A method and device for assessing terrace damage and loss of function

Abstract

This application proposes a method and apparatus for assessing terraced field damage and functional loss, relating to the field of ecological assessment. The method includes: dividing the target area into terraced patches based on terraced field spatial distribution data to obtain assessment units; expanding physical features and screening factors based on initial multidimensional driving factor data to obtain standardized feature vectors; inputting the standardized feature vectors into a multi-task neural network for damage prediction to obtain predicted damage probability and predicted damage area ratio; multiplying the predicted damage probability and predicted damage area ratio to obtain the expected damage intensity; acquiring terraced field type and erosion factor set; dynamically correcting engineering measures factors and analyzing soil loss based on terraced field type, expected damage intensity, and erosion factor set to obtain the newly added soil erosion; and acquiring sediment transport ratio and soil data set; performing functional loss analysis based on the newly added soil erosion, sediment transport ratio, and soil data set to obtain the functional loss assessment results.

Description

Technical Field

[0001] This application relates to the field of ecological assessment, and in particular to a method and apparatus for assessing damage and functional loss of terraced fields. Background Technology

[0002] Terraced fields are an important soil and water conservation engineering measure in my country, playing an irreplaceable role in intercepting surface runoff, reducing soil erosion, and improving soil fertility. However, due to factors such as frequent extreme rainstorms, insufficient field maintenance, and human activities, structural damage problems such as terrace embankment collapse and field surface damage are becoming increasingly prominent. Accurately assessing the degree of terrace damage and its impact on soil and water conservation functions is a prerequisite for scientifically formulating terrace restoration and management plans.

[0003] Existing methods for assessing terraced field damage and functional loss cannot simultaneously predict the probability of damage and the extent of spatial coverage, making it difficult to comprehensively reflect the damage status of terraced fields; they cannot respond to dynamic changes in the degree of terraced field damage, and there is a lack of a seamless transmission mechanism between damage identification and soil erosion; and a multidimensional functional loss analysis mechanism has not been established, resulting in low accuracy in assessing terraced field damage and functional loss, which affects the restoration and management of terraced fields. Summary of the Invention

[0004] To address the aforementioned problems, in a first aspect, the present invention provides a method for assessing the damage and functional loss of terraced fields, comprising: Acquire the spatial distribution data of terraced fields in the target area, and divide the target area into terraced patches based on the spatial distribution data to obtain evaluation units; Obtain the initial multidimensional driving factor data of the evaluation unit, and perform physical feature expansion and factor screening based on the initial multidimensional driving factor data to obtain a standardized feature vector; The standardized feature vector is input into a multi-task neural network for damage prediction to obtain the predicted damage probability and the predicted damage area ratio. The predicted damage probability and the predicted damage area ratio are multiplied to obtain the expected damage intensity. Obtain the terrace type and erosion factor set of the assessment unit, and perform dynamic correction of engineering measures factors and soil loss analysis based on terrace type, expected damage intensity and erosion factor set to obtain the newly added soil erosion amount; Obtain the sediment transport ratio and soil data set of the assessment unit, and conduct functional loss analysis based on the newly added soil erosion, sediment transport ratio and soil data set to obtain the functional loss assessment results.

[0005] In one embodiment, the target area is divided into terraced patches based on the spatial distribution data of terraced fields to obtain evaluation units, including: Acquire the digital elevation model data of terraced fields, slope classification standards, spectral characteristics of field boundaries, topographic texture characteristics, spatial continuity characteristics, and field ridge dividing lines from the spatial distribution data of terraced fields; Based on the slope grading standard, the spectral characteristics of field boundaries and the topographic texture characteristics, the terrace patch boundaries of the digital elevation model data of terraced fields are identified to obtain the vector data of terrace patch boundaries. Based on the spatial continuity characteristics and the field ridge dividing lines, the boundary vector data of the terraced fields are divided into independent fields to obtain multiple initial terraced field patch units; The boundaries of the initial terraced patch units are smoothed and topological consistency is screened to obtain evaluation units.

