A method for identifying and modeling the driving mechanism of temperature effect on cropland change

Abstract

This invention belongs to the field of remote sensing data processing and land surface temperature monitoring technology, and relates to a method for identifying and attributing the driving mechanism of temperature effects on cultivated land. The method includes: collecting and preprocessing multi-source data; constructing an index of temperature change in cultivated land relative to the background land surface; constructing a set of dominant driving factors; identifying the dominant driving factors and constructing a sample set; introducing an intrinsic biophysical mechanism model to construct a comprehensive attribution model, and using the constructed sample set for model training and validation; using the validated comprehensive attribution model to perform seasonal-scale contribution decomposition, comparative analysis of typical transformation types, identification of irrigation cooling efficiency thresholds, and assessment of crop cooling potential. This invention can quantify the modulating effect of agricultural management measures on cultivated land temperature effects, achieving synergistic attribution of natural and anthropogenic processes. It maintains the physical constraints of energy balance while capturing the influence of management factors through a parameterized scheme, improving attribution accuracy and interpretability.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing data processing and land surface temperature monitoring technology, specifically involving a method for identifying and attributing the driving mechanism of temperature effect on farmland change. Background Technology

[0002] Arable land is the land use type most heavily influenced by human activities. Arable land change includes the conversion of arable land to other land types, as well as changes in crop types, irrigation patterns, and vegetation growth within arable land. These changes disrupt the surface energy balance by altering surface biophysical properties (albedo, roughness, evapotranspiration, and heat flux distribution), thereby causing local surface temperature variations. Accurately understanding the formation mechanism of the temperature effect of arable land change, quantifying the relative contributions of radiative and non-radiative processes (such as evapotranspiration and turbulent exchange), and analyzing the moderating effects of climate background and agricultural management practices are of significant scientific importance for assessing the agricultural climate effects and guiding adaptive management.

[0003] However, the temperature effect on cultivated land is the result of the synergistic effects of multiple biophysical processes, and is influenced by both natural background and human management, resulting in a complex driving mechanism. Current research lacks a systematic and explainable attribution framework. Existing methods for attributing changes in land surface temperature mainly fall into the following three categories: (1) The first type is based on statistical regression. By establishing a multiple linear regression model of surface temperature and multiple potential driving factors (such as albedo, vegetation index and evapotranspiration), the regression coefficients are used to explain the contribution of each factor. This method is simple and easy to implement, but it is difficult to handle the nonlinear relationship and collinearity between factors, and it lacks physical mechanism constraints, resulting in insufficient reliability and interpretability of the attribution results.

[0004] The second category is based on energy balance decomposition methods, such as the Intrinsic Biophysical Mechanism (IBM) model, which decomposes surface temperature changes into two parts: radiation-driven (albedo and downdraft radiation changes) and non-radiation-driven (evapotranspiration, roughness, and aerodynamic impedance changes). This method has clear physical meaning, but it usually assumes that each factor acts independently, does not fully consider the interaction effects between factors, and does not adequately characterize agricultural management factors (such as irrigation and crop type), thus failing to quantify the modulating effect of human management on farmland temperature.

[0005] The third category is importance assessment methods based on machine learning, such as models like random forests and XGBoost, which can be used to identify dominant driving factors and quantify their contributions. These methods can handle nonlinear relationships, but the models lack interpretability, are difficult to separate the physical contributions of radiative and non-radiative processes, and do not provide sufficient explicit modeling of management factors, thus failing to provide direct scientific basis for agricultural management decisions.

[0006] In summary, the existing technology has the following main shortcomings: (1) Existing attribution models mostly focus on natural factors and lack explicit representation of agricultural management measures (irrigation intensity, crop type and farming system), and cannot quantify the modulating effect of human management on temperature; (2) There are complex interaction effects between management factors and natural factors (such as the difference in evapotranspiration response to temperature under different crop types). Existing models are unable to capture such interactions, resulting in reduced attribution accuracy. (3) The dominant driving factors and their contributions vary with climate zones, agricultural zones and seasons. Existing studies lack systematic spatial differentiation and seasonal dynamic analysis, which limits the regional application of attribution results; (4) The cooling effect of irrigation on the surface temperature has a saturation phenomenon, but existing studies have failed to identify the key threshold of the saturation phenomenon, which limits the scientific basis for the optimal allocation of agricultural water resources.

