A method for predicting spatio-temporal variation of future soil carbon at regional scale based on a geographically weighted regression model

Abstract

The present application relates to the technical field of ecological environment management, and provides a regional scale future soil carbon spatio-temporal variation prediction method based on a geographic weighted regression model, comprising the following steps: S1: collecting data; S2: constructing a soil carbon future spatio-temporal variation prediction model and performing future spatio-temporal variation prediction of soil carbon density in different periods, comprising: S21: calculating the annual variation rate of soil carbon density; S22: selecting a prediction model auxiliary variable; S23: constructing a soil carbon future spatio-temporal variation prediction model; S24: future spatio-temporal variation prediction of soil carbon density in different periods. The present application uses the geographic weighted regression model to capture the empirical relationship between the annual variation rate of soil organic carbon density and the influencing factors and the spatial difference thereof, integrates the influence of the initial soil carbon density on the annual variation rate of soil carbon density, and effectively realizes the prediction of the spatio-temporal distribution trend of soil carbon density under different future temperature change scenarios in a region.

Description

Technical Field

[0001] This invention relates to the field of ecological environment governance technology, and in particular to a regional-scale method for predicting the spatiotemporal changes of future soil carbon based on a geographically weighted regression model. Background Technology

[0002] Soil stores 2-3 times more carbon (C) than the atmosphere and vegetation, playing a crucial role in the global C cycle. Even small changes in soil C storage can alter atmospheric CO2 concentrations, thus impacting global climate change. Promoting C fixation in terrestrial ecosystems, particularly through soil C fixation, is considered a vital green pathway to mitigate global warming. Arable land is the largest land use type globally, covering approximately 37% of the land surface. Because soil C flux is relatively lower than that of natural soils such as forests and grasslands, arable land is considered to possess enormous C fixation potential. Therefore, predicting the future spatiotemporal changes in arable land soil C under climate change scenarios (especially global warming) is crucial for developing reasonable long-term arable land management measures to achieve soil carbon fixation and "carbon neutrality" goals.

[0003] Currently, models for predicting future soil carbon (C) changes mainly include process and mechanism models based on soil formation and evolution, such as the Century model, RothC model, and DNDC model, as well as prediction models based on empirical relationships between soil C and influencing factors. Process and mechanism models require numerous and complex parameters for calibration, are easily affected by human intervention, and show significant differences in results when predicting soil C responses to climate change across regions. Different models predict inconsistent magnitudes and even directions of soil C change. Furthermore, existing studies often use extreme treatment conditions (beyond the range of possible climate change in the next few decades), raising questions about the validity of their data extrapolation and potentially leading to biased predictions of future soil carbon evolution trends. Empirical relationship prediction models predict future soil C changes by fitting the relationship between current soil C and its influencing factors. Although they only reflect empirical relationships and are difficult to generalize to other regions, their modeling process is relatively simple, and they can achieve satisfactory results when used to predict future soil C changes in the regions where models are established. For example, multiple linear regression models can be used to establish empirical relationships between various factors and soil C, thereby predicting the trend of soil C changes under different environmental change conditions. However, the relationship between soil C and influencing factors is usually spatially non-stationary. The study area only establishes a multiple linear regression model, whose regression coefficients are consistent in space. This ignores the differences in the influence of various factors on soil C at different spatial locations, which inevitably leads to high uncertainty in the prediction results.

[0004] In geographically weighted regression (GWR) models, the regression coefficients of independent variables vary with spatial location, enabling the detection of different degrees of influence of influencing factors on the target variable at different spatial locations. This allows for a better understanding of the spatial non-stationarity of the relationship between soil properties and various influencing factors. However, changes in soil C are also related to the initial soil C density; as soil C density increases, soil C gradually approaches saturation, and the rate of change slows down. Therefore, accurately predicting future trends in soil C requires incorporating the influence of initial soil C density into the model. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a regional-scale method for predicting the spatiotemporal variation of future soil carbon based on a geographically weighted regression model. This method aims to more accurately express the spatial nonstationarity of the relationship between the rate of soil carbon change and various influencing factors. Furthermore, by introducing the initial soil carbon density and incorporating its influence, the method predicts the rate of soil carbon change and soil carbon density year by year, thereby providing a methodological reference for better predicting the future spatiotemporal variation pattern of soil carbon at the regional scale.

[0006] This invention provides the following technical solution:

[0007] This invention provides a method for predicting the spatiotemporal variation of future soil carbon at a regional scale based on a geographically weighted regression model, which includes the following steps:

[0008] S1: Data Collection

[0009] Collect soil data and influencing factor data for the study area;

[0010] S2: Construct a model for predicting the future spatiotemporal changes of soil carbon and predict the spatiotemporal changes of soil carbon density in different future periods, including the following steps:

[0011] S21: Calculation of the annual rate of change of soil carbon density

[0012] The annual rate of change of soil carbon density was calculated using soil organic or inorganic carbon density from paired sampling points across two historical periods. Based on the number of years between the two historical sampling periods, the annual rate of change of soil organic or inorganic carbon density was calculated for all paired sampling points, denoted as the annual rate of change of soil organic carbon density (SOCD). c and the annual change rate of soil inorganic carbon density SICD c ;

[0013] S22: Selecting auxiliary variables for the prediction model

[0014] Data on factors influencing the annual rate of change of soil organic or inorganic carbon density were selected as auxiliary variables for modeling. Correlation analysis was used to analyze the relationship between the annual rate of change of soil organic or inorganic carbon and the selected influencing factor data and the initial soil organic or inorganic carbon density in the first sampling period. Then, variance expansion factor was used to conduct collinearity analysis of auxiliary variables to select the auxiliary variables to be used for modeling.

[0015] S23: Constructing a predictive model for future spatiotemporal changes in soil carbon

[0016] We used a geographically weighted regression model to establish the empirical relationship between the annual change rate of soil organic or inorganic carbon and each auxiliary variable, and obtained the regression coefficients of the geographically weighted regression model on each auxiliary variable. We used the soil organic or inorganic carbon density of the second period sampling point as the initial soil C density, and predicted the spatial distribution pattern of the annual change rate of soil organic or inorganic carbon density under different temperature change scenarios year by year.

[0017] S24: Prediction of Spatiotemporal Variation of Soil Carbon Density in Future Periods

[0018] The spatiotemporal variation pattern of soil organic or inorganic carbon density under different temperature change scenarios is predicted year by year. The annual change rate of soil organic or inorganic carbon in the predicted year is added to the soil organic or inorganic carbon density in the previous year to obtain the spatiotemporal variation pattern of soil organic or inorganic carbon density in the predicted year.

[0019] According to some implementation methods, in step S22, the data on influencing factors to be screened include parent rock, average annual temperature, average annual rainfall, topography, type of cultivated land use, fertilization intensity, soil type, soil particle composition, and carbon input.