[0006] In one embodiment, initial multidimensional driving factor data of the evaluation unit is obtained, and physical feature expansion and factor screening are performed based on the initial multidimensional driving factor data to obtain a standardized feature vector, including: For each assessment unit, initial multidimensional driving factor data including natural environment, ecological characteristics and human activity disturbance are obtained, and initial physical characteristic extended variables including slope micro-topography and hydrological characteristics are calculated based on the initial multidimensional driving factor data. By using a composite screening index weighted by point-binary correlation coefficient and mutual information, factor screening is performed on the initial multidimensional driving factor data and the initial physical feature extended variables to obtain optimized multidimensional driving factor data and optimized physical feature extended variables. By optimizing multidimensional driving factor data and physical feature extension variables, a standardized feature vector for the evaluation unit is formed.

[0007] In one embodiment, the multi-task neural network includes a shared feature extraction layer, a first fully connected layer, a second fully connected layer, a damage probability prediction layer, and a damage extent prediction layer. The process of obtaining the predicted damage probability and the predicted damage area ratio includes: The standardized feature vector is input into the shared feature extraction layer, and the standardized feature vector is recalibrated through a self-attention mechanism to obtain attention-enhanced features; The attention-enhanced features are sequentially input into the first fully connected layer and the second fully connected layer, and the output of the second fully connected layer is added to the input of the first fully connected layer through a residual connection to obtain the shared hidden features; The shared hidden features are input into the damage probability prediction layer to predict the damage probability. The shared hidden features are input into the damage prediction layer to predict the damaged area and obtain the predicted proportion of the damaged area.

[0008] In one embodiment, dynamic correction of engineering measures and soil loss analysis are performed based on terrace type, expected damage intensity, and erosion factor set to obtain the newly added soil erosion amount, including: Based on the terrace type, determine the normal state engineering measures factor for the assessment unit under normal conditions, and the completely damaged state engineering measures factor and degradation response parameters under completely damaged conditions; A continuous nonlinear correction function is constructed based on the engineering measure factors under normal conditions, the engineering measure factors under completely damaged conditions, and the degradation response parameters. The expected damage intensity is substituted into the continuous nonlinear correction function as the independent variable to calculate the change in the engineering measure factors. Obtain the rainfall erosivity factor, soil erodibility factor, slope length factor, slope factor, biological measure factor, and tillage measure factor from the erosivity factor set. Substitute the changes in engineering measure factor, rainfall erosivity factor, soil erodibility factor, slope length factor, slope factor, biological measure factor, and tillage measure factor into the soil loss model to calculate the newly added soil erosion modulus of the assessment unit. Obtain the area of ​​the assessment unit, multiply the newly added soil erosion modulus by the area, and obtain the newly added soil erosion amount of the assessment unit.

[0009] In one embodiment, a functional loss analysis is performed based on the increased soil erosion, sediment transport ratio, and soil dataset to obtain a functional loss assessment result, including: The effective mass of sediment lost into river channels and valleys is calculated by multiplying the newly added soil erosion amount by the sediment transport ratio. Obtain the topsoil bulk density, topsoil price, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price from the soil dataset; The composite ecological function loss of the assessment unit is calculated based on the effective sediment mass, topsoil bulk density, topsoil price, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price. The sum of the combined ecological function losses of all assessment units is taken as the total functional loss, and the functional loss assessment result is determined based on the total functional loss.

[0010] In one embodiment, the calculation process for the combined ecological function loss includes: The loss of topsoil resource substitution function is calculated based on the effective sediment mass, topsoil bulk density, and topsoil price. The loss of soil nutrient retention ecological function is calculated based on the effective sediment mass, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price. The loss of topsoil resource substitution function and the loss of soil nutrient retention ecological function are added together to obtain the composite ecological function loss.

[0011] Secondly, the present invention provides a device for assessing the damage and functional loss of terraced fields, used to implement the aforementioned method for assessing the damage and functional loss of terraced fields, the device comprising: The evaluation unit acquisition module is used to acquire the spatial distribution data of terraced fields in the target area, and to divide the target area into terraced patches based on the spatial distribution data of terraced fields to obtain evaluation units. The standardized feature vector acquisition module is used to acquire the initial multidimensional driving factor data of the evaluation unit, and to perform physical feature expansion and factor screening based on the initial multidimensional driving factor data to obtain the standardized feature vector. The expected damage intensity acquisition module is used to input the standardized feature vector into the multi-task neural network for damage prediction, obtain the predicted damage probability and the predicted damage area ratio, and multiply the predicted damage probability and the predicted damage area ratio to obtain the expected damage intensity. A new soil erosion acquisition module has been added to obtain the terrace type and erosion factor set of the assessment unit. Based on the terrace type, expected damage intensity and erosion factor set, dynamic correction of engineering measures factors and soil loss analysis are performed to obtain the new soil erosion amount. The module for obtaining functional loss assessment results is used to acquire the sediment transport ratio and soil data set of the assessment unit. Based on the newly added soil erosion, sediment transport ratio and soil data set, functional loss analysis is performed to obtain functional loss assessment results.

