A method for quantitatively evaluating and extracting spatial distribution of composite high-temperature drought history analogy deviation degree

Abstract

The present application belongs to the technical field of disaster meteorological risk assessment, and particularly relates to a kind of composite high temperature drought history analogy deviation quantitative evaluation and spatial distribution extraction method, based on multiple source data such as monthly air temperature, vapor pressure difference and soil humidity, first determine the warm season window dominated by high temperature per grid and per year;Subsequently, the composite high temperature drought event is identified by using the high value threshold of VPD in the warm season and the low value threshold of SM;Then, the composite high temperature drought state vector is constructed from three dimensions of frequency, severity and duration;Finally, the historical reference library is formed by all effective grids in the historical reference period, and the historical analogy deviation of each grid in the target period is extracted by the minimum value of standardized Euclidean distance, and it is mapped to a visual raster layer.The present application can more accurately represent the composite climate stress formed by the combined action of high evaporation demand and soil moisture deficit, and realize the quantitative measurement of the composite high temperature drought state of the target period relative to the historical experience envelope.

Description

Technical Field

[0001] This invention belongs to the field of disaster meteorological risk assessment technology, and in particular relates to a method for quantitative assessment and spatial distribution extraction of the deviation of historical analogy between composite high temperature and drought. Background Technology

[0002] Against the backdrop of global warming, the frequency, duration, and intensity of extreme weather events are all showing an increasing trend. Drought, as a natural disaster with a wide impact, slow evolution, and persistent and significant harm, poses a long-term threat to agricultural production, water resource security, terrestrial ecosystem stability, and regional socio-economic sustainable development. Traditional drought monitoring methods typically focus on precipitation deficit, reduced soil moisture, abnormal evapotranspiration, or single drought indices. While these methods can reflect the abnormal state of a particular physical process, they are insufficient to fully describe the complex extreme risks formed when multiple climate stresses occur simultaneously and reinforce each other.

[0003] Combined high-temperature drought refers to a combined extreme event in which increased atmospheric aridity and soil moisture deficit occur simultaneously under a high-temperature or warm-season background. In this process, a higher vapor pressure gradient (VPD) indicates increased atmospheric evaporation demand, while lower soil moisture (SM) indicates limited surface water supply. The resulting coupled state of atmospheric and soil drought typically has a stronger impact on vegetation photosynthesis, ecosystem productivity, crop yield, and regional carbon sequestration than a single drought event.

[0004] In recent years, the complex high-temperature droughts that have occurred in some regions have not only exhibited increased intensity or prolonged duration, but also shown structural changes in multidimensional characteristics such as frequency, severity, and duration, which differ from historical experience. In other words, the complex high-temperature drought conditions that occur during certain target periods are difficult to find similar references in historical observation samples, demonstrating a clear lack of historical analogy. If only single statistical measures such as drought frequency, single-event intensity, or duration are used for analysis, it is difficult to determine whether the complex high-temperature drought during the target period has deviated from the historical state, and it is also difficult to identify key areas where the complex drought risk pattern has been restructured.

[0005] Based on this, the present invention defines the degree of lack of similarity references between the multidimensional characteristics of the target period's combined high temperature and drought and the historical sample space as the "historical analogy deviation of combined high temperature and drought". This index is used to characterize the minimum statistical distance between a certain geographic grid in the target period and the known combined high temperature and drought states represented by all valid grids in the historical baseline period in the multidimensional feature space composed of frequency, severity, and duration. The higher the historical analogy deviation, the more difficult it is to find a similar state in the historical baseline period for the combination of combined high temperature and drought in the target period; the lower the historical analogy deviation, the more the state in the target period is still within the range of historical experience. Summary of the Invention

[0006] The purpose of this invention is to overcome the technical shortcomings of existing complex drought assessment methods, such as difficulty in characterizing multidimensional structural changes, difficulty in measuring the degree of deviation of the target period state from historical samples, and difficulty in forming spatially expressible indicators. This invention provides a method for quantitative assessment and spatial distribution extraction of the deviation of historical analogy of complex high temperature drought.