[0007] Therefore, there is an urgent need to develop a method for identifying and attributing the temperature effect driving mechanism of farmland change that can couple natural and human processes, fully consider factor interactions, reveal regional differentiation and seasonal patterns, and identify irrigation cooling efficiency thresholds. Summary of the Invention

[0008] To address the aforementioned technical problems, the present invention proposes a solution. The present invention provides a method for identifying and attributing the temperature effect driving mechanism of cultivated land change, comprising: Collect and preprocess multi-source data; the multi-source data includes remote sensing surface temperature data, radiation factor data, hydrothermal factor data, agricultural management data, and auxiliary data; the auxiliary data includes topographic data, climate zoning data, and agricultural zoning data. Using a spatiotemporal substitution method, an index of temperature change of cultivated land relative to the background reference surface is constructed based on remote sensing surface temperature data. Construct a set of dominant driving factors; the dominant driving factors include radiation factors, hydrothermal factors, and agricultural management factors; Identify the dominant driving factors, pair temperature change indicators with each dominant driving factor in time series to construct a sample set, and conduct spatial differentiation analysis of each dominant driving factor according to the region to reveal the differences in dominant driving factors in different climate zones and agricultural areas, providing a basis for targeted agricultural management in the region. Based on the principle of surface energy balance, an intrinsic biophysical mechanism model is introduced to decompose the temperature change index into radiation-driven and non-radiation-driven terms. A comprehensive attribution model is constructed and trained using a sample set. Simultaneously, model validation and sensitivity analysis are performed. Using the validated integrated attribution model, seasonal-scale contribution decomposition, comparative analysis of typical transformation types, identification of irrigation cooling efficiency thresholds, and assessment of crop cooling potential were performed.

[0009] In some optional embodiments, constructing a temperature change index of cultivated land relative to the background surface includes: selecting target pixels within a local window whose topographic features are similar to the background climate features and whose land cover type is the background reference land type, and calculating the average temperature of the target pixels; assuming the temperature change index is... The surface temperature of cultivated land is The average temperature of the background reference surface pixels is The temperature change index of the target pixel is expressed as: .

[0010] In some alternative embodiments, radiation factors include albedo and downwave radiation; hydrothermal factors include evapotranspiration, soil moisture, and leaf area index; and agricultural management factors include irrigation intensity, crop type, and crop rotation pattern.

[0011] In some optional embodiments, dominant driving factors are identified, and sample sets are constructed by pairing temperature change indicators with each dominant driving factor in time series. Spatial differentiation analysis of each dominant driving factor is then performed according to the region, including: screening factors related to temperature change indicators through Pearson correlation analysis; determining dominant factors by ranking features by importance; and analyzing the spatial differentiation of dominant driving factors by climate zone and agricultural zone.

[0012] In some alternative embodiments, based on the principle of surface energy balance, an intrinsic biophysical mechanism model is introduced, including: Let the surface temperature be... The surface albedo is Downlink shortwave radiation is Downlink longwave radiation is Upward longwave radiation is The sensible heat flux is The latent heat flux is The Stefan-Boltzmann constant is Surface emissivity ,but: ; Let the temperature change index be The surface temperature of cultivated land is The background reference surface temperature is ,but ;set up The difference in albedo between cultivated land and the background reference surface. The difference in downwave radiation between cultivated land and the background reference surface. The difference in evapotranspiration between cultivated land and the background reference surface. To represent the aerodynamic impedance difference between the cultivated land and the background reference surface, the energy balance equations for the cultivated land and the background reference surface are subtracted and linearized to obtain the decomposition form of the temperature change index. Retaining the radiation-driven term, evapotranspiration term, and aerodynamic impedance term, it is expressed as: .

[0013] In some alternative embodiments, radiation-driven terms include albedo and downdraft radiation; non-radiation-driven terms include evapotranspiration, soil moisture, and vegetation structure parameters.

[0014] In some optional embodiments, an irrigation impact factor is introduced to calculate evapotranspiration, assuming the actual evapotranspiration is... Potential evaporation The irrigation impact coefficient is Irrigation intensity is Irrigation intensity reflects the proportion of increase in evapotranspiration per unit of irrigation intensity. Therefore, evapotranspiration after considering the increase in irrigation intensity is expressed as: .

[0015] In some optional embodiments, the irrigation impact coefficient is calibrated as follows: under the same climatic background and crop type, paired pixels of irrigated farmland and adjacent rainfed farmland are selected, with evapotranspiration differences caused by irrigation; the irrigation impact coefficient is set as follows. Evapotranspiration in the irrigation area Evapotranspiration in rain-fed areas Irrigation intensity is ,but: .

[0016] In some optional embodiments, a comprehensive attribution model is constructed, including: set up To account for changes in evapotranspiration after enhanced irrigation, To account for changes in soil moisture, the irrigation intensity is: , Leaf area index, For crop type variables, Let be the interaction term between leaf area index and crop type, used to characterize the nonradiative moderating effect of different crop canopy structures on land surface temperature. The comprehensive attribution model is then expressed as: .