[0012] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for assessing the damage and functional loss of terraced fields.

[0013] Fourthly, the present invention provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for assessing the damage and functional loss of terraced fields.

[0014] The present invention has the following beneficial effects: 1. The initial multidimensional driving factor data is processed into standardized feature vectors. Damage prediction is performed on the standardized feature vectors through a multi-task neural network, outputting the predicted damage probability and the predicted damage area ratio. The expected damage intensity is obtained by multiplying the two, realizing a unified quantitative representation of the probability of damage and the degree of spatial coverage. According to the terrace type and the expected damage intensity, the engineering measure factors are dynamically corrected to obtain the new soil erosion, realizing the seamless transfer of the terrace structure damage state to the physical model of soil and water loss. Combined with the sediment transport ratio and soil dataset, multidimensional functional loss analysis is performed on the new soil erosion, which significantly improves the accuracy of terrace damage and functional loss assessment.

[0015] 2. Initial multidimensional driving factor data were obtained for each evaluation unit, and physical feature extension variables characterizing slope micro-topography and hydrological properties were calculated, enriching the physical information dimension of the driving factors of terrace damage. A composite screening index was used to remove redundancy from the high-dimensional factor space, reducing feature redundancy while retaining key information, resulting in information-condensed and numerically stable standardized feature vectors, effectively improving the convergence efficiency and generalization performance of subsequent neural network training. A multi-task neural network was constructed, with a shared feature extraction layer capturing the long-range dependencies between driving factors in the standardized feature vectors through a self-attention mechanism. The first and second fully connected layers, in conjunction with residual connections, extracted shared hidden features, alleviating the gradient vanishing problem in deep network training. Furthermore, by sharing the underlying feature representation, the damage probability prediction layer and the damage degree prediction layer could utilize the correlation between tasks to improve joint prediction accuracy.

[0016] 3. By using the expected damage intensity as the independent variable to drive continuous nonlinear correction, the change in engineering measure factors is obtained. These factors, along with the erosion factor set, are then substituted into the soil loss model to calculate the new soil erosion. This achieves a continuous nonlinear gradual response of engineering measure factors with damage intensity, and quantitatively connects the damage identification results with the soil loss model. The new soil erosion is multiplied by the sediment transport ratio to obtain the effective sediment mass lost into the river channel and valley. Combined with the soil data set, the composite ecological function loss is calculated. All units are summed to obtain the total functional loss, and the functional loss assessment result is determined accordingly. This establishes a quantitative conversion method from physical erosion to ecological value loss. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method according to an embodiment of the present invention; Figure 2 This is a structural diagram of the device according to an embodiment of the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0020] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to and includes any or all possible combinations of one or more of the listed items.

[0021] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0022] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0023] Reference Figure 1 The present invention provides a method for assessing the damage and functional loss of terraced fields, including steps S100-S500.

[0024] S100 acquires the spatial distribution data of terraced fields in the target area, divides the target area into terraced patches based on the spatial distribution data, and obtains evaluation units.

[0025] In some embodiments, a typical terraced field distribution area in the Loess Plateau (hereinafter referred to as the "study area") is used as the implementation area to assess the damage to terraced fields and the loss of soil and water conservation functions within the area. The total area of ​​the study area is approximately The terraced fields include three types: stone terraces, earthen terraces, and horizontal terraces. The soil erosion in this area is mainly caused by water erosion, making it a typical representative area for conducting terraced field damage assessments.

[0026] In some embodiments, the target area is divided into terraced patches based on the spatial distribution data of terraced fields to obtain evaluation units, including: Acquire the digital elevation model data of terraced fields, slope classification standards, spectral characteristics of field boundaries, topographic texture characteristics, spatial continuity characteristics, and field ridge dividing lines from the spatial distribution data of terraced fields; Based on the slope grading standard, the spectral characteristics of field boundaries and the topographic texture characteristics, the terrace patch boundaries of the digital elevation model data of terraced fields are identified to obtain the vector data of terrace patch boundaries. Based on the spatial continuity characteristics and the field ridge dividing lines, the boundary vector data of the terraced fields are divided into independent fields to obtain multiple initial terraced field patch units; The boundaries of the initial terraced patch units are smoothed and topological consistency is screened to obtain evaluation units.