[0007] This invention is based on multi-source data such as monthly air temperature, vapor pressure difference, and soil moisture. First, it determines the warm season window dominated by high temperature grid by grid and year by year. Then, it uses the high value threshold of VPD and the low value threshold of SM within the warm season to jointly identify complex high temperature and drought events. Next, it constructs a state vector of complex high temperature and drought from three dimensions: frequency, severity, and duration. Finally, it uses all effective grids of the historical baseline period to form a historical reference library, extracts the historical analogy deviation of each grid in the target period by the minimum value of standardized Euclidean distance, and maps it to a visual raster layer.

[0008] To achieve the above objectives, the present invention provides a method for quantitative assessment and spatial distribution extraction of the deviation of historical analogy of composite high temperature and drought, comprising the following steps.

[0009] S1: Acquisition of multi-source basic data;

[0010] The study area and assessment period were defined, and multi-source monthly environmental data covering the historical baseline period t1 and the target evaluation period t2 were collected. This data includes, but is not limited to, air temperature data, vapor pressure differential data, soil moisture data, study area boundary data, and auxiliary geographic data used for masking. Air temperature data was used to identify periods dominated by high temperatures, vapor pressure differential data was used to characterize atmospheric aridity, and soil moisture data was used to characterize the degree of surface water deficit.

[0011] The historical baseline period t1 can be a continuous period of 5, 10, or longer preceding the target evaluation period; the target evaluation period t2 can be a continuous period immediately following the historical baseline period, or it can be a future or near-term period to be evaluated. The historical baseline period is used to establish a sample space of known climate states, and the target evaluation period is used to extract the historical analogy deviation of the composite high temperature and drought of the grid to be evaluated.

[0012] S2: Multi-source data preprocessing and grid isomorphism;

[0013] The data obtained in step S1 are subjected to coordinate system unification, time series alignment, study area pruning, outlier removal, and spatial resampling. Preferably, all data are unified to the same geographic coordinate system, the same monthly time step, and the same spatial resolution, and a unified grid is used as the basic spatial unit for subsequent calculations.

[0014] A validity marker is established for each grid point. If a grid point has missing key variables, is located in a non-land area, has water cover, permanent snow cover, or otherwise fails to meet the calculation requirements during the evaluation period, it is marked as an invalid grid and removed. After preprocessing, a set of valid grid points that are fully aligned in spatial, temporal, and variable dimensions between the historical baseline period and the target evaluation period is obtained.

[0015] S3: Dynamic determination of the warm season window at grid scale;

[0016] For each valid grid point g and each year y, the average temperature for each window within the year is calculated using a sliding window of three consecutive months based on monthly air temperature data. If the starting month of the window is r, the three-month average temperature can be expressed as: ;

[0017] in, Representing grid points The air temperature in the r-th month of year y, where r is the starting month of the window. (Select) The longest consecutive three months serve as the warm season window W(g,y) for that grid point that year;

[0018] By determining the warm season window grid by grid and year by year, it can adapt to the differences in high temperature seasons at different latitudes, altitudes and climate zones, avoid the drought identification bias caused by using a fixed monthly window, and concentrate the identification of subsequent complex high temperature and drought events on the period when high temperature stress is most obvious.

[0019] S4: Joint identification of complex high-temperature and drought events;

[0020] Based on the warm season window determined in step S3, VPD and SM data for the warm season months of the historical baseline period or the entire study period are extracted, and the 90th percentile threshold VPD is calculated grid by grid. 90 And the 10th percentile threshold of SM 10 Among them, VPD 90 This indicates the high-critical state of atmospheric evaporation demand for this grid, SM. 10 This indicates the low-critical state of soil moisture deficit in this grid.

[0021] For any grid point g in any warm season month t, a combined high-temperature drought event is determined to have occurred at that grid point in that month if the following condition is met simultaneously: VPD(g,t) >= VPD 90， SM(g,t) <= SM 10；

[0022] This allows the establishment of a composite high-temperature drought occurrence marker sequence C(g,t). C(g,t) = 1 when the joint threshold condition is met; otherwise, C(g,t) = 0. This sequence serves as the basis for subsequent calculations of frequency, severity, and duration.

[0023] S5: Three-dimensional feature extraction of composite high temperature and drought;

[0024] Based on the composite high-temperature drought occurrence marker sequence obtained in step S4, three core features—frequency, severity, and duration—were extracted for each effective grid point within the historical baseline period t1 and the target evaluation period t2, respectively.