[0017] In some optional embodiments, a validated integrated attribution model is used to perform seasonal-scale contribution decomposition, comparative analysis of typical transformation types, identification of irrigation cooling efficiency thresholds, and assessment of crop cooling potential. Seasonal contribution decomposition includes: quantifying the contribution of each driving factor to temperature change indicators at the monthly scale, and identifying the switching patterns of dominant factors in different seasons. A comparative analysis of typical conversion types was conducted, including the selection of three typical conversion types: forest land to cultivated land, grassland to cultivated land, and paddy field to dry land. The differences in net temperature effects and biophysical mechanisms among the different conversion types were compared. Irrigation cooling efficiency threshold identification includes: simulating temperature change indicators by varying irrigation intensity, plotting irrigation cooling efficiency curves, and identifying the saturation point of efficiency decline; Crop cooling potential assessment includes: comparing differences in temperature change indices among different crop types under the same background, and incorporating the interaction term between leaf area index and crop type. (Used to characterize the nonradiative regulation of surface temperature by different crop canopy structures, among which) Leaf area index, dimensionless; As a crop type variable, we assess the potential of crop type selection to moderate local temperature and regional differences.

[0018] The beneficial effects of this invention are: (1) This invention is the first to explicitly introduce irrigation correction coefficient and crop type interaction term under the energy balance framework, and constructs an attribution model that couples natural and human processes. It can quantify the modulating effect of agricultural management measures on the temperature effect of cultivated land, fill the gap in the insufficient characterization of management factors in the prior art, and realize the synergistic attribution of natural and human processes. (2) The present invention adopts a combination of physical mechanism model and data-driven approach, which maintains the physical constraints of energy balance and captures the influence of management factors through parameterization scheme, thereby improving the accuracy and interpretability of attribution and making the attribution results more reliable. (3) This invention clarifies the dominant factors and transformation patterns of different climate zones and seasons through zoning analysis and seasonal scale decomposition, providing a basis for the regional application of attribution results; (4) This invention proposes an irrigation cooling efficiency threshold identification method, which identifies the saturation point of marginal benefit decline by simulating the temperature response under different irrigation intensities, providing a quantitative scientific basis for the optimal allocation of agricultural water resources and the formulation of irrigation policies; (5) This invention can be used to assess the potential impact of different crop layouts and irrigation expansion on local climate, and guide the selection of agricultural measures to adapt to climate change, which has important application value. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of a method for identifying and attributing the temperature effect driving mechanism of farmland change, provided in Embodiment 1 of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] Example 1 As an example, in order to solve the problems existing in the prior art, this embodiment provides a method for identifying and attributing the driving mechanism of temperature effect on farmland change.

[0022] The implementation details of the method in this embodiment are described below. The following content is only for the convenience of understanding and is not necessary for implementing this solution.

[0023] The method of this embodiment can be applied to electronic devices with communication, computing, and data storage capabilities. (See attached...) Figure 1 As shown, the method provided in this embodiment includes steps 110-160.

[0024] Step 110: Collect and preprocess multi-source data; multi-source data includes remote sensing surface temperature data, radiation factor data, hydrothermal factor data, agricultural management data, and auxiliary data; auxiliary data includes topographic data, climate zoning data, and agricultural zoning data.

[0025] Specifically, all data were uniformly converted to the WGS84 geographic coordinate system through projection transformation; all data were uniformly resampled to a 5km spatial resolution using bilinear interpolation, which matches the scale of the attribution model and balances computational efficiency and spatial accuracy; all data were uniformly converted to a monthly time resolution through time matching to ensure consistency of different data sources over time; outliers in the data were removed using the 3σ criterion, and missing values ​​were filled using neighbor interpolation or time series interpolation.

[0026] Step 120: Using the spatiotemporal replacement method, construct an index of temperature change of cultivated land relative to the background reference surface based on remote sensing surface temperature data.

[0027] Specifically, constructing a temperature change index for cultivated land relative to a background reference surface includes: selecting background reference surface pixels within a local window that have similar topographic and climatic features and whose land cover type is the reference land class; calculating the average temperature of the background reference surface pixels; and assuming the temperature change index is... The surface temperature of cultivated land is The average temperature of the background reference surface pixels is The temperature change index of the target pixel is then expressed as: .

[0028] The method for determining the background reference surface is as follows: (1) For each 5km resolution farmland pixel, establish a local sliding window of 30km×30km; (2) Filter the pixels that meet the following conditions in the window as background reference pixels: the land cover type is forest or grassland and remains stable during the study period; the elevation difference between the background reference pixel and the central cultivated land pixel does not exceed 100m; the slope is less than 5° to avoid the influence of the terrain on the surface temperature.