[0020] According to some implementation methods, in step S23, the GWR model is used to estimate the local spatial changes of the study geographical area and detect its spatial nonstationarity. The calculation formula is as follows:

[0021]

[0022] In the formula, Y i Let X be the dependent variable for sample point i. ik Let (u) be the observed value of the k-th auxiliary variable at the i-th point. i ,v i Let ) be the position coordinates of the i-th point, and β0(u i ,v i ) is the intercept, β k (u i ,v i ) represents the regression coefficient at the i-th point, and ε i This is the error term;

[0023] SOCD was established using the GWR model.c and SICD c An empirical relationship model between the variables and auxiliary variables is used to predict SOCD under different climate scenarios at different times in the future. c and SICD c The spatiotemporal changes include the following steps:

[0024] First, a GWR model was constructed with SOCDc and SIDCC as dependent variables and initial auxiliary variables with variance inflation coefficients less than 5 as independent variables. An adaptive Gaussian function was used to fit the regression coefficients of each regression point, the bandwidth was selected by golden section search, and the modified Akaike information content criterion was used as the bandwidth selection criterion.

[0025] Secondly, the MAE, RMSE, and correction determination coefficient between the predicted and measured values ​​of the sample points are calculated and compared with the prediction results of the OLS model to evaluate the prediction accuracy of the GWR model.

[0026] Finally, keeping other auxiliary variables constant, the annual average temperature was increased by 1.5℃ and 2℃ respectively, and the regression coefficients of the GWR model at each sampling point were extracted to predict the spatial distribution patterns of SOCDc and SICDc under the two warming scenarios.

[0027] According to some implementations, step S24 includes the following steps:

[0028] Calculate annual SOCD cn The calculation formula is as follows:

[0029]

[0030] In the above formula, n represents the year; SOCDcn is the annual change rate of soil organic carbon in the nth year; (u i ,v i ) represents the position coordinates of the i-th sample point; SOCD(u i ,v i ) cn The change rate of soil organic carbon at sampling point i in year n is expressed in kg·m³. -2 ·yr -1 ;β0(u i ,v i ) is the intercept at sample point i; SOCD(u i ,v i ) (n-1) The soil organic carbon density at sampling point i in the previous year is expressed in kg·m³. -2 ;X ik βi represents the observed value of the k-th auxiliary variable at the i-th sample point, determined after collinearity detection; β1 represents the regression coefficient of the constructed GWR model on the initial carbon density; βi represents the regression coefficient of the ... k (ui ,v i ) represents the regression coefficient of the k-th auxiliary variable at the i-th sample point in the constructed GWR model; ε i This is the error term;

[0031] Calculate annual SICD cn The calculation formula is as follows:

[0032]

[0033] In the above formula, n represents the year; SICDcn is the annual change rate of soil inorganic carbon in the nth year; (u i ,v i ) represents the position coordinates of the i-th sample point; SICD(u i ,v i ) cn The change rate of soil inorganic carbon at sample point i in year n is expressed in kg·m³. -2 ·yr -1 ;β0(u i ,v i ) is the intercept at sample point i; SICD(u i ,v i ) (n-1) The soil inorganic carbon density at sampling point i in the previous year is expressed in kg·m³. -2 ;X ik βi represents the observed value of the k-th auxiliary variable at the i-th sample point, determined after collinearity detection; β1 represents the regression coefficient of the constructed GWR model on the initial carbon density; βi represents the regression coefficient of the ... k (u i ,v i ) represents the regression coefficient of the k-th auxiliary variable at the i-th sample point in the constructed GWR model; ε i This is the error term;

[0034] Calculate annual SOCD n and SICD n The calculation formula is:

[0035] SOCD(u i v i ) n =SOCD(u i v i ) n-1 +SOCD(u i v i ) cn ×1 (6)

[0036] SICD(u i v i ) n =SICD(ui v i ) n-1 +SICD(u i v i ) cn ×1 (7)

[0037] In the above formula, (u i ,v i ) represents the position coordinates of the i-th sample point; SOCD(u i ,v i ) n and SICD(u i ,v i ) n These represent the soil organic carbon density and soil inorganic carbon density at sampling point i in year n, respectively, in kg·m³. -2 SOCD(u i ,v i ) cn and SICD(u i ,v i ) cn These represent the annual rates of change of soil organic carbon density and soil inorganic carbon density at sampling point i in year n, respectively, in kg·m³. -2 ·yr -1 SOCD(u i ,v i ) (n-1) and SICD(u i ,v i ) (n-1) These are the soil organic carbon density and soil inorganic carbon density at sampling point i in the previous year, respectively, in kg·m³. -2 .

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] (1) This invention uses a geographically weighted regression model to express the relationship between the rate of change of soil C and the changes of various influencing factors with spatial location. In the embodiments, the rate of change of soil organic carbon density (SOCD) is used. c ) and the rate of change of soil inorganic carbon density (SICD) c The coefficient of determination R for the fitting correction) 2 adj The values ​​were 0.370 and 0.557, respectively, which improved by 36.53% and 85.05% compared with the least squares (OLS) model, and more accurately captured the differences in the influence of different factors on soil C variation at different spatial locations.

[0040] (2) This invention introduces the initial soil C density as the model input variable and predicts the soil C change rate and soil carbon density year by year. It fully considers the influence of the initial soil C density and is more in line with the law that the soil C change rate changes with the initial soil C density, so as to more accurately predict the future change trend of soil C. Attached Figure Description

[0041] Figure 1 The location of the Sichuan Basin and the distribution of soil sample points at different times are provided for embodiments of the present invention.

[0042] Figure 2 The correlation coefficients between the annual change rate of soil organic carbon density (a) and the annual change rate of soil inorganic carbon density (b) in cultivated land in the Sichuan Basin and the influencing factors provided in the embodiments of the present invention (Note: *, ** and *** represent significance level (P<0.05), significance level (P<0.01) and highly significant level (P<0.001), respectively).

[0043] Figure 3 The relationship between the predicted and measured values ​​of the annual change rate of soil organic carbon density under the OLS model (a), the annual change rate of soil organic carbon density under the GWR model (b), the annual change rate of soil inorganic carbon density under the OLS model (c), and the annual change rate of soil inorganic carbon density under the GWR model (d) provided in the embodiments of the present invention.

[0044] Figure 4 The spatial distribution of standardized regression coefficients for predicting the rate of change of soil organic carbon density based on the GWR model provided in this embodiment of the invention.

[0045] Figure 5 Spatial distribution of standardized regression coefficients for predicting the rate of change of soil inorganic carbon density based on the GWR model, provided in this embodiment of the invention.

[0046] Figure 6a Statistical characteristics of the annual variation rate of soil organic carbon density under different climatic scenarios provided in embodiments of the present invention.

[0047] Figure 6b The statistical characteristics of the annual variation rate of soil inorganic carbon density under different climatic scenarios are provided in the embodiments of the present invention.

[0048] Figure 7 The spatial distribution of the annual variation rate of soil organic carbon density under different climatic scenarios provided in the embodiments of the present invention.

[0049] Figure 8 The spatial distribution of the annual variation rate of soil inorganic carbon density under different climatic scenarios provided in the embodiments of the present invention.

[0050] Figure 9aThe annual variation trend of soil organic carbon density under different climate scenarios from 2020 to 2060 is provided for embodiments of the present invention.