[0027] In some embodiments, the slope grading standard adopts the terrace slope grading system in the People's Republic of China industry standard "Classification and Grading Standard for Soil Erosion" (SL190-2007); the spectral characteristics of the field boundary include the normalized difference vegetation index (NDVI). ) and Normalized Difference Water Index (NDDI) The threshold range of ) Threshold range is , Threshold range is Topographic texture features include local slope variation coefficient and profile curvature.

[0028] An object-oriented segmentation method is employed, overlaying digital elevation model data with multispectral remote sensing imagery. A multi-scale segmentation algorithm is then used, setting the shape factor to... Tightness factor is Initial patch boundaries are generated. The initial patch boundaries are then thresholded using spectral and topographic texture features: retaining... exist interval and exist The boundaries of patches within the interval, and removing patches with a local slope variation coefficient less than [value missing]. The patch boundaries are used to filter out interfering patches that are not part of the terraced land, and obtain the vector data of the terraced patch boundaries.

[0029] Spatial continuity is defined as the distance between the boundaries of adjacent patches being less than 100%. And the height difference between patches is less than The dividing line between the paddy fields is based on the difference in elevation between the paddy fields being greater than 1. And the angle of abrupt slope change is greater than The location of the terraced patches is extracted. Based on the above characteristics, the boundary vector data of the terraced patches is divided into multiple independent fields, resulting in multiple initial terraced patch units.

[0030] The boundaries of the initial terraced patch units were smoothed using cubic spline interpolation to remove jagged edges. The smoothing parameter was set to [value missing]. Topological consistency screening is performed on the smoothed boundaries to remove overlapping and gap patches, and patches with areas smaller than [a certain size] are also considered. Furthermore, spatially fragmented patches that are too far from other patches are deleted to ensure that each evaluation unit is a spatially closed and topologically consistent independent computational unit. After the above processing, a total of [number] evaluation units were obtained for the study area. Each terraced field spatial assessment unit is denoted as... ( Each evaluation unit Each element has attribute information, including element number, spatial location, boundary vector, and total area. (unit: and terraced field types This provides a unified spatial basis for the subsequent unit-by-unit calculations.

[0031] S200 acquires the initial multidimensional driving factor data of the evaluation unit, performs physical feature expansion and factor screening based on the initial multidimensional driving factor data, and obtains standardized feature vectors.

[0032] In some embodiments, initial multidimensional driving factor data of the evaluation unit is obtained, and physical feature expansion and factor screening are performed based on the initial multidimensional driving factor data to obtain a standardized feature vector, including: For each assessment unit, initial multidimensional driving factor data including natural environment, ecological characteristics and human activity disturbance are obtained, and initial physical characteristic extended variables including slope micro-topography and hydrological characteristics are calculated based on the initial multidimensional driving factor data. By using a composite screening index weighted by point-binary correlation coefficient and mutual information, factor screening is performed on the initial multidimensional driving factor data and the initial physical feature extended variables to obtain optimized multidimensional driving factor data and optimized physical feature extended variables. By optimizing multidimensional driving factor data and physical feature extension variables, a standardized feature vector for the evaluation unit is formed.

[0033] In some embodiments, for each evaluation unit We obtained its initial multidimensional driving factor data, which includes three dimensions: natural environment factors, ecological characteristic factors, and human activity disturbance factors, comprehensively covering various factors that affect the stability of terraced structures and soil and water conservation functions.

[0034] Natural environmental factors include: average annual rainfall (unit: Rainfall erosivity factor (unit: Soil erodibility factors (unit: ), slope length factor (Dimensionless), Slope Factor (dimensionless), aspect (unit: ), altitude (unit: ), terrain relief (unit: Soil effective cohesion (unit: ), soil density (unit: Terraced fields with high ridges (unit: ), pore water pressure (unit: Effective internal friction angle (unit: Ecological characteristic factors include: Normalized Difference Vegetation Index (NDVI). (dimensionless), vegetation cover (unit: Soil organic matter mass fraction (unit: Human activity interference factors include: distance from the road (unit: ), distance from residential areas (unit: ), land use type (post-processing of categorical variable coding).