[0029] (3) Calculate the average surface temperature of all background reference pixels that meet the conditions, and use it as the background reference surface temperature of the cultivated land pixel; (4) Calculate the temperature change index of the entire study area pixel by pixel using a sliding window.

[0030] Step 130: Construct a set of dominant driving factors; the dominant driving factors include radiation factors, hydrothermal factors, and agricultural management factors.

[0031] In some alternative embodiments, radiation factors include albedo and downwave radiation; hydrothermal factors include evapotranspiration, soil moisture, and leaf area index; and agricultural management factors include irrigation intensity, crop type, and crop rotation pattern.

[0032] Albedo is the proportion of shortwave solar radiation reflected by the Earth's surface, directly affecting the radiant energy absorbed by the surface. Evapotranspiration is the sum of surface water evaporation and plant transpiration, regulating surface temperature through latent heat exchange. Soil moisture affects soil heat capacity and thermal conductivity, and also influences the evapotranspiration process. Leaf area index reflects the density of the vegetation canopy, affecting surface albedo, roughness, and evapotranspiration.

[0033] Irrigation intensity ranges from 0 to 1, with 1 for irrigated areas and 0 for rainfed areas, representing the degree to which irrigation improves water conditions. Crop type is a categorical variable, including major crop types such as maize, wheat, and rice, with different crops exhibiting different canopy structures and physiological characteristics. Crop rotation pattern represents the planting system of cultivated land, such as single-crop or double-crop annually, influencing vegetation growth cycles and ground cover.

[0034] This invention employs a combination of physical mechanism models and data-driven approaches, which maintains the physical constraints of energy balance while capturing the influence of management factors through parameterization schemes, thereby improving attribution accuracy and interpretability and making the attribution results more reliable.

[0035] Step 140: Identify the dominant driving factors, pair temperature change indicators with each dominant driving factor in time series to construct a sample set, and conduct spatial differentiation analysis of each dominant driving factor according to the region to reveal the differences in dominant driving factors in different climate zones and agricultural areas, providing a basis for targeted agricultural management in the region.

[0036] In some optional embodiments, dominant driving factors are identified, and sample sets are constructed by pairing temperature change indicators with each dominant driving factor in time series. Spatial differentiation analysis of each dominant driving factor is then performed according to the region, including: screening factors related to temperature change indicators through Pearson correlation analysis; determining dominant factors by ranking features by importance; and analyzing the spatial differentiation of dominant driving factors by climate zone and agricultural zone.

[0037] Specifically, the temperature change index obtained in step 120 is paired with the dominant driving factors constructed in step 130 on a time series basis to construct a sample set. The Pearson correlation coefficient between the temperature change index and each driving factor is calculated, and factors that are significantly correlated at a Pearson correlation coefficient level of less than 0.05 are selected as preliminary dominant driving factors.

[0038] The dominant factors are identified by ranking features by importance. This involves constructing a random forest regression model with temperature change as the dependent variable and all significantly relevant driving factors as independent variables. The features importance scores output by the random forest regression model are used to rank the factors and identify the dominant driving factors. The random forest model can handle nonlinear relationships and collinearity issues among factors, improving the accuracy of key factor identification. In practical applications, in addition to random forest importance ranking, the Boruta algorithm, recursive feature elimination, or SHAP value analysis can also be used to identify dominant driving factors. The Boruta algorithm can more robustly identify all relevant factors; the RFE method can recursively eliminate unimportant factors to obtain the optimal subset of factors; and SHAP value analysis can provide the contribution of each factor to each sample, providing better interpretability.

[0039] The study area was divided into arid, semi-arid, and humid zones based on climate zones. Spatial differentiation analysis was conducted within each climate zone to compare the differences in dominant driving factors across different climate zones. The study area was also divided into different agricultural regions based on agricultural zones. Zonal analysis clarified the spatial differentiation patterns of dominant driving factors, providing a basis for targeted regional agricultural management.

[0040] Step 150: Based on the principle of surface energy balance, an intrinsic biophysical mechanism model is introduced to decompose the temperature change index into radiation-driven and non-radiation-driven terms, construct a comprehensive attribution model, and use the constructed sample set to train the comprehensive attribution model for parameter fitting. At the same time, model validation and sensitivity analysis are performed.

[0041] This step is based on the principle of surface energy balance. It introduces an improved Intrinsic Biophysical Mechanism (IBM) model to decompose temperature change indicators into radiation-driven and non-radiation-driven contributions. It also explicitly introduces a parameterization scheme for agricultural management factors to construct a comprehensive attribution model. Simultaneously, model validation and sensitivity analysis are conducted.