[0025] (1) Frequency F: This represents the proportion of months with compound high-temperature drought to the total number of warm months in the assessment period, reflecting the normalization of compound high-temperature drought occurrence. The calculation formula is: F(g) = N drought (g) / N warm (g);

[0026] Where, N drought (g) represents the number of warm season months during the assessment period for grid point g that were identified as experiencing compound high-temperature drought, N warm (g) represents the total number of warm season months within the assessment period.

[0027] (2) Severity I: Represents the degree of joint standardized deviation between the VPD exceeding the threshold and the SM deficit within the month of the combined high-temperature drought. For any month t of the combined high-temperature drought, its instantaneous severity I can be expressed as: ;

[0028] in, and denoted as the standard deviations of VPD and SM for grid point g during the warm season months of the historical baseline period.

[0029] (3) Duration D: Represents the longest duration of the continuous occurrence of the combined high temperature and drought process within the assessment period, used to characterize the temporal persistence of the combined high temperature and drought event. If no combined high temperature and drought occurs at a certain grid point within the assessment period, F, I and D can be assigned a value of 0, or set to null values ​​according to actual application and excluded in subsequent calculations.

[0030] S6: Construct the feature matrix of the historical benchmark period and the target evaluation period;

[0031] The frequency F, severity I, and duration D obtained in step S5 are combined into a three-dimensional composite high-temperature drought state vector. For any valid grid point i in the historical baseline period t1, its state vector is denoted as: X i = [Fi , I i D i ];

[0032] For any valid grid point j in the target evaluation period t2, its state vector is denoted as: Y j = [F j , I j D j ];

[0033] The state vectors of all valid grid points in the historical baseline period are summarized to obtain the historical baseline period feature matrix X; the state vectors of all valid grid points in the target evaluation period are summarized to obtain the target evaluation period feature matrix Y. The historical baseline period feature matrix X constitutes a composite high-temperature drought historical state reference database.

[0034] S7: Calculation of characteristic variance for the historical baseline period;

[0035] To eliminate differences in the dimensions, ranges, and spatial dispersion of the three features—frequency, severity, and duration—the spatial variance of each feature dimension in the historical baseline feature matrix X is calculated. For the k-th feature dimension, its variance Var(X) is calculated. k The calculation formula is: ;

[0036] Where m is the total number of valid grid points in the historical baseline period, and X i,k Let i be the value of the i-th grid point in the k-th feature dimension during the historical baseline period. This represents the average value of the k-th feature across all valid grid points in the historical baseline period. k=1, 2, and 3 correspond to frequency, severity, and duration, respectively.

[0037] If the variance of a certain feature dimension is zero or close to zero in the historical reference period, a very small positive number is introduced as a stabilizing term in the calculation to avoid the denominator being zero and to maintain the numerical stability of the distance calculation.

[0038] S8: Calculation of the standardized distance between the target period and the historical baseline period;

[0039] For any valid grid point j during the target evaluation period, its state vector Y j Compared with the state vector X of all valid grid points in the historical reference period i Each dimension is compared individually, and the historical variance obtained in step S7 is used to standardize and correct the differences. The standardized Euclidean distance between grid point j in the target period and grid point i in the historical baseline period is calculated. The calculation formula is: ;

[0040] in, This represents the value of the j-th grid point in the k-th feature dimension during the target evaluation period. This represents the value of the i-th grid point in the k-th feature dimension during the historical baseline period. Variance standardization effectively reduces the interference of different feature dimensions and numerical fluctuation ranges on the overall distance result, allowing comparison of the three feature dimensions within a unified distance framework.

[0041] S9: Historical analogy deviation extraction;

[0042] For each valid grid point j in the target evaluation period, calculate its standardized Euclidean distance with all valid grid points in the historical baseline period, and extract the minimum value from all distances as the composite high-temperature drought historical analogy deviation of that grid point. The calculation formula is as follows: ;

[0043] in, The larger the value, the more difficult it is to find a similar reference for the composite high temperature and drought state vector of grid point j in the target evaluation period in all samples of the historical baseline period. The higher the degree of historical non-analogy, the more likely the region may have experienced a structural deviation from composite high temperature and drought. The smaller the value, the closer the target evaluation period is to the existing state in the historical baseline period, indicating that the complex high temperature and drought risk pattern has strong historical continuity.

