate the comprehensive impact of drought on the ecosystem.

[0053] Example (7): Extracting driving variables such as hydrological and meteorological variables at different drought development periods; In this embodiment, the time window method is used to extract the hydrometeorological data, vegetation characteristics and drought duration of the drought development and recovery periods with the drought apex as the center. Specifically, according to the set time window, the time window before the drought apex is defined as the drought development period, and the time window after the apex is the drought recovery period. In these two stages, the main hydrometeorological data of concern include key meteorological factors such as temperature, precipitation, radiation and soil moisture; while vegetation characteristics mainly cover vegetation coverage fraction and humidity. The calculation of drought duration is based on the change process of soil moisture anomalies. The duration of the drought is defined by identifying the soil moisture anomalies before and after the drought apex and the time period from the extreme value to the normal level.

[0054] Example (8): Driving variable attribution analysis based on the SHAP method; In this example, the SHapley Additive exPlanations (SHAP) method was used to perform attribution analysis of hydrological and meteorological data, vegetation, and drought characteristics for three types of vegetation function variables, namely photosynthesis function, transpiration function, and water regulation function. This method aims to quantify the contribution of each driving variable to the functional characteristics of vegetation, and help to deeply understand the impact of these variables on vegetation functions during the development and recovery period of drought.

[0055] During the development stage of drought, studies have shown that the main driving variables affecting vegetation functions are hydro-meteorological factors such as temperature and soil moisture (e.g. Figure 6 ). These variables show significant correlation during the drought process, especially during the peak temperature period, when the response of vegetation functions is significantly affected. At the same time, although geographical characteristics such as vegetation cover also have a certain influence, their importance is relatively small compared with meteorological driving variables. Although the impact of drought duration on vegetation functions is relatively limited, in some cases, changes in soil moisture still affect the water supply of vegetation. During the drought recovery period, the impact of each driving variable is roughly similar to that during the drought development period, but the importance of driving variables is generally higher than that during the drought development period. This shows that as vegetation gradually recovers, the influence of hydrometeorological data such as temperature and radiation drop gradually increases.

[0056] By analyzing the impact of driving variables, it was found that ambient temperature has the greatest impact on ET, while the duration of drought and ambient humidity also play an important role. Especially in the drought recovery stage, the increase in ambient humidity helps to improve vegetation transpiration and enhance water use efficiency. In addition, the effects of soil moisture and environmental dryness on VOD ratio cannot be ignored, reflecting the ability of vegetation to regulate water in response to drought. Although the impact of drought duration on VOD ratio is relatively small, it still shows its importance to the moisture status of vegetation under certain circumstances. This multi-dimensional impact assessment provides strong support for a deep understanding of the functional response of vegetation under drought conditions.

[0057] The analysis results based on the SHAP method show that during the development of drought, temperature and soil moisture, as the main hydrological and meteorological data, significantly affect the photosynthesis and transpiration of vegetation. In the drought recovery stage, the impact of these variables is still significant, and their importance is higher than that in the drought development stage. This shows that meteorological driving factors play a key role in the recovery of vegetation functions. The study also found that there are differences in the response patterns of different vegetation function variables. The abnormal responses of SIF, ET and VOD ratio in drought events reveal the complex relationship between the physiological regulation ability and structural changes of vegetation. This dynamic response mechanism further emphasizes the combined impact of soil moisture and meteorological conditions on vegetation functions.

[0058] The present application also discloses an electronic device. Figure 7 , Figure 7 The electronic device 500 may include: at least one processor 501 , at least one network interface 504 , a user interface 503 , a memory 505 , and at least one communication bus 502 .

[0059] The communication bus 502 is used to realize the connection and communication between these components.

[0060] The user interface 503 may include a display screen, and the optional user interface 503 may also include a standard wired interface or a wireless interface.

[0061] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0062] The present application also discloses a computer-readable storage medium storing a plurality of instructions suitable for loading by a processor to execute the above-mentioned quantitative analysis method of physiological and structural responses of vegetation under extreme dry-heat combined stress.

[0063] The above are only exemplary embodiments of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure.

[0064] This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art not described in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A quantitative analysis method for the physiological and structural responses of vegetation under extreme dry-heat combined stress, characterized in that: The method comprises the following steps: S1: Collect hydrological and meteorological data, vegetation structure data and vegetation function data of the study area and pre-process them; S2: Obtain soil moisture data of the study area; determine the drought peak by calculating the soil drought index of the soil moisture data; S3: Extract the outlier form of hydrometeorological data by combining hydrometeorological data with local weighted regression method; S4: Set a time window centered on the drought apex and calculate the average change trajectory of each variable of the hydrological and meteorological data during the drought period; S5: Taking hydrological and meteorological data and vegetation structure data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, a random forest model is established; based on the time window, the random forest model is trained and the prediction results of extremely drought areas are obtained; S6: According to the prediction results, the physiological signal ratios of the three types of vegetation function variables of the vegetation function data are obtained; The three types of vegetation functional variables include: photosynthesis functional variables, transpiration functional variables and water regulation functional variables; S7: Extract driving variables of drought development and drought recovery periods according to drought apex and outlier forms; driving variables include: hydrological and meteorological data, vegetation characteristics and drought duration; S8: Taking the three types of vegetation functional variables as the objects and combining them with the proportion of physiological signals, the driving variables were attribution analyzed to obtain the relative importance ranking of the driving variables to the three types of vegetation functional variables, and to complete the quantitative analysis of the physiological response and structural response of vegetation under the combined stress of extreme dryness and heat.