[0042] Among them, based on the principle of surface energy balance, an intrinsic biophysical mechanism model is introduced, including: Let the surface temperature be... The surface albedo is Downlink shortwave radiation is Downlink longwave radiation is Upward longwave radiation is The sensible heat flux is The latent heat flux is The Stefan-Boltzmann constant is Surface emissivity ,but: ; Let the temperature change index be The surface temperature of cultivated land is The background reference surface temperature is Background reference land surface refers to stable pixels selected within a local window that have similar terrain and climate to cultivated land pixels and whose land cover type is woodland or grassland. The energy balance equations for cultivated land and the background reference surface are subtracted and linearized to obtain the decomposition form of the temperature change index. Retaining the radiation-driven term, evapotranspiration term, and aerodynamic impedance term, it is expressed as: .

[0043] in, The difference in albedo between cultivated land and the background reference surface. The difference in downwave radiation between cultivated land and the background reference surface. The difference in evapotranspiration between cultivated land and the background reference surface. This represents the aerodynamic impedance difference between the cultivated land and the background reference surface. Aerodynamics (unit: s / m) characterizes the resistance to turbulent exchange between the surface and the atmosphere, affecting the transport efficiency of sensible and latent heat fluxes. Aerodynamic impedance can be estimated using the wind profile formula: ; in, Let be the Karman constant (approximately 0.4), u be the wind speed, z be the reference height, and d be the zero-plane displacement. The length of the surface roughness. This is the stability correction function.

[0044] In practical applications, in addition to the IBM model, physical models based on surface energy balance systems can also be used for attribution.

[0045] This invention constructs an attribution model that couples natural and anthropogenic processes by explicitly introducing irrigation correction coefficients and crop type interaction terms within an energy balance framework. This model can quantify the modulating effect of agricultural management measures on the temperature effect of cultivated land, filling the gap in the insufficient characterization of management factors in existing technologies and achieving synergistic attribution of natural and anthropogenic processes.

[0046] Specifically, radiation-driven factors include albedo and downdraft radiation; non-radiative factors include evapotranspiration, soil moisture, and vegetation structure parameters.

[0047] By further decomposing the non-radiative driving terms into evapotranspiration, soil moisture, and vegetation structure terms, the impact of non-radiative processes can be characterized in greater detail.

[0048] In this calculation, an irrigation impact coefficient is introduced, and the actual evapotranspiration is assumed to be... Potential evaporation The irrigation impact coefficient is (Dimensionless), irrigation intensity is The irrigation intensity value ranges from 0 to 1, with 1 for irrigated areas and 0 for rain-fed areas. Irrigation intensity reflects the proportion of increase in evapotranspiration per unit of irrigation intensity. Therefore, evapotranspiration after considering the increase in irrigation intensity is expressed as: .

[0049] The calibration method for the irrigation impact coefficient is as follows: Under the same climatic background and crop type, paired pixels of irrigated farmland and adjacent rainfed farmland are selected, and the difference in evapotranspiration is caused by irrigation; let the irrigation impact coefficient be... Evapotranspiration in the irrigation area Evapotranspiration in rain-fed areas Irrigation intensity is ,but: .

[0050] In practical applications, either the irrigation experimental data method or the process model simulation method can be used. The irrigation experimental data method estimates evapotranspiration by utilizing the differences in evapotranspiration under different irrigation levels observed in field controlled experiments; the process model simulation method uses land surface process models such as the VIP model and SiB2 to simulate the differences in evapotranspiration under different irrigation scenarios, thus indirectly estimating evapotranspiration, and is suitable for areas lacking observational data.

[0051] Irrigation significantly enhances evapotranspiration by increasing soil moisture supply, thereby producing a cooling effect. Different crop types have different canopy structures (height, leaf tilt angle distribution, and leaf area index, etc.), which affect surface roughness and aerodynamic impedance, and thus influence the distribution of sensible and latent heat fluxes. An interaction term between leaf area index and crop type is introduced. This interaction term characterizes the modulation of nonradiative processes by crop type through canopy structure. It captures the differences in leaf area index response to temperature effects across different crop types, improving the accuracy of characterizing nonradiative contributions.

[0052] In addition to using the interaction term between leaf area index and crop type Alternatively, crop type one-hot encoding can be used to construct a product term with the leaf area index, or a hierarchical model can be used to fit the response functions of different crop types separately. One-hot encoding can more flexibly capture the differences between different crop types; hierarchical models can improve the fitting accuracy of the model, but have higher computational complexity.