[0051] Figure 9b The annual variation trend of soil inorganic carbon density under different climate scenarios from 2020 to 2060 is provided for embodiments of the present invention.

[0052] Figure 10 Spatial distribution of annual variation rate of soil organic carbon density under different climate scenarios from 2020 to 2060, provided for embodiments of the present invention.

[0053] Figure 11 The spatial distribution of the annual variation rate of soil inorganic carbon density under different climate scenarios from 2020 to 2060 is provided for embodiments of the present invention.

[0054] Figure 12 The variation trends of soil organic carbon density (a), soil inorganic carbon density (b), and soil carbon density (c) under different climate scenarios from 2020 to 2060 are provided for embodiments of the present invention.

[0055] Figure 13 Spatial distribution of soil organic carbon density under different warming scenarios from 2020 to 2060, provided for embodiments of the present invention.

[0056] Figure 14 Spatial distribution of soil inorganic carbon density under different climate scenarios from 2020 to 2060, provided for embodiments of the present invention.

[0057] Figure 15 Spatial distribution of soil carbon density under different climate scenarios from 2020 to 2060, provided for embodiments of the present invention.

[0058] Figure 16 The flowchart illustrates a regional-scale method for predicting the spatiotemporal variation of future soil carbon based on a geographically weighted regression model, as provided in this embodiment of the invention. Detailed Implementation

[0059] The present invention will now be described in detail with reference to embodiments and accompanying drawings. However, it should be understood that the embodiments and drawings are for illustrative purposes only and do not constitute any limitation on the scope of protection of the present invention. All reasonable modifications and combinations included within the inventive spirit of the present invention fall within the scope of protection of the present invention.

[0060] The present invention will be further described below with reference to the accompanying drawings.

[0061] Example 1

[0062] The Sichuan Basin, with its high proportion of arable land and dense population, plays a vital role in my country's grain production. Previous studies found that the region's average annual temperature rose significantly between the 1980s and 2010s, increasing by an average of 0.36℃ per decade; average annual rainfall decreased slightly but without a significant trend. Of the 111 meteorological stations within the basin, only Yunyang and Neijiang stations showed no significant change in average annual temperature, while the remaining stations showed significant or even extremely significant increases. Regarding average annual rainfall, only eight stations located in the western and southern parts of the basin showed significant changes, while the remaining 103 stations showed no significant changes. The patterns of change in average annual temperature and rainfall in the Sichuan Basin are consistent with the trends of average annual temperature and rainfall across my country.

[0063] Therefore, the regional-scale future soil carbon spatiotemporal variation prediction method based on the geographically weighted regression model provided in this embodiment, taking the Sichuan Basin as the research object, is of typical significance in predicting the spatiotemporal variation pattern of soil C under future climate change scenarios.

[0064] like Figure 16 The prediction method includes the following steps:

[0065] S1: Data Collection

[0066] Collect soil data and influencing factor data for the study area;

[0067] Soil data were obtained from compilations of soil species records from various districts and counties within the basin obtained during the Second National Soil Survey in the 1980s (1980-1985) and from field soil sampling points in the 2010s (2017-2019). Specifically, data from 4219 farmland soil sampling points representative of the main soil species in the basin were obtained in the 1980s. Figure 1 b) These historical soil sampling data record detailed environmental information and soil physicochemical indicators, including the location, topography, parent material, soil type, farmland use type, soil organic matter, and soil pH value of each sampling point. Soil sampling from the 2010s began in 2016 and was completed between 2017 and 2019. Sampling sites were laid out at approximately a 1:1 scale, referencing the locations of soil sampling points from the 1980s. Sampling plots were designed to be as close as possible to the locations of the 1980s sampling points (within the same village or nearby towns), sharing similar farmland use, soil types, and topography. At each sampling location, 4-5 soil samples were collected within a 5m radius of the recorded geographic coordinates. These samples were then thoroughly mixed using a quartering method to form a composite sample, and the latitude, longitude, and altitude were recorded using a handheld GPS. Considering the significant changes in land use over 40 years, additional sampling points were added around some sampling points, resulting in a total of 4409 topsoil (0-20cm) sampling points collected. Figure 1c) Approximately 1 kg of soil sample was collected from each sampling point. After being brought back to the laboratory, impurities were removed, and the sample was allowed to air dry in a ventilated area. After grinding and sieving, the soil particle composition, pH value, organic matter, carbonate, and other physicochemical properties were determined using indoor analytical methods identical to those used in the 1980s. For some missing soil bulk density and carbonate content values, prediction methods were established using random forest models and artificial neural network models to establish relationships between environmental factors and existing soil properties with soil bulk density and carbonate content. To detect outliers, the carbonate content of soil samples from different periods was assessed for outliers by adding or subtracting three times the standard deviation from the mean, for both soil types and cultivated land use types. Samples simultaneously marked as outliers were replaced with the maximum or minimum normal value obtained under both classification methods.

[0068] Due to the long sampling intervals and drastic land use changes, the locations of sampling points in the two periods cannot be completely identical. This embodiment, based on soil genetics theory, pairs soil samples from the two periods according to the principles of "spatial proximity, similar soil type and cultivated land use type, and similar topographic location." Specifically, using soil samples from the 1980s as the center, spatial analysis in ArcGIS software was used to generate buffer distances for these historical samples. Soil samples from 2017 to 2019 with similar topographic location, soil type, and planting system, and the smallest distance from the historical samples were selected as paired soil samples, resulting in 3697 paired samples.

[0069] The influencing factors data were selected based on the soil formation process, using soil-forming factors and soil properties closely related to soil carbon changes as input variables to construct a predictive model for future soil carbon changes. These included parent rock, average annual temperature and rainfall, topographic factors, farmland use type, fertilization intensity (nitrogen fertilizer), soil type, soil mechanical composition, and initial soil carbon density in the 1980s. The average annual temperature and rainfall were calculated from daily temperature and rainfall data from 144 county-level meteorological stations in the study area and its surrounding areas from 1961 to 1980. Figure 1a) First, the annual and seasonal mean temperatures and rainfall values ​​for each meteorological station were calculated. Then, in ArcGIS 10.6 software, ordinary kriging was used to interpolate these values, resulting in raster data of the spatial distribution of annual and seasonal mean temperatures and rainfall values ​​in the Sichuan Basin. Topographic factors were calculated from ALOS 12.5m DEM data, sourced from the Resource and Environmental Science and Data Center of the Chinese Academy of Sciences (https: / / www.resdc.cn / ). Based on this data, topographic factors such as elevation, slope, aspect, plane curvature, profile curvature, and topographic humidity index for the study area were calculated in ArcGIS 10.6 software. Using the fertilizer application rates (calculated per acre for N, P, and K) and crop yields for different planting seasons from 143,552 plots in the study area obtained from the survey, as well as the straw-to-grain ratio, straw-to-root ratio, grain moisture content, carbon content, and straw return ratio of different crops obtained from existing studies, the average annual fertilization intensity and carbon input of each plot were calculated. Based on the principle of complete consistency in planting system and soil type, the fertilizer application rate and carbon input of the plot closest to the soil sampling point were assigned to that soil sampling point.