[0035] The initial multidimensional driving factor data sources included: rainfall-related data from multi-year continuous observation data of meteorological stations in and around the study area; soil parameters from field sampling and indoor geotechnical test analysis results; topographic parameters extracted from digital elevation model data; vegetation parameters derived from multispectral remote sensing image inversion; and human activity-related factors extracted from land use maps and basic geographic information data. All data were uniformly projected and transformed to the same coordinate system before entering factor screening to achieve accurate spatial overlay.

[0036] Extended variables of physical characteristics include topographic humidity index Geometric critical stability index of terraced slopes These two variables respectively characterize the water catchment and retention characteristics of the assessment unit and the critical stability state of the slope under gravity, deepening the physical meaning of the original topographic factors, including the topographic humidity index. The calculation formula is:

[0037] In the formula, As an evaluation unit upstream catchment area (unit: ), As an evaluation unit Terraced field slope (unit: ), For a small positive number, take This is used to avoid the denominator being zero.

[0038] Geometric critical stability index of terraced slopes The calculation formula is:

[0039] The meanings and units of the symbols in the formula are the same as before. This index reflects the stability margin of the terraced slope under given geotechnical and geometric parameters. The larger the value, the closer the slope is to the critical instability state.

[0040] Composite screening indicators A measure that combines linear correlation (point-bivariate correlation coefficient) and nonlinear correlation (mutual information) can more comprehensively evaluate the contribution of each feature to terrace damage. For any feature... Composite screening indicators The calculation formula is:

[0041]

[0042] In the formula, Features in damaged samples The mean, Features in normal samples The mean, Features The standard deviation in the whole sample. For the number of damaged samples, This is the normal sample size. Total sample size ( ); Features With damaged labels Mutual information between them Features Information entropy To adjust the weights, it is used to control the relative contribution of the two types of screening indicators in the composite indicator.

[0043] In this embodiment, based on existing field survey data of the study area (including...) One damaged sample and (Number of normal samples) That is, linear and nonlinear screening indicators are equally weighted in the comprehensive evaluation. The calculation of each feature... Value, by Sort values ​​from largest to smallest and retain Features, remove The characteristics were used to filter out redundant factors with weak correlation to terrace damage. After screening, from the initial... Retain from each factor Several optimization factors, including: rainfall erosivity factor Soil erodibility factors Slope length factor Slope factor Terraced slope Terraced fields with high ridges Topographic humidity index Geometric critical stability index of terraced slopes Normalized Difference Vegetation Index , Soil organic matter content, distance from the road, distance from residential areas.

[0044] Filtered Several optimization factors are used to optimize the multidimensional driving factor data, and each feature is then analyzed. - Standardization is applied to eliminate the influence of different units and orders of magnitude on subsequent neural network training, making each feature numerically comparable. For features... The standardized formula is:

[0045]

[0046] In the formula, As a standardized feature, Features The mean across all evaluation units, Features Standard deviation across all assessment units. (All) The standardized feature values ​​are arranged in a fixed order to form an evaluation unit. Standardized feature vectors , for The column vector completes the full transformation from raw multi-source data to normalized numerical input.

[0047] The S300 inputs a standardized feature vector into a multi-task neural network to predict damage, obtaining the predicted damage probability and the predicted damage area ratio. The predicted damage probability and the predicted damage area ratio are then multiplied to obtain the expected damage intensity.

[0048] In some embodiments, the multi-task neural network includes a shared feature extraction layer, a first fully connected layer, a second fully connected layer, a damage probability prediction layer, and a damage extent prediction layer. The process of obtaining the predicted damage probability and the predicted damage area ratio includes: The standardized feature vector is input into the shared feature extraction layer, and the standardized feature vector is recalibrated through a self-attention mechanism to obtain attention-enhanced features; The attention-enhanced features are sequentially input into the first fully connected layer and the second fully connected layer, and the output of the second fully connected layer is added to the input of the first fully connected layer through a residual connection to obtain the shared hidden features; The shared hidden features are input into the damage probability prediction layer to predict the damage probability. The shared hidden features are input into the damage prediction layer to predict the damaged area and obtain the predicted proportion of the damaged area.