[0053] Step 160: Using the validated integrated attribution model, perform seasonal-scale contribution decomposition, comparative analysis of typical transformation types, identification of irrigation cooling efficiency thresholds, and assessment of crop cooling potential.

[0054] The construction of a comprehensive attribution model includes: set up To account for changes in evapotranspiration after enhanced irrigation, To account for changes in soil moisture, the irrigation intensity is: , Leaf area index, For crop type variables, Let be the interaction term between leaf area index and crop type, used to characterize the nonradiative moderating effect of different crop canopy structures on land surface temperature. The comprehensive attribution model is then expressed as: .

[0055] The model can take the form of a multiple linear regression model or a random forest machine learning model. Multiple linear regression models have clear physical meaning, and the regression coefficients can directly explain the contribution of each factor, making them suitable for interpretive analysis. Random forest models can handle nonlinear relationships and interaction effects between factors, making them suitable for prediction and the detection of complex interactions.

[0056] This invention prioritizes the use of a multiple linear regression model to maintain physical interpretability, while utilizing a random forest model to verify the existence of nonlinear relationships and their impact on attribution results.

[0057] Using observational data from typical farmland stations, including irrigated farmland, rain-fed farmland, and paddy field stations, as well as surface temperature observation data from agricultural meteorological stations, the accuracy of the temperature change indicators estimated by the model was validated. The coefficient of determination (R²) and root mean square error (RMSE) were calculated to evaluate the model's fit and predictive ability.

[0058] Sensitivity analysis of key parameters was conducted using the perturbation method: the irrigation influence coefficient was perturbed within ±20% of its calibration value, and the changes in the estimated results of temperature change indicators were observed; the leaf area index and soil moisture were perturbed within ±10% and ±15% of their observed values, respectively, to assess their impact on the model results. Through sensitivity analysis, the sensitivity intervals of each parameter were determined, the uncertainty of the model was assessed, and a reference was provided for the application of the model.

[0059] Specifically, using the validated integrated attribution model, seasonal-scale contribution decomposition, typical transformation type comparative analysis, irrigation cooling efficiency threshold identification, and crop cooling potential assessment were conducted. Seasonal-scale contribution decomposition includes: quantifying the contribution of each driving factor to temperature change indicators at the monthly scale, and identifying the transition patterns of dominant factors in different seasons; by quantifying the contribution of each dominant driving factor to temperature change indicators at the monthly scale, calculating the contribution ratio of each factor in different months, and identifying the transition patterns of dominant factors in different seasons. For example, during the growing season (May-September), evapotranspiration is vigorous, and it is usually the dominant factor for the temperature effect of cultivated land; while during the non-growing season (October-April of the following year), vegetation cover is low, and albedo changes become the dominant factor. The specific steps are as follows: (1) The monthly surface temperature data and driving factor data were divided into four groups according to the month: spring (March-May), summer (June-August), autumn (September-November) and winter (December-February); (2) For each season, the contribution value of each driving factor is calculated using the trained comprehensive attribution model. If the model adopts a multiple linear regression form, then the driving factors... Contribution value The calculation is as follows: ; in, The regression coefficient for this factor represents the change in the factor of cultivated land relative to the background reference surface. Temperature change index. It can be decomposed into the sum of the contribution values ​​of each driving factor and the residual terms: ; in, The residual term represents the amount of temperature change that the comprehensive attribution model failed to explain. The smaller the value of the residual term, the better the model fits. (3) Calculate the contribution value of each driving factor to the total contribution in each season ( or The proportions of each season can be used to create bar charts or pie charts to visually represent the dominant factors in each season. (4) By comparing the changes in dominant factors in different seasons, we can identify the transformation patterns and provide a scientific basis for agricultural management in different seasons.

[0060] A comparative analysis of typical conversion types was conducted, including: selecting conversions between forest and cultivated land, grassland and cultivated land, and paddy fields and dry land; comparing the output results of the comprehensive attribution model; and analyzing the differences in biophysical mechanisms among different conversion types. For example, conversion from forest to cultivated land typically leads to increased albedo (producing a cooling effect) and decreased evapotranspiration (producing a warming effect), with the net effect depending on the climate zone and season; while conversion from paddy fields to dry land primarily produces a warming effect by reducing evapotranspiration. The specific steps are as follows: (1) Based on two or more periods of land cover data, extract three types of conversion pixels: forest land to cultivated land, grassland to cultivated land, and paddy field to dry land; (2) For each converted pixel, calculate the changes in each driving factor before and after the conversion, input the data into the model to obtain the net temperature change and the contribution share of each factor, and compare the differences in the biophysical mechanisms of different conversion types. (3) Through comparative analysis, the impact mechanism of different transformation types on local climate is revealed, providing a scientific basis for farmland protection and utilization policies.