[0070] S2: Construction of a model for predicting future spatiotemporal changes in soil carbon, including the following steps:

[0071] S21: Calculation of the annual rate of change of soil carbon density

[0072] Soil samples were collected from 1980–1985 and 2017–2019, respectively. Based on the collected data from the Second National Soil Census of various districts and counties in the Sichuan Basin, soil sampling in the 1980s was mainly conducted from 1980 to 1982; therefore, 1982 is used as the initial year for this study. Soil sampling in the 2010s was mainly conducted from 2017 to 2018; therefore, 2018 is used as the final year for this study. Thus, the two sampling periods in this study are approximately 36 years apart. Based on this, the formula for calculating the annual change rate of soil organic carbon (SOCDc) at each sampling point is:

[0073] SOCD c =(SOCD1-SOCD0)÷36 (1)

[0074] In formula (1): SOCD c The annual change rate of soil organic carbon density is expressed in kg·m³. -2 ·yr -1 SOCD0 represents the soil organic carbon density at the beginning of the study period, which in this example is the soil organic carbon density in 1982; SOCD1 represents the soil organic carbon density at the end of the study period, which in this example is the soil organic carbon density in 2018; 36 represents the time interval between soil sampling in the two periods, in yr.

[0075] Annual change rate of soil inorganic carbon at each sampling point (SICD) c The formula for calculating ) is:

[0076] SICD c =(SICD1-SICD0)÷36 (2)

[0077] In equation (2): SICD c The annual change rate of soil inorganic carbon density, in kg·m³. -2 ·yr -1 SICD0 represents the soil inorganic carbon density at the beginning of the study period, which in this example is the soil inorganic carbon density in 1982; SICD1 represents the soil inorganic carbon density at the end of the study period, which in this example is the soil inorganic carbon density in 2018; 36 represents the time interval between soil sampling in the two periods, in yr.

[0078] S22: Select auxiliary variables for the prediction model

[0079] Based on the influencing factor data obtained in S1, factors related to the changes in SOCD and SICD were selected as the SOCD data for the study area. c and SICD c The initial auxiliary variables for the predictive model include climate (annual mean temperature and annual mean rainfall), topography (elevation, slope, aspect, slope length, profile curvature, planar curvature, and topographic humidity index), parent rock, soil type, soil physical properties (sand, silt, and clay), farmland use type, fertilization intensity, C input, and initial SOCD (SOCD0 is the initial year's soil organic carbon density) and SICD (SICD0 is the initial year's soil inorganic carbon density) from the 1980s. To facilitate model calculation, the class means were used instead of class names for qualitative categorical variables such as parent rock, soil type, and farmland use type, and these were used as variable values ​​in model construction. Multicollinearity among the initial variables was analyzed using the variance inflation factor (VIF). Initial variables with a VVIF greater than 5 were removed, and those with a VVIF less than 5 were retained as the final auxiliary variables for the predictive model. This value was set smaller than the commonly used 7.5 to minimize multicollinearity among the final auxiliary variables selected for modeling.

[0080] S23: Construction of a Spatiotemporal Variation Prediction Model for Annual Changes in Soil Carbon Density

[0081] Geographically Weighted Regression (GWR) is an extension of Ordinary Least Squares (OLS) and is a typical local regression model. By embedding the geographical location information of sample data into the regression parameters through weights, the GWR model can estimate the local spatial variation of the dependent variable and effectively detect its spatial nonstationarity. The formula for calculating GWR is as follows:

[0082]

[0083] In equation (3): Y i Let X be the dependent variable for sample point i. ik Let (u) be the observed value of the k-th auxiliary variable at the i-th point. i ,v i Let ) be the position coordinates of the i-th point, and β0(u i ,v i ) is the intercept, β k (u i ,v i ) represents the regression coefficient at the i-th point, and ε i This is the error term.

[0084] SOCD was established using the GWR model. c and SICD c An empirical relationship model between the variables and auxiliary variables is used to predict SOCD under different climate scenarios at different times in the future. c and SICD c The spatiotemporal changes of [the subject / object]. The specific modeling process is as follows:

[0085] (a) A GWR model was constructed with SOCDc and SIDCC as dependent variables and initial auxiliary variables with variance inflation coefficients less than 5 as independent variables. An adaptive Gaussian function was used to fit the regression coefficients of each regression point. The bandwidth was selected by the golden section search and the modified Akaike Information Criterion (AICc) was used as the bandwidth selection criterion.

[0086] (b) Calculate the MAE, RMSE, and adjusted R-squared coefficient of determination between the predicted and measured values ​​of the sample points. 2 adj The prediction results of the GWR model were compared with those of the OLS model to evaluate the prediction accuracy of the GWR model.

[0087] (c) Keeping other auxiliary variables constant, the annual average temperature is increased by 1.5℃ and 2℃ respectively. The regression coefficients of the GWR model constructed in (a) are extracted at each sampling point to predict the spatial distribution patterns of SOCDc and SICDc under the two warming scenarios.

[0088] S24: Prediction of Spatiotemporal Variation of Soil Carbon Density in Future Periods

[0089] Based on the regression coefficients of the constructed GWR model at each sampling point, keeping other auxiliary variables constant, and using 2018 as the base year, the spatiotemporal distribution of SOCDc and SICDc in the study area were predicted annually under three climate scenarios: constant annual average temperature, annual average temperature increase of 1.5℃, and annual average temperature increase of 2℃. SOCDc and SICDc were added to the SOCD and SICD of the previous year to obtain the SOCD and SICD of the study area in the current year, and SOCD and SICD were added to SICD to obtain the STCD of the current year.

[0090] SOCD annually from 2019 to 2060 cn The calculation formula is as follows:

[0091]

[0092] In formula (4): n represents the year; SOCDcn is the annual change rate of soil organic carbon in the nth year; (u i ,v i ) represents the position coordinates of the i-th sample point; SOCD(u i ,v i ) cn The change rate of soil organic carbon at sampling point i in year n is expressed in kg·m³. -2 ·yr -1 ;β0(u i ,v i ) is the intercept at sample point i; SOCD(u i ,v i ) (n-1) The soil organic carbon density at sampling point i in the previous year (i.e., the initial carbon density SOCD0, such as the predicted SOCD0 for 2019) is given. c2019 The initial carbon density values ​​are based on observations from 2017 to 2019, and the units are kg·m³. -2 ;X ik βi represents the observed value of the k-th auxiliary variable at the i-th sample point, determined after collinearity detection; β1 represents the regression coefficient of the constructed GWR model on the initial carbon density; βi represents the regression coefficient of the ... k (u i ,v i ) represents the regression coefficient of the k-th auxiliary variable at the i-th sample point in the constructed GWR model; ε i This is the error term.