[0044] Unlike methods that directly subtract between earlier and later periods of the same grid, this step employs a one-to-many global matching strategy between the target period grid and all historical baseline grids. This strategy does not require the target grid to be compared only with its own historical state, but rather seeks the most similar reference in the entire historical state sample library, making it more suitable for identifying novel complex high-temperature and drought combinations that lack historical analogies at the regional scale.

[0045] S10: Extraction of spatial distribution of historical analogy deviation;

[0046] All valid grid points for the target evaluation period obtained in step S9 The numerical values ​​are remapped to a spatial raster matrix based on the original latitude and longitude or projected coordinates of each grid point, generating a composite spatial distribution layer of historical high-temperature and drought analogy deviation. This layer can be further divided into low-deviation, medium-deviation, and high-deviation zones using quantile, natural breakpoint, standard deviation, or preset threshold methods.

[0047] Areas with high historical analogy deviation indicate that the combination of frequency, severity, and duration of the target period's complex high-temperature drought has significantly deviated from the historical reference state, and can be used as regional drought risk reconstruction areas, key monitoring areas for ecosystem vulnerability, and priority areas for water resource adaptive management.

[0048] The beneficial effects of this invention are:

[0049] 1. This invention takes the joint anomaly of VPD and SM during the high-temperature-dominant period as the identification object, and expands the complex high-temperature drought from single-factor drought monitoring to the identification of land-atmosphere coupled extreme events, which can more accurately characterize the complex climate stress formed by the combined effects of high evaporation demand and soil moisture deficit.

[0050] 2. This invention constructs a three-dimensional state vector that unifies the occurrence frequency, severity and duration of complex high-temperature drought, transforming drought risk assessment from a single statistical quantity into a multi-dimensional structural state assessment, which can reveal the frequency-intensity-duration combination changes that cannot be captured by a single indicator.

[0051] 3. This invention constructs a historical reference library by using all valid grids of the historical baseline period, and uses the minimum standardized Euclidean distance between the target period grid and the historical reference library as the historical analogy deviation, thereby realizing a quantitative measurement of the target period's complex high temperature and drought state relative to the historical experience envelope.

[0052] 4. This invention introduces the characteristic variance of the historical reference period as a standardization factor, which can reduce the impact of inconsistencies in dimensions, fluctuation ranges, and dispersion of different physical quantities on the comprehensive distance calculation, and improve the stability of cross-regional, cross-period, and cross-climate zone comparisons.

[0053] 5. This invention can convert abstract multidimensional statistical distances into grid-scale spatial distribution layers, intuitively identifying high-value areas of structural deviations in complex high-temperature drought, and providing an operable spatial basis for drought monitoring and early warning, water resource allocation, ecological risk prevention and control, and climate adaptation management. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention.

[0055] Figure 2 This is a flowchart of the steps of the present invention;

[0056] Figure 3 This is a flowchart illustrating the dynamic determination of the warm season window, the monthly identification of complex high temperature and drought conditions, and the construction of three-dimensional feature vectors in this invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the invention will be further described below with reference to specific embodiments. These embodiments are only for explaining the invention and are not intended to limit the scope of protection of the invention.

[0058] This embodiment follows Figure 1 and Figure 2The process shown takes the Chinese mainland region as the study area, selects 2001-2010 as the historical baseline period t1, and selects 2011-2020 as the target evaluation period t2. Monthly temperature, vapor pressure difference and soil moisture data are used as the main inputs to quantitatively evaluate the deviation of the historical analogy of the combined high temperature and drought.

[0059] S1. Collect monthly air temperature data, monthly vapor pressure difference data, and monthly soil moisture data for the study area, and obtain the boundary vector data of the study area. The data can be derived from meteorological observation interpolation products, reanalysis products, or remote sensing inversion products, as long as they can provide continuous monthly time series covering the study area.

[0060] S2. In the GIS spatial analysis environment, all variables are unified to the WGS84 geographic coordinate system. The study area boundaries are used for cropping, and bilinear interpolation or other suitable resampling methods for continuous variables are employed to unify the spatial resolution to 0.25°×0.25°. Grid points with numerous missing values, those not belonging to land or water bodies, or those that do not meet the analysis requirements are removed, resulting in a valid grid set.