[0061] Irrigation cooling efficiency threshold identification includes: simulating temperature change indicators by varying irrigation intensity, plotting an irrigation cooling efficiency curve, and identifying the saturation point where efficiency declines. Specifically, it identifies the point where the curve slope begins to flatten significantly, i.e., the irrigation intensity saturation point where the marginal cooling benefit begins to decrease significantly. This irrigation intensity saturation point provides a quantitative basis for the optimal allocation of agricultural water resources. Increasing irrigation intensity below the saturation point yields significant cooling benefits, while further increasing irrigation intensity above the saturation point provides limited cooling benefits. The specific steps are as follows: (1) Select a typical irrigation area (such as the North China Plain), keep other driving factors unchanged, and adjust the irrigation intensity. Discretize the data from 0 to 1 with a set step size (e.g., 0.05), and substitute the values ​​into the model to calculate the corresponding temperature change index. ,draw With irrigation intensity The changing irrigation cooling efficiency curve; the curve is usually steep at first and then flat, indicating that the marginal cooling effect of irrigation gradually decreases. (2) Calculate the marginal cooling efficiency of the curve, and its first derivative (difference approximation): ; Where δ is the step size; (3) When the absolute value of marginal efficiency is lower than the preset threshold (e.g., 0.05 °C / 0.1, meaning that when the temperature drop is less than 0.05 °C for every 0.1 increase in irrigation intensity, the benefit is considered insignificant), the corresponding The value represents the saturation point at which irrigation cooling efficiency decreases, providing a basis for agricultural water resource allocation.

[0062] Crop cooling potential assessment includes: comparing differences in temperature change indices among different crop types under the same background, combined with... The regression coefficients of the interaction terms assess the potential of crop type selection to moderate local temperature and regional differences, providing a scientific basis for optimizing crop layout in different regions. For example, due to its high evapotranspiration rate, rice typically has a strong cooling potential; in areas with abundant water resources, expanding the rice planting area can alleviate the local heat island effect. The specific steps are as follows: (1) Select cultivated land pixels of different crop types (such as rice, wheat and corn) under the same climate background, topography and irrigation pattern; (2) Encode crop types and correlate them with leaf area index. Constructing interactive items Substitute the values ​​into the comprehensive attribution model for regression analysis to obtain the regression coefficients of the interaction terms; (3) Let This is an indicator of temperature change for the first crop type. This serves as an indicator of temperature change for the second crop type. The regression coefficients for the interaction term, For the first crop type, the crop type variable is... For the second crop type variable, the differences in temperature change indices between different crop types under the same LAI are compared using the following formula: ; By comparing the differences in temperature change indices of different crop types under the same LAI, the cooling potential and spatial distribution of different crop types can be assessed, providing a basis for crop layout optimization.

[0063] This invention, through zonal analysis and seasonal-scale decomposition, clarifies the dominant factors and transformation patterns in different climatic zones and seasons, providing a foundation for the regional application of attribution results. This invention can be used to assess the potential impacts of different crop layouts and irrigation expansion on local climate, guiding the selection of agricultural adaptation measures to climate change, and has significant application value.

[0064] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying and attributing the temperature-driven mechanism of farmland change, characterized in that, include: Collect and preprocess multi-source data; the multi-source data includes remote sensing surface temperature data, radiation factor data, hydrothermal factor data, agricultural management data, and auxiliary data; Supporting data include topographic data, climate zoning data, and agricultural zoning data; Using a spatiotemporal substitution method, an index of temperature change of cultivated land relative to the background reference surface is constructed based on remote sensing surface temperature data. Construct a set of dominant driving factors; The dominant driving factors include radiation factors, hydrothermal factors, and agricultural management factors; Identify the dominant driving factors, pair temperature change indicators with each dominant driving factor in time series to construct a sample set, and perform spatial differentiation analysis of each dominant driving factor according to partitions. Based on the principle of surface energy balance, an intrinsic biophysical mechanism model is introduced to decompose the temperature change index into radiation-driven and non-radiation-driven terms. A comprehensive attribution model is constructed and trained using a sample set. Simultaneously, model validation and sensitivity analysis are performed. Using the validated integrated attribution model, seasonal-scale contribution decomposition, comparative analysis of typical transformation types, identification of irrigation cooling efficiency thresholds, and assessment of crop cooling potential were performed.