[0093] SICD annually from 2019 to 2060 cn The calculation formula is as follows:

[0094]

[0095] In formula (5): n represents the year; SICDcn is the annual change rate of soil inorganic carbon in the nth year; (u i ,v i ) represents the position coordinates of the i-th sample point; SICD(u i ,v i ) cn The change rate of soil inorganic carbon at sample point i in year n is expressed in kg·m³. -2 ·yr -1 ;β0(u i ,v i ) is the intercept at sample point i; SICD(u i ,v i ) (n-1) The soil inorganic carbon density at sampling point i in the previous year (i.e., the initial carbon density SICD0, such as the predicted SICD0 for 2019) is given. c2019 The initial carbon density values ​​are based on observations from 2017 to 2019, and the units are kg·m³. -2 ;X ik βi represents the observed value of the k-th auxiliary variable at the i-th sample point, determined after collinearity detection; β1 represents the regression coefficient of the constructed GWR model on the initial carbon density; βi represents the regression coefficient of the ... k (u i ,v i ) represents the regression coefficient of the k-th auxiliary variable at the i-th sample point in the constructed GWR model; ε i This is the error term.

[0096] SOCD annually from 2019 to 2060 n and SICD n The calculation formula is:

[0097] SOCD(u i v i ) n =SOCD(u i v i ) n-1 +SOCD(u i v i ) cn ×1 (6)

[0098] SICD(u i v i ) n =SICD(ui v i ) n-1 +SICD(u i v i ) cn ×1 (7)

[0099] In equations (6) and (7): (u i ,v i ) represents the position coordinates of the i-th sample point; SOCD(u i ,v i ) n and SICD(u i ,v i ) n These represent the soil organic carbon density and soil inorganic carbon density at sampling point i in year n, respectively, in kg·m³. -2 SOCD(u i ,v i ) cn and SICD(u i ,v i ) cn These represent the annual rates of change of soil organic carbon density and soil inorganic carbon density at sampling point i in year n, respectively, in kg·m³. -2 ·yr -1 SOCD(u i ,v i ) (n-1) and SICD(u i ,v i ) (n-1) These are the soil organic carbon density and soil inorganic carbon density at sampling point i in the previous year, respectively, in kg·m³. -2 .

[0100] The following provides an evaluation of the prediction model parameters and prediction accuracy in the above prediction results of annual soil carbon density change rate. The evaluation process is as follows:

[0101] First, the selection of auxiliary variables:

[0102] The correlation analysis results show that ( Figure 1 The data included annual average temperature, annual average rainfall, elevation, slope, aspect, plan curvature, profile curvature, topographic humidity index, parent rock, soil type, sand, silt, clay, farmland use type, fertilization intensity, C input, and SOCD0 and SOCD of the study area. c Related variables can be used as initial auxiliary variables; SICD of the study area cThe initial auxiliary variables are annual average temperature, annual average rainfall, elevation, slope, aspect, topographic humidity index, parent rock, soil type, sand, silt, clay, farmland use type, fertilization intensity, C input and SICD0.

[0103] To reduce collinearity among auxiliary variables, an OLS model was used to detect collinearity in the initially selected auxiliary variables, and variables with a variance inflation coefficient greater than 5 were removed based on the detection results. As shown in Table 1, 10 factors were ultimately selected as the SOCD for the study area. c The auxiliary variables for the prediction model are annual average temperature, annual average rainfall, slope, topographic humidity index, sand grains, parent rock, soil type, farmland use type, fertilization intensity, and SOCD0; while the SICD of the study area... c The prediction model has eight auxiliary variables: annual average temperature, annual average rainfall, elevation, slope, sand grains, parent rock, fertilization intensity, and SICD0.

[0104] Table 1. Collinearity diagnosis of different auxiliary variables based on the OLS model.

[0105]

[0106]

[0107] Secondly, the prediction accuracy of the soil carbon density annual change rate prediction model was evaluated.

[0108] The SOCD of cultivated land in the Sichuan Basin over the past 40 years was analyzed using both OLS and GWR models. c and SICD c Make predictions ( Figure 3 (Table 2). The results show that the GWR model is effective for SOCD in the study area. c Predicted R 2 adj The value is 0.370, which is better than the OLS model (R²). 2 adj =0.271); MAE and RMSE were reduced by 30.89% and 23.65% respectively compared to the OLS model. Compared to the OLS model, the GWR model improved SICD... c Predicted R 2 adj The GWR value was 0.557, an increase of 85.05%; MAE and RMSE decreased by 32.29% and 26.87%, respectively. This indicates that, compared to the OLS model, the GWR model significantly improved the SOCD of the study area. c and SICD c Both have better predictive performance.

[0109] Table 2 Evaluation of Prediction Model Results

[0110]

[0111] Spatial distribution of regression coefficients of pre-explanatory variables in the model

[0112] GWR model results show that the SOCD of cultivated land in the Sichuan Basin c The explanatory variables have different spatial nonstationarities. Figure 4 When the regression coefficient of an explanatory variable is greater than 0, it indicates that the variable has a positive impact on SOCD. c A positive effect is observed; however, when the regression coefficient of the explanatory variable is <0, it indicates that the variable has a negative effect on SOCDc. Specifically, the parent rocks of the Minjiang and Qujiang river basins have a positive effect on SOCDc. c It has a significant negative impact, while the parent rocks of the Yangtze River main stream have a significant negative impact on SOCD. c The positive impact is the greatest. Figure 4 a) The impact of farmland use type on SOCD c The spatial impact generally increases gradually in a positive direction from the southwest and east of the basin to the north. Figure 4 b), while soil types show a distribution trend from low in the north and south to high in the center. Figure 4 c). The effect of fertilization intensity on SOCD in the Minjiang and Fujiang river basins. c The negative impact is greatest in the central region, but the degree of this impact gradually decreases towards the surrounding areas. Figure 4 d). Sand particles and SOCD0 with SOCD c All have negative effects, and the area with the smallest sand grain regression coefficient is around the basin. Figure 4 e), and the region with the smallest SOCD0 regression coefficient also includes the Chengdu Plain area ( Figure 4 g). The regression coefficients of slope and topographic humidity index show similar spatial distribution patterns, generally exhibiting a "low-medium-low-high-low" banded distribution pattern from northeast to southwest. Figure 4 f and g). Annual average temperature and SOCD c The effects are predominantly negative, with only the northern and southeastern parts of the basin showing positive effects. Figure 4 h). Annual average rainfall is inversely related to annual average temperature; the regression coefficient for annual average rainfall in most areas of the Sichuan Basin is greater than 0, consistent with SOCD. c It has a positive effect; only in parts of the Tuojiang and Jialingjiang river basins in the northwest of the basin is the regression coefficient of annual average rainfall less than 0, i.e., it is consistent with SOCD. c It has a negative effect.

[0113] Sichuan Basin Cultivated Land SICD c The regression coefficients of each explanatory variable are as follows: Figure 5As shown, the results indicate that the spatial nonstationarity of each explanatory variable differs. Specifically, the regression coefficient of the parent rock exhibits a spatial distribution trend of high in the center and low around the edges, and its relationship with SICD is evident in the central basin region. c It has a positive effect ( Figure 5 a) In most areas of the Sichuan Basin, the regression coefficients for fertilization intensity and sand grain size are >0, except for the area near the Chengdu Plain, where the coefficients for SICD are not significantly different. c It has a negative impact ( Figure 5 a) The regression coefficients of elevation and annual average rainfall show similar spatial distribution patterns, and in most areas, the regression coefficients of both are <0, which is similar to SICD. c It has a negative effect ( Figure 5 (d and g). The regression coefficients for slope and annual average temperature generally exhibit a patchy distribution trend, with higher values ​​in the center and lower values ​​around the edges. However, the annual average temperature has a greater influence on SICD. c The area with negative impact is wider. Figure 5 e and f). The regression coefficients of SICD0 are all <0, meaning that SICD0 and SICD... c All of these have negative effects, especially in the Qujiang River Basin, the Minjiang River Basin, and the main stream area of ​​the Yangtze River. Figure 5 h).