[0061] The process of dynamically determining the warm season window, identifying the combined high temperature and drought on a monthly basis, and constructing the three-dimensional feature vector is as follows: Figure 3 As shown, the specific steps are as follows:

[0062] S3 involves performing a three-month sliding window calculation for each valid grid point annually to obtain the three consecutive months with the highest average annual temperature for that grid point, which are then used as the warm season window for that year. The warm season window may vary for different regions due to different climatic backgrounds, thus avoiding spatial bias caused by fixed-month windows.

[0063] S4, VPD is calculated grid-by-grid based on the VPD and SM sequences of the warm season months of the historical baseline period. 90 and SM 10 Then, iterate through all warm season months from 2001 to 2020, and when a certain grid cell meets the condition that its VPD is not less than VPD in a certain month... 90 And SM is no higher than SM 10 At that time, the month was marked as the month of the occurrence of complex high temperature and drought.

[0064] S5 calculates the composite high-temperature drought frequency F, severity I, and duration D for each effective grid point in the two periods of 2001-2010 and 2011-2020. The frequency is the ratio of the number of drought months to the total number of warm season months, the severity is the normalized Euclidean distance mean of the VPD over-threshold amplitude and SM deficit amplitude, and the duration is the longest consecutive drought months.

[0065] S6. Combine the [F,I,D] of each effective grid point from 2001 to 2010 into a historical baseline feature matrix X, and combine the [F,I,D] of each effective grid point from 2011 to 2020 into a target evaluation period feature matrix Y. Calculate the variance of the three feature dimensions in X to obtain the variance parameters required for standardized distance calculation.

[0066] S7. For each valid grid point j in the target evaluation period, calculate its standardized Euclidean distance with all valid grid points i in the historical baseline period in the three-dimensional feature space, and extract the minimum value as the historical analogy deviation (HAD) of that grid point. j The historical grid point corresponding to this minimum value is the most similar reference state of the target period grid point in the historical baseline period.

[0067] S8, HAD of all target evaluation period grid points j The numerical values ​​are reassigned to the spatial raster according to their latitude and longitude positions, outputting a composite high-temperature drought historical analogy deviation raster layer with a resolution of 0.25°. Further adjustments can be made based on HAD... j The numerical values ​​are used to classify different levels of regions, which are used to identify key areas with significant structural changes in complex high-temperature drought.

[0068] S9, in a specific embodiment, using 2001–2010 as the historical baseline period and 2011–2020 as the target evaluation period, a raster layer showing the spatial distribution of the historical analogy deviation of the composite high-temperature drought on a ten-year scale is output. Furthermore, using a five-year time window, the historical analogy deviation of the composite high-temperature drought from 2006–2010 to 2001–2005 and from 2016–2020 to 2011–2015 is calculated, and the corresponding spatial distribution raster layer is output. Through the above multi-time-window comparative analysis, the degree of deviation of the multi-dimensional characteristics of composite high-temperature drought from the historical baseline state and its spatial evolution pattern under multiple time windows can be identified. Thus, the quantitative extraction and spatial distribution mapping of the historical analogy deviation of the composite high-temperature drought in the target area are completed.

[0069] Through the above implementation process, this invention can automatically complete the identification of complex high-temperature and drought events, quantification of three-dimensional features, calculation of historical analogy deviation, and spatial mapping without relying on subjective human interpretation. The results can not only show the intensity of complex high-temperature and drought events, but also reveal whether the combination of complex high-temperature and drought characteristics in the target period has deviated from the historical reference state, thereby providing quantitative technical support for identifying new extreme climate risk areas.

Claims

1. A method for quantitative assessment and spatial distribution extraction of deviation from historical analogies of composite high temperature and drought, characterized in that, Includes the following processes: Based on multi-source data such as monthly temperature, vapor pressure difference and soil moisture, the warm season window dominated by high temperature is first determined grid by grid and year by year; Subsequently, the combined high-value threshold of VPD and low-value threshold of SM during the warm season were used to identify complex high-temperature and drought events; Then, a composite high-temperature drought state vector is constructed from three dimensions: frequency, severity, and duration. Finally, a historical reference library is constructed using all valid grids from the historical baseline period. The historical analogy deviation of each grid in the target period is extracted by using the minimum value of the standardized Euclidean distance, and then mapped to a visual raster layer.