2. The quantitative analysis method of physiological and structural response of vegetation under extreme dry-heat combined stress according to claim 1, characterized in that: Step S1 includes: The hydrological and meteorological data include: temperature, rainfall, soil moisture and solar radiation data; The vegetation structure data include: leaf area index and near-infrared vegetation reflectance; The vegetation function data include: solar induced chlorophyll fluorescence, vegetation optical thickness and evapotranspiration; Solar induced chlorophyll fluorescence includes: clear sky instantaneous data and daily average data; Vegetation optical depth includes: daytime vegetation optical depth data and nighttime vegetation optical depth data; The specific steps of the pretreatment include: Perform quality screening and coordinate conversion on hydro-meteorological data, vegetation structure data and vegetation function data; The hydrological and meteorological data, vegetation structure data and vegetation function data are mean synthesized and resampled to unify the temporal and spatial resolutions.

3. The quantitative analysis method of physiological and structural response of vegetation under extreme dry-heat combined stress according to claim 1, characterized in that: Step S2 includes: S21: Using the soil moisture data of the historical preset years in the study area, the soil drought index is calculated as follows: in Indicates Month and Annual soil moisture, i.e. soil drought index; and They are Mean and standard deviation across all years; Filter out Pixels with ,are considered as extremely drought areas; S22: The time when the minimum value of soil moisture in the extremely drought area occurs is taken as the drought peak, and the time when the drought peak occurs is obtained.

4. The quantitative analysis method of physiological and structural response of vegetation under extreme dry-heat combined stress according to claim 1, characterized in that: Step S3 includes: S31: Extracting long-term trends of hydrological and meteorological data using a local weighted regression method based on LOWESS; S32: Calculate the monthly average of hydrological and meteorological data to obtain the average seasonal cycle; S33: Based on the hydrological and meteorological data, the long-term change trend and the average seasonal cycle are removed to obtain the outlier form of the hydrological and meteorological data.

5. The quantitative analysis method of physiological and structural response of vegetation under extreme dry-heat combined stress according to claim 1, characterized in that: Step S5 includes: S51: Taking hydrological and meteorological data and vegetation structure data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, the first random forest model was established; S52: Taking hydrological and meteorological data as independent variables and vegetation function data as dependent variables, combined with the average change trajectory, the second random forest model was established; S53: training the first random forest model and the second random forest model by using the hydrological and meteorological data, the vegetation structure data, and the vegetation function data outside the time window; S54: Predicting the hydrological and meteorological data and vegetation structure data outside the time window through the trained first random forest model and the second random forest model to obtain prediction results.

6. The quantitative analysis method of physiological and structural response of vegetation under extreme dry-heat combined stress according to claim 5, characterized in that: Step S6 includes: Calculating the residual between the prediction results of the first random forest model and the prediction results of the second random forest model to separate the physiological signals of the vegetation function data; According to different drought development stages and wetness levels, combined with physiological signals, the physiological signal ratios of the three types of vegetation function variables in vegetation function data were statistically analyzed.

7. The quantitative analysis method of physiological and structural response of vegetation under extreme dry-heat combined stress according to claim 1, characterized in that: Step S7 includes: The time window before the drought peak is defined as the drought development period, and the time window after the drought peak is defined as the drought recovery period; Hydrometeorological data include: temperature, rainfall, solar radiation and soil moisture; Vegetation characteristics include: vegetation cover fraction and wetness; The drought duration was obtained by identifying the time when soil moisture recovered from abnormal values ​​to normal levels before and after the peak of drought.

8. The quantitative analysis method of physiological and structural response of vegetation under extreme dry-heat combined stress as claimed in claim 1, characterized in that: Step S8 includes: The SHAP method was used to conduct attribution analysis on the driving variables, quantify the characteristic contributions of hydrological and meteorological data, vegetation characteristics and drought duration to the three types of vegetation functional variables, and obtain the relative importance ranking of the driving variables to the three types of vegetation functional variables.

9. An electronic device, characterized in that: The electronic device comprises a processor (501), a memory (505), a user interface (503) and a network interface (504), wherein the memory (505) is used to store instructions, the user interface (503) and the network interface (504) are used to communicate with other devices, and the processor (501) is used to execute the instructions stored in the memory (505) so that the electronic device executes the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed by a computer, the method according to any one of claims 1 to 8 is executed.

Images