[0114] Based on the spatiotemporal variation prediction model of topsoil carbon in the Sichuan Basin constructed in this embodiment, the SOCD of the study area is predicted under the scenarios of constant annual average temperature, an increase of 1.5℃, and an increase of 2℃. c and SICD c The spatiotemporal variation pattern of the Sichuan Basin (Figure 6). The final prediction results are the prediction results under different temperature change scenarios. c Overall, the trend is downward as the annual average temperature increases. Figure 6a When the annual average temperature increases by 1.5℃ and 2℃, the SOCD of the study area... c The mean values ​​decreased by 32.67% and 43.82%, respectively. Meanwhile, the Sichuan Basin SICD... c The trend increases with the rise in average annual temperature. Figure 6b When the annual average temperature increases by 1.5℃ and 2℃, the SICD of the study area... c The average values ​​increased by 2.70 times and 3.63 times, respectively. Analysis shows that the prediction results are consistent with the actual conditions in different regions.

[0115] Predicting the spatial distribution characteristics of SOCDc in the surface layer of cultivated land in the Sichuan Basin under different climatic scenarios. Figure 7 The results show that the SOCD in the study area c It exhibits significant spatial heterogeneity. Under constant annual mean temperature, the SOCD in the southwestern part of the basin... c The SOCD is relatively high in the northeastern part of the basin and around the basin. cLow, or even negative ( Figure 7 a). Under a global warming scenario ( Figure 7 b and c), Sichuan Basin SOCD c The overall trend is downward, SOCD c The number of areas with negative SOC values ​​is also gradually increasing, especially in the northeastern part of the basin. This indicates that climate warming has accelerated the decomposition rate of surface SOC in the Sichuan Basin, and some areas may experience SOC loss due to warming.

[0116] Figure 8 SICD of cultivated land surface layer in Sichuan Basin was displayed c Spatial distribution characteristics under different climate scenarios. Figure 8 It can be seen from a that the SICD of the study area at the current temperature is... c Overall, the SICD is relatively low, but exhibits significant spatial variations. Specifically, the SICD in the central and western part of the basin (near the Chengdu Plain) is... c Higher, while SICD in the northeastern and southern parts of the basin is higher. c Lower ( Figure 8 a). Under a global warming scenario ( Figure 8 b and c), Sichuan Basin SICD c The spatial distribution pattern of SICD in the central basin is undergoing further changes. With the increase in average annual temperature, the SICD in the central basin... c The positive value area gradually expands, and the SIC retention capacity increases; however, the SIC in the northeastern and southwestern parts of the basin... c The decline is significant, further exacerbating the risk of SIC attrition.

[0117] Spatiotemporal projection of annual variation rate of soil carbon density from 2020 to 2060

[0118] Predict SOCD of topsoil in the Sichuan Basin from 2020 to 2060 under different climate scenarios. c and SICD c The spatiotemporal variation trend (Figure 9). From Figure 9a It can be seen that at the current temperature, the SOCD in the study area is... c Slightly lower but not significantly different; as the average annual temperature rises, SOCD... c A significant decrease. And SICD c It increases slowly at the current temperature but remains negative; as the annual average temperature rises, SICD... c From negative to positive ( Figure 9b This indicates that climate warming has a significant and different impact on the changes in SOCD and SICD of cultivated land in the Sichuan Basin. After the annual average temperature increases, the SOC retention capacity in the study area decreases significantly, while the SIC retention capacity further increases.

[0119] Figure 10This study presents the SOCD of the topsoil in the Sichuan Basin under different climate scenarios from 2020 to 2060. c Spatial distribution trend. Under constant annual average temperature ( Figure 10 ae), SOCD study area c The values ​​are mostly positive, indicating that SOCD is generally accumulating. Over time (2020-2060), the SOCD in the study area... c The trend is downward but the change is not significant. This indicates that, assuming the average annual temperature remains constant, the cultivated land in the Sichuan Basin still has the capacity to retain SOC.

[0120] When the annual average temperature rises by 1.5℃ ( Figure 10 fj), SOCD of the study area c The average value is significantly lower than before the warming. Specifically, the SOCD values ​​in the northeastern, central, and southwestern parts of the basin are significantly lower. c All values ​​dropped significantly to negative values; while SOCD values ​​in the western and southern parts of the basin... c The levels remain relatively high or even slightly increased. This indicates that with an average annual temperature increase of 1.5℃, the soil organic matter (SOC) sequestration capacity of cultivated land in the Sichuan Basin decreases and exhibits significant spatial variations.

[0121] When the annual average temperature rises by 2°C ( Figure 10 ko), SOCD of the study area c The mean value continuously decreases when the annual average temperature remains constant or increases by 1.5°C. Spatially, this decrease is observed in the northeastern, central, and southwestern parts of the SOCD basin. c The increased area of ​​the declining region indicates a wider extent of SOC loss; while SOCD in the western and southern parts of the basin... c It remains positive and still has a certain SOC retention capacity.

[0122] SICD of topsoil in Sichuan Basin under different climate scenarios, 2020–2060 c The spatial distribution trend is shown in Figure 6-11. Under the current annual average temperature (Figure 6-11a-e), the SICD in the study area... c Maintaining at a low level, SICD c The average value is generally between -0.01 and 0.01 kg·m. -2 ·yr -1 Between these ranges, and with little change over time (2020–2060). Spatially, the northwestern and central SICD areas of the basin... c Relatively high, while SICD in the southwest and east of the basin. c Relatively low.

[0123] When the annual average temperature increases by 1.5℃ (Figure 6-11f-j), the SICD of the study area... cThe mean value increased significantly compared to when the annual average temperature remained constant, turning from negative to positive, and the spatial distribution showed significant differences. Specifically, the SICD value in the central part of the basin... c Significant growth; while SICD in the northern and southwestern parts of the basin c Significantly lower than before the warming. SICD in the study area from 2020 to 2060. c The average value shows an overall downward trend. This indicates that there are significant differences in the SIC (Sedimentation Capacity) of cultivated land in the Sichuan Basin under different geographical conditions.

[0124] When the annual average temperature increases by 2℃ (Figure 6-11 K-O), the SICD of the study area at different periods... c The mean is significantly higher than when the annual average temperature remains constant or increases by 1.5°C. Over time (2020–2060), SICD... c The mean continues to decline. Spatially, SICD c The spatial distribution pattern is similar to that when the annual average temperature increases by 1.5℃, but exhibits more pronounced spatial differences. This suggests that warming only enhances the soil's ability to retain SiC in the central part of the basin, while further reducing its retention capacity in the northern and southwestern parts of the basin.