[0049] In some embodiments, the parameters of each layer of the multi-task neural network are configured as follows: the shared feature extraction layer includes a multi-head self-attention module, and the number of attention heads is set to... Dimensions of each attention head The first fully connected layer contains One neuron; the second fully connected layer contains The damage probability prediction layer contains 10 neurons; One neuron, using Activation function; damage prediction layer includes One neuron, using Activation function.

[0050] The shared feature extraction layer uses a self-attention mechanism to process the standardized feature vectors. Perform feature recalibration and attention-enhanced features. The calculation formula is:

[0051] In the formula, As an evaluation unit The standardized feature vector, , , These are the query weight matrices (all dimensions are...) ), key weight matrix (all dimensions are...) ) and value weight matrix (all dimensions are ), In this embodiment, the scaling factor is used. . This is a row-wise normalized exponential function. After self-attention computation... The dimension is dimension.

[0052] Attention-enhancing features Input to the first fully connected layer, the first fully connected layer contains There are neurons, and the activation function is... Output The calculation formula is:

[0053] In the formula, The weight matrix of the first fully connected layer (dimension 1) ), The bias vector of the first fully connected layer ( dimension), The output of the first fully connected layer ( dimension). To modify the activation function of the linear unit, .

[0054] Will Input the second fully connected layer, the second fully connected layer contains There are neurons, and the activation function is... The output of the second fully connected layer is denoted as... The calculation formula is:

[0055]

[0056] In the formula, The weight matrix of the second fully connected layer (dimension 1) ), The bias vector of the second fully connected layer ( dimension), To share hidden features ( (Dimension). In this formula, the output of the second fully connected layer is the same as the input of the first fully connected layer. The residuals are connected and added together to form a residual structure.

[0057] Shared hidden features Input damage probability prediction layer Predicted probability of damage The calculation formula is:

[0058] In the formula, The weight vector of the damage probability prediction layer ( dimension), This is the bias term (scalar) for the damage probability prediction layer. for Type activation function, . The range of values ​​is , indicating the evaluation unit The probability of damage to the terraced field structure.

[0059] Shared hidden features Input damage prediction layer, output evaluation unit The predicted proportion of damaged area The calculation formula is:

[0060] In the formula, The weight vector of the damage prediction layer ( dimension), This is the bias term (scalar) for the damage prediction layer. The range of values ​​is , indicating the evaluation unit The predicted proportion of damaged area.

[0061] Evaluation Unit Expected damage intensity Expected damage intensity It also encodes the probability information of damage occurrence and the spatial coverage information of damage occurrence, which can more comprehensively reflect the overall damage status of the assessment unit. close to This indicates that the assessment unit has no risk of damage or the degree of damage is extremely minor. close to This indicates that the assessed unit is almost certainly damaged and the damage covers nearly the entire unit area.

[0062] The S400 acquires the terrace type and erosion factor set of the assessment unit, and performs dynamic correction of engineering measures factors and soil loss analysis based on the terrace type, expected damage intensity and erosion factor set to obtain the newly added soil erosion.

[0063] In some embodiments, dynamic correction of engineering measures factors and soil loss analysis are performed based on terrace type, expected damage intensity, and erosion factor set to obtain the newly added soil erosion, including: Based on the terrace type, determine the normal state engineering measures factor for the assessment unit under normal conditions, and the completely damaged state engineering measures factor and degradation response parameters under completely damaged conditions; A continuous nonlinear correction function is constructed based on the engineering measure factors under normal conditions, the engineering measure factors under completely damaged conditions, and the degradation response parameters. The expected damage intensity is substituted into the continuous nonlinear correction function as the independent variable to calculate the change in the engineering measure factors. Obtain the rainfall erosivity factor, soil erodibility factor, slope length factor, slope factor, biological measure factor, and tillage measure factor from the erosivity factor set. Substitute the changes in engineering measure factor, rainfall erosivity factor, soil erodibility factor, slope length factor, slope factor, biological measure factor, and tillage measure factor into the soil loss model to calculate the newly added soil erosion modulus of the assessment unit. Obtain the area of ​​the assessment unit, multiply the newly added soil erosion modulus by the area, and obtain the newly added soil erosion amount of the assessment unit.

[0064] In some embodiments, according to the evaluation unit Terraced fields Determine the normal state engineering measure factors under normal conditions. And the engineering measures factor for the completely destroyed state. and degradation response parameters Different types of terraced fields , and The values ​​are shown in Table 1.