2. The method for quantitative assessment and spatial distribution extraction of the deviation of historical analogy of composite high temperature and drought as described in claim 1, characterized in that, Specifically, the following steps are included: S1: Acquisition of multi-source basic data; The study area and assessment period were determined, and multi-source monthly environmental data covering the historical baseline period t1 and the target evaluation period t2 were collected. The historical baseline period is used to establish a sample space of known climate states, while the target evaluation period is used to extract the historical analogy deviation of the composite high temperature and drought of the grid to be evaluated. S2: Multi-source data preprocessing and grid isomorphism; The data obtained in step S1 are processed by coordinate system 1, time series alignment, study area cropping, outlier removal and spatial resampling. Establish a validity identifier for each grid point; If a grid point has missing key variables, is located in a non-land area, has water cover, is covered by permanent ice and snow, or otherwise does not meet the calculation requirements during the evaluation period, it will be marked as an invalid grid and removed. After preprocessing, an effective grid set that is fully aligned in spatial, temporal, and variable dimensions is obtained between the historical baseline period and the target evaluation period; S3: Dynamic determination of the warm season window at grid scale; For each valid grid point g and each year y, the average temperature of each window within the year is calculated using a sliding window of three consecutive months based on monthly air temperature data. By determining the warm season window grid year by year, it adapts to the differences in high temperature seasons in different latitudes, altitudes and climate zones; S4: Joint identification of complex high-temperature and drought events; Based on the warm season window determined in step S3, VPD and SM data for the warm season months of the historical baseline period or the entire study period are extracted, and the 90th percentile threshold VPD is calculated grid by grid. 90 And the 10th percentile threshold of SM 10 ; Among them, VPD 90 This indicates the high-critical state of atmospheric evaporation demand for this grid, SM. 10 This indicates the low-critical state of soil moisture deficit in this grid. For any grid point g in any warm season month t, a combined high-temperature drought event is determined to have occurred at that grid point in that month if the following condition is met simultaneously: VPD(g,t) >= VPD 90 SM(g,t) <= SM 10； Therefore, a composite high-temperature drought occurrence marker sequence C(g,t) is established; when the joint threshold condition is met, C(g,t)=1; otherwise, C(g,t)=0. S5: Three-dimensional feature extraction of composite high temperature and drought; Based on the composite high-temperature drought occurrence marker sequence obtained in step S4, three core features—frequency, severity, and duration—were extracted for each effective grid point during the historical baseline period t1 and the target evaluation period t2, respectively. S6: Construct the feature matrix of the historical benchmark period and the target evaluation period; The frequency F, severity I, and duration D obtained in step S5 are combined into a three-dimensional composite high-temperature drought state vector; for any valid grid point i in the historical baseline period t1, its state vector is denoted as: X i = [F i , I i , D i ]; For any valid grid point j in the target evaluation period t2, its state vector is denoted as: Y j = [F j , I j , D j ]; The state vectors of all valid grid points in the historical baseline period are summarized to obtain the historical baseline period feature matrix X; the state vectors of all valid grid points in the target evaluation period are summarized to obtain the target evaluation period feature matrix Y; the historical baseline period feature matrix X constitutes a composite high-temperature drought historical state reference library. S7: Calculation of characteristic variance for the historical baseline period; The spatial variance of each feature dimension in the historical baseline feature matrix X is calculated. S8: Calculation of the standardized distance between the target period and the historical baseline period; For any valid grid point j during the target evaluation period, its state vector Y j Compared with the state vector X of all valid grid points in the historical reference period i Compare them one by one, and use the historical variance obtained in step S7 to standardize and correct the differences in each dimension; S9: Historical analogy deviation extraction; For each valid grid point j in the target evaluation period, calculate its standardized Euclidean distance with all valid grid points in the historical baseline period, and extract the minimum value from all distances as the composite high-temperature drought historical analogy deviation of that grid point. ; S10: Extraction of spatial distribution of historical analogy deviation; All valid grid points for the target evaluation period obtained in step S9 The numerical values ​​are remapped to a spatial raster matrix based on the original latitude and longitude or projected coordinates of each grid point to generate a composite high temperature and drought historical analogy deviation spatial distribution layer. This layer is divided into low deviation zone, medium deviation zone and high deviation zone by quantile method, natural breakpoint method, standard deviation classification method or preset threshold method. Areas with high historical analogy deviation indicate that the combination of frequency, severity, and duration of the target period's complex high-temperature drought has significantly deviated from the historical reference state, and should be designated as areas for regional drought risk reconstruction, key monitoring areas for ecosystem vulnerability, and priority areas for water resource adaptive management.