[0125] Spatiotemporal distribution pattern of soil carbon density from 2020 to 2060

[0126] Figure 6-12 shows the trends of SOCD, SICD, and STCD in the surface farmland of the Sichuan Basin from 2020 to 2060 under different climate scenarios. Under all climate change scenarios, SOCD in the study area shows an increasing trend (Figure 6-12a); however, with increasing annual average temperature, the accumulation of SOCD in the study area decreases. This indicates that current temperatures are more conducive to SOCD accumulation in the study area.

[0127] As shown in Figure 6-12b, under the condition of constant annual average temperature, the SICD in the study area decreased slightly from 2020 to 2060; with the increase of annual average temperature, the SICD in the study area showed a gradual increasing trend. This indicates that warming can improve the ability of cultivated land in the study area to retain SIC.

[0128] Considering the changes in SOC and SIC, the STCD in the study area showed an increasing trend under different climate scenarios (Figure 6-12c), indicating that the cultivated land soils in the Sichuan Basin will continue to have a "carbon sink" function in the future. Under the scenario of constant annual average temperature, although SIC is lost year by year, the amount of loss is far less than the increase in SOC. Under the warming scenario, although SIC changes from loss to storage, the increase in SOCD is significantly reduced, resulting in a lower cumulative amount of STCD under the warming scenario compared to the scenario of constant annual average temperature, and this difference increases over time.

[0129] Figure 6-13 shows the spatial distribution trend of surface SOCD in cultivated land in the Sichuan Basin from 2020 to 2060 under different climate scenarios. When the annual mean temperature remains constant (Figure 6-13a-e), the SOCD in the study area generally shows a gradual increasing trend, with the mean SOCD increasing by 4.02-13.75%, consistent with statistical results. Spatially, SOCD is relatively high in the northwest and southwest of the study area, while it is relatively low in the central and southeastern parts of the basin.

[0130] When the annual average temperature increased by 1.5℃ (Figure 6-13f-j), the average SOCD value in the study area decreased compared to before the temperature rise, decreasing by 0.46-6.77%, mainly in the southwest, central, and northeastern parts of the basin. From 2020 to 2060, the average SOCD value also showed an overall increasing trend (1.89-6.55%), but the increase was less than when the annual average temperature remained constant. This indicates that with an annual average temperature increase of 1.5℃, the soil SOC sequestration capacity of cultivated land in the Sichuan Basin slightly decreases.

[0131] When the annual average temperature increases by 2℃ (Figure 6-13 K-0), the SOCD in the study area still shows an increasing trend from 2020 to 2060, but the growth rate decreases significantly (1.17–4.13%). Compared with the conditions of constant annual average temperature and an annual average temperature increase of 1.5℃, the average SOCD values ​​in the study area decreased by 0.61–9.02% and 0.15–2.42% respectively in each period. Spatially, the area of ​​SOCD decline in the southwest, central, and northeastern parts of the basin expanded significantly. This indicates that when the annual average temperature increases by 2℃, the SIC (Sodium Carbonate) retention capacity in parts of the Sichuan Basin continues to decrease, and some areas may even experience SOC (Sodium Carbonate) loss.

[0132] Figure 6-13 shows the spatial distribution trend of surface soil SICD in the Sichuan Basin under different climatic scenarios from 2020 to 2060. Under the current temperature (Figure 6-14a-e), the overall SICD in the study area is slightly reduced (0.79-2.16%), consistent with the statistical results. Spatially, the SICD in the study area shows a distribution pattern of high in the center and low around the edges. This indicates that under constant annual average temperature, the soil's ability to retain SIC in the Sichuan Basin's cultivated land does not change significantly.

[0133] When the temperature increased by 1.5℃ (Figure 6-14f-j), the average SICD value in the study area increased compared to before the temperature rise (2.12–39.46%). From 2020 to 2060, the SICD value in the study area showed an overall increasing trend, with the average SICD value increasing by 9.11–33.60%. Spatially, the SICD value in the study area still showed a distribution pattern of high value in the center and low value around the edges, but the high-value area in the central part and the low-value area in the southwest and northeast of the basin expanded significantly. This indicates that when the annual average temperature increases by 1.5℃, the sequestration capacity of SIC in cultivated land in the Sichuan Basin is enhanced, and there are significant spatial differences.

[0134] When the temperature increased by 2℃ (Figure 6-14k-o), the spatial distribution pattern of SICD in the study area did not change significantly.

[0135] The Silicity Index (SICD) of the study area continued to increase from 2020 to 2060, with the growth rate further increasing (12.67–47.17%), consistent with statistical results. Compared to when the annual average temperature remained constant and when the annual average temperature increased by 1.5℃, the average SICD of the study area increased by 2.83–54.69% and 0.69–10.92% respectively during each period. This indicates that under a 2℃ increase in the annual average temperature, the overall capacity of cultivated land in the Sichuan Basin to retain Silicity (SIC) further enhanced. However, in areas with high annual average rainfall (such as the northeastern part of the basin), the risk of SIC loss also increased further due to factors such as leaching.

[0136] Figure 6-15 shows the spatial distribution trend of surface soil carbon dioxide (STCD) in the Sichuan Basin under different climate scenarios from 2020 to 2060. When the annual average temperature remains constant (Figures 6-15a-e), the STCD in the study area generally shows a steady increasing trend over time (2020-2060), with the mean STCD increasing by 3.28-11.33%, consistent with statistical results. Spatially, the STCD is relatively high in the central part of the basin, while it is relatively low in the northeastern and southwestern parts. This indicates that, under constant annual average temperature, the soil carbon fixation capacity of cultivated land in the Sichuan Basin is relatively strong.

[0137] When the annual average temperature increases by 1.5℃ (Figure 6-15f-j), the mean STCD value in the study area decreases slightly compared to when the annual average temperature remains constant (0.06–0.52%). From 2020 to 2060, the STCD in the study area shows an overall increasing trend (3.02–10.79%). Spatially, the STCD distribution pattern is generally similar to that before the warming, but the low-value areas in the central, southwestern, and northeastern parts of the basin have significantly expanded. This indicates that with an annual average temperature increase of 1.5℃, the increase in SICD can only partially offset the loss of SOCD, resulting in a decrease in the carbon fixation capacity of cultivated soil in the Sichuan Basin.

[0138] When the annual average temperature increases by 2℃ (Figure 6-15k-o), the mean STCD value in the study area decreases slightly (0.08–0.44%) compared to when the annual average temperature remains constant. Over time (2020–2060), the STCD in the study area continues to show an increasing trend, but the growth rate slows down (2.99–10.94%), consistent with statistical results. Spatially, the STCD distribution pattern is not significantly different from that when the temperature increases by 1.5℃, but the area of ​​extreme STCD values ​​further expands. This indicates that under a 2℃ increase in annual average temperature, STCD in the central, northeastern, and southwestern parts of the Sichuan Basin is mainly characterized by losses, possibly indicating a "carbon source" effect.