[0065] Table 1 Engineering measure factor parameters for different terrace types

[0066] The basis for determining each parameter is as follows: The values ​​were taken with reference to the recommended values ​​of engineering measures factors for different types of terraces under good soil and water conservation conditions in the "Classification and Grading Standard for Soil Erosion" (SL190-2007); The values ​​were taken from damaged terraced fields in the same area (with a similar proportion of damaged area). Fitted values ​​of field monitoring data; degradation response parameters Based on historical damage observation data of different terrace types in the study area, the parameters were calibrated using a grid search method with the goal of minimizing the root mean square error between the soil erosion modulus calculated by substituting the corrected engineering measures factors into the Chinese soil loss model and the measured erosion modulus.

[0067] The complete expression for the continuous nonlinear correction function is:

[0068] Changes in engineering measures factors The calculation formula is:

[0069] In the formula, The revised engineering measures factor; This represents the change in the engineering measure factor, i.e., the difference between the corrected engineering measure factor and the normal state engineering measure factor.

[0070] Obtain the rainfall erosivity factor from the erosivity factor set. Soil erodibility factors Slope length factor Slope factor Biological factors and farming practices The change in engineering measures factors Substituting the above six factors into the Chinese Soil Loss Equation (CSLE) model, the assessment unit is calculated. The newly added soil erosion modulus The calculation formula is:

[0071] In the formula, To add soil erosion modulus (unit: The methods for obtaining each factor are as follows: Based on multi-year average monthly rainfall data of the study area, Calculation using empirical formulas; Based on soil texture and organic matter content, Model formula calculation; and Slope length and slope calculation based on digital elevation model data; Determined by referring to tables based on vegetation cover and land use type; The method and season of cultivation are determined by referring to a table.

[0072] Evaluation Unit New soil erosion ,in As an evaluation unit Area (unit: ), new soil erosion The unit is .

[0073] The S500 acquires the sediment transport ratio and soil dataset of the assessment unit, performs functional loss analysis based on the new soil erosion, sediment transport ratio, and soil dataset, and obtains the functional loss assessment results.

[0074] In some embodiments, functional loss analysis is performed based on the increase in soil erosion, sediment transport ratio, and soil dataset to obtain functional loss assessment results, including: The effective mass of sediment lost into river channels and valleys is calculated by multiplying the newly added soil erosion amount by the sediment transport ratio. Obtain the topsoil bulk density, topsoil price, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price from the soil dataset; The composite ecological function loss of the assessment unit is calculated based on the effective sediment mass, topsoil bulk density, topsoil price, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price. The sum of the combined ecological function losses of all assessment units is taken as the total functional loss, and the functional loss assessment result is determined based on the total functional loss.

[0075] In some embodiments, sediment transport ratio Defined as an evaluation unit The proportion of sediment lost into river channels and valleys to the total soil erosion in that unit is dimensionless and ranges from [value missing]. . According to the evaluation unit The watershed area, gully density, and slope were calculated using empirical formulas for the assessment units within the study area. The range of values ​​is The effective sediment quality lost into river channels and valleys The calculation formula is:

[0076]

[0077] In the formula, Units are .

[0078] Obtain the topsoil bulk density from the soil dataset (unit: ), topsoil price (unit: Soil total nitrogen mass fraction (Dimensionless, expressed as mass fraction), total phosphorus mass fraction (dimensionless), organic matter mass fraction (Dimensionless) Price of Total Nitrogen Fertilizer (unit: ), price of all-phosphate fertilizer (unit: ) and organic fertilizer prices (unit: ).

[0079] In some embodiments, the calculation process for the loss of complex ecological functions includes: The loss of topsoil resource substitution function is calculated based on the effective sediment mass, topsoil bulk density, and topsoil price. The loss of soil nutrient retention ecological function is calculated based on the effective sediment mass, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price. The loss of topsoil resource substitution function and the loss of soil nutrient retention ecological function are added together to obtain the composite ecological function loss.

[0080] Evaluation Unit Loss of surface soil resources as a substitute The calculation formula is:

[0081] In the formula, The unit is This loss quantifies the value of topsoil resources corresponding to the effective sediment lost into the river channel, reflecting the direct economic loss of the soil parent material resources themselves