2. The method for identifying and attributing the temperature effect driving mechanism of farmland change according to claim 1, characterized in that, Constructing a temperature change index for cultivated land relative to a background reference surface includes: selecting background reference surface pixels within a local window that have similar topographic and climatic features and whose land cover type is the reference land type; calculating the average temperature of the background reference surface pixels; and assuming the temperature change index is... The surface temperature of cultivated land is The average temperature of the background reference surface pixels is The temperature change index of the target pixel is then expressed as: 。 3. The method for identifying and attributing the temperature effect driving mechanism of cultivated land change according to claim 2, characterized in that, Radiation factors include albedo and downwave radiation; hydrothermal factors include evapotranspiration, soil moisture, and leaf area index; agricultural management factors include irrigation intensity, crop type, and crop rotation pattern.

4. The method for identifying and attributing the temperature effect driving mechanism of farmland change according to claim 1, characterized in that, The dominant driving factors were identified, and a sample set was constructed by pairing temperature change indicators with each dominant driving factor in the time series. Spatial differentiation analysis of each dominant driving factor was then performed according to the region, including: screening factors related to temperature change indicators through Pearson correlation analysis; determining dominant factors by ranking feature importance; and analyzing the spatial differentiation of dominant driving factors by climate zone and agricultural zone.

5. The method for identifying and attributing the temperature effect driving mechanism of farmland change according to claim 1, characterized in that, Based on the principle of surface energy balance, an intrinsic biophysical mechanism model is introduced, including: Let the surface temperature be... The surface albedo is Downlink shortwave radiation is Downlink longwave radiation is Upward longwave radiation is The sensible heat flux is The latent heat flux is The Stefan-Boltzmann constant is Surface emissivity ,but: ； Let the temperature change index be The surface temperature of cultivated land is The background reference surface temperature is ,but ;set up The difference in albedo between cultivated land and the background reference surface. The difference in downwave radiation between cultivated land and the background reference surface. The difference in evapotranspiration between cultivated land and the background reference surface. To represent the aerodynamic impedance difference between the cultivated land and the background reference surface, the energy balance equations for the cultivated land and the background reference surface are subtracted and linearized to obtain the decomposition form of the temperature change index. Retaining the radiation-driven term, evapotranspiration term, and aerodynamic impedance term, it is expressed as: 。 6. The method for identifying and attributing the temperature effect driving mechanism of farmland change according to claim 5, characterized in that, Radiation-driven factors include albedo and downdraft radiation; non-radiation-driven factors include evapotranspiration, soil moisture, and vegetation structure parameters.

7. The method for identifying and attributing the temperature effect driving mechanism of cultivated land change according to claim 6, characterized in that, The irrigation impact factor is introduced to calculate evapotranspiration, assuming the actual evapotranspiration is... Potential evaporation The irrigation impact coefficient is Irrigation intensity is Irrigation intensity reflects the proportion of increase in evapotranspiration per unit of irrigation intensity. Therefore, evapotranspiration after considering the increase in irrigation intensity is expressed as: 。 8. The method for identifying and attributing the temperature effect driving mechanism of cultivated land change according to claim 7, characterized in that, The calibration method for the irrigation impact coefficient is as follows: Under the same climatic background and crop type, paired pixels of irrigated farmland and adjacent rainfed farmland are selected, and the difference in evapotranspiration is caused by irrigation; let the irrigation impact coefficient be... Evapotranspiration in the irrigation area Evapotranspiration in rain-fed areas Irrigation intensity is ,but: 。 9. The method for identifying and attributing the temperature effect driving mechanism of cultivated land change according to claim 1, characterized in that, Constructing a comprehensive attribution model includes: set up To account for changes in evapotranspiration after enhanced irrigation, To account for changes in soil moisture, the irrigation intensity is... , Leaf area index, For crop type variables, Let be the interaction term between leaf area index and crop type, used to characterize the nonradiative moderating effect of different crop canopy structures on land surface temperature. The comprehensive attribution model is then expressed as: 。 10. The method for identifying and attributing the temperature effect driving mechanism of cultivated land change according to claim 1, characterized in that, Seasonal contribution decomposition includes: quantifying the contribution of each driving factor to temperature change indicators at the monthly scale, and identifying the switching patterns of dominant factors in different seasons. A comparative analysis of typical conversion types was conducted, including the selection of three typical conversion types: forest land to cultivated land, grassland to cultivated land, and paddy field to dry land. The differences in net temperature effects and biophysical mechanisms among the different conversion types were compared. Irrigation cooling efficiency threshold identification includes: simulating temperature change indicators by varying irrigation intensity, plotting irrigation cooling efficiency curves, and identifying the saturation point of efficiency decline; The assessment of crop cooling potential includes comparing the differences in temperature change indicators among different crop types under the same background, and evaluating the potential of crop type selection to regulate local temperature and regional differences.

Images