[0139] The above embodiments are merely preferred embodiments of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for predicting the spatiotemporal variation of future soil carbon at a regional scale based on a geographically weighted regression model, characterized in that, Specifically, the following steps are included: S1: Data Collection Collect soil data and influencing factor data for the study area; S2: Construct a model for predicting the future spatiotemporal changes of soil carbon and predict the spatiotemporal changes of soil carbon density in different future periods, including the following steps: S21: Calculation of the annual rate of change of soil carbon density The annual rate of change of soil carbon density was calculated using soil organic or inorganic carbon density from paired sampling points across two historical periods. Based on the number of years between the two historical sampling periods, the annual rate of change of soil organic or inorganic carbon density was calculated for all paired sampling points, denoted as the annual rate of change of soil organic carbon density (SOCD). c and the annual change rate of soil inorganic carbon density SICD c ; S22: Selecting auxiliary variables for the prediction model Data on factors influencing the annual rate of change of soil organic or inorganic carbon density were selected as auxiliary variables for modeling. Correlation analysis was used to analyze the relationship between the annual rate of change of soil organic or inorganic carbon and the selected influencing factor data and the initial soil organic or inorganic carbon density in the first sampling period. Then, variance expansion factor was used to conduct collinearity analysis of auxiliary variables to select the auxiliary variables to be used for modeling. S23: Constructing a predictive model for future spatiotemporal changes in soil carbon We used a geographically weighted regression model to establish the empirical relationship between the annual change rate of soil organic or inorganic carbon and each auxiliary variable, and obtained the regression coefficients of the geographically weighted regression model on each auxiliary variable. Using the soil organic or inorganic carbon density of the sampling point in the second sampling period as the initial soil carbon density, we predicted the spatial distribution pattern of the annual change rate of soil organic or inorganic carbon density under different temperature change scenarios year by year. S24: Prediction of Spatiotemporal Variation of Soil Carbon Density in Future Periods Using the soil organic or inorganic carbon density at the sampling point in the second sampling period as the initial carbon density, and based on the geographically weighted regression model, the soil organic or inorganic carbon density of the previous year is used as the independent variable to iteratively predict the annual change rate of soil organic carbon density or soil inorganic carbon density under different temperature change scenarios in the future. The predicted annual change rate of soil organic or inorganic carbon density is added to the soil organic or inorganic carbon density of the previous year, thereby deducing the spatiotemporal change pattern of soil organic or inorganic carbon density in different future periods year by year.

2. The method for predicting the spatiotemporal variation of future soil carbon at a regional scale based on a geographically weighted regression model according to claim 1, characterized in that, In step S22, the data of influencing factors to be screened include parent rock, average annual temperature, average annual rainfall, topography, type of cultivated land use, fertilization intensity, soil type, soil particle composition and carbon input.

3. The method for predicting the spatiotemporal variation of future soil carbon at a regional scale based on a geographically weighted regression model according to claim 1, characterized in that, In step S23, the GWR model is used to estimate the local spatial changes in the study area and detect its spatial non-stationarity. The calculation formula is as follows: （3） In the formula, Y i Let X be the dependent variable for sample point i. ik Let (u) be the observed value of the k-th auxiliary variable at the i-th point. i ,v i Let ) be the position coordinates of the i-th point, and β0(u i ,v i ) is the intercept, β k (u i ,v i ) represents the regression coefficient at the i-th point, and ε i This is the error term; SOCD was established using the GWR model. c and SICD c An empirical relationship model between the auxiliary variables and the model is used to predict SOCD under different temperature change scenarios at different future periods. c and SICD c The spatiotemporal changes include the following steps: First, with SOCD c and SICD c The GWR model was constructed using the auxiliary variables selected as dependent variables and the variance inflation coefficients of the auxiliary variables as independent variables. The regression coefficients of each regression point were fitted with an adaptive Gaussian function, the bandwidth was selected by the golden section search, and the modified Akaike information content criterion was used as the bandwidth selection criterion. Secondly, the MAE, RMSE, and correction determination coefficient between the predicted and measured values ​​of the sample points are calculated and compared with the prediction results of the OLS model to evaluate the prediction accuracy of the GWR model. Finally, keeping other auxiliary variables constant, the annual average temperature was increased by 1.5℃ and 2℃ respectively, and the regression coefficients of the GWR model at each sampling point were extracted to predict SOCD under the two warming scenarios. c and SICD c Spatial distribution pattern.

4. The method for predicting the spatiotemporal variation of future soil carbon at a regional scale based on a geographically weighted regression model according to claim 3, is characterized in that... Step S24 includes the following steps Calculate annual SOCD cn The calculation formula is as follows: （4） In the above formula, n represents the year; SOCD cn Let u be the annual change rate of soil organic carbon in year n; i ,v i ) represents the position coordinates of the i-th sample point; SOCD(u i ,v i ) cn β0(u) represents the rate of change of soil organic carbon at sampling point i in year n, in kg·m⁻²·yr⁻¹; i ,v i ) is the intercept at sample point i; SOCD(u i ,v i ) （n-1） X represents the soil organic carbon density at sampling point i in the previous year, expressed in kg·m⁻². ik The observed value of the auxiliary variable at the i-th sample point k after collinearity detection; The regression coefficients of the constructed GWR model on the initial carbon density; β k (u i ,v i ) represents the regression coefficient of the k-th auxiliary variable at the i-th sample point in the constructed GWR model; ε i This is the error term; Calculate annual SICD cn The calculation formula is as follows: （5） In the above formula, n represents the year; SICD cn Let u be the annual change rate of soil inorganic carbon in year n; i ,v i ) represents the position coordinates of the i-th sample point; SICD(u i ,v i ) cn β0(u) represents the rate of change of soil inorganic carbon at sampling point i in year n, in kg·m⁻²·yr⁻¹; i ,v i ) is the intercept at sample point i; SICD(u i ,v i ) （n-1） X represents the soil inorganic carbon density at sampling point i in the previous year, in kg·m⁻². ik β1 represents the observed value of the k-th auxiliary variable at the i-th sample point, determined after collinearity detection; β2 represents the regression coefficient of the constructed GWR model on the initial carbon density; β3 represents the regression coefficient of the GWR model on the initial carbon density. k (u i ,v i ) represents the regression coefficient of the k-th auxiliary variable at the i-th sample point in the constructed GWR model; ε i This is the error term; Calculate annual SOCD n and SICD n， The calculation formula is: （6） （7） In the above formula, (u i ,v i ) represents the position coordinates of the i-th sample point; SOCD(u i ,v i ) n and SICD(u i ,v i ) n These represent the soil organic carbon density and soil inorganic carbon density at sampling point i in year n, respectively, in kg·m⁻²; SOCD(u i ,v i ) cn and SICD(u i ,v i ) cn , respectively, represent the annual change rates of soil organic carbon density and soil inorganic carbon density at sampling point i in year n, in kg·m⁻²·yr⁻¹; SOCD(u i ,v i ) （n-1） and SICD(u i ,v i ) （n-1） These are the soil organic carbon density and soil inorganic carbon density at sample point i in the previous year, respectively, in kg·m⁻².

Images