3. The method for quantitative assessment and spatial distribution extraction of the deviation of historical analogy of composite high temperature and drought as described in claim 2, characterized in that, In S1, the data includes air temperature data, vapor pressure difference data, soil moisture data, study area boundary data, and auxiliary geographic data used for masking. Air temperature data is used to identify the high-temperature dominance period, vapor pressure difference data is used to characterize atmospheric aridity, and soil moisture data is used to characterize the degree of surface water deficit.

4. The method for quantitative assessment and spatial distribution extraction of deviation from historical high-temperature drought data according to claim 2, characterized in that, In S2, all data are unified to the same geographic coordinate system, the same monthly time step, and the same spatial resolution, and a unified grid is used as the basic spatial unit for subsequent calculations.

5. The method for quantitative assessment and spatial distribution extraction of the deviation of historical analogy of composite high temperature and drought as described in claim 2, characterized in that, In S3, if the starting month of the window is r, then the average temperature over the three months is expressed as: ; in, Representing grid points The air temperature in the r-th month of year y, where r is the starting month of the window; select The longest consecutive three months serve as the warm season window W(g,y) for that grid point that year.

6. The method for quantitative assessment and spatial distribution extraction of deviation from historical high-temperature drought data according to claim 2, characterized in that, The specific calculation methods for the three core features of S5—frequency, severity, and duration—are as follows: (1) Frequency F: represents the proportion of months with compound high-temperature drought to the total number of warm season months in the assessment period, used to reflect the normalization of compound high-temperature drought occurrence; its calculation formula is: F(g) = N drought (g) / N warm (g); Where, N drought (g) represents the number of warm season months during the assessment period for grid point g that were identified as experiencing compound high-temperature drought, N warm (g) is the total number of warm season months within the assessment period; (2) Severity I: represents the degree of joint standardized deviation between the VPD exceeding the threshold and the SM deficit within the month of the combined high temperature and drought; for any month t of the combined high temperature and drought, its instantaneous severity I is expressed as: ； in, and These are the standard deviations of VPD and SM for grid point g during the warm season months of the historical baseline period, respectively. (3) Duration D: represents the longest number of months of continuous occurrence of compound high temperature and drought during the assessment period, used to characterize the temporal persistence of compound high temperature and drought events; if no compound high temperature and drought occurs at a certain grid point during the assessment period, F, I and D are assigned the value 0, or set to null values ​​according to actual application and excluded in subsequent calculations.

7. The method for quantitative assessment and spatial distribution extraction of deviation from historical high-temperature drought data according to claim 2, characterized in that, In S7, for the k-th feature dimension, its variance Var(X) k The calculation formula is: ； Where m is the total number of valid grid points in the historical baseline period, and X i,k Let i be the value of the i-th grid point in the k-th feature dimension during the historical baseline period. The value of the k-th feature is the average value across all valid grid points in the historical baseline period; k=1, 2, and 3 correspond to frequency, severity, and duration, respectively. If the variance of a certain feature dimension is zero or close to zero in the historical baseline period, then a very small positive number is introduced as a stabilizing term in the calculation.

8. The method for quantitative assessment and spatial distribution extraction of deviation from historical high-temperature drought data according to claim 2, characterized in that, In S8, the standardized Euclidean distance between grid point j in the target period and grid point i in the historical baseline period is... The calculation formula is: ； in, This represents the value of the j-th grid point in the k-th feature dimension during the target evaluation period. This represents the value of the i-th grid point in the k-th feature dimension during the historical baseline period.

9. The method for quantitative assessment and spatial distribution extraction of deviation from historical high-temperature drought data according to claim 2, characterized in that, In S9, the deviation of the composite high temperature and drought historical analogy of grid points The calculation formula is: ； in, The larger the value, the more difficult it is to find a similar reference for the composite high temperature and drought state vector of grid point j in the target evaluation period in all samples of the historical baseline period. The higher the degree of historical non-analogy, the more likely the region may have experienced a structural deviation from composite high temperature and drought. The smaller the value, the closer the target evaluation period is to the existing state in the historical baseline period, indicating that the complex high temperature and drought risk pattern has strong historical continuity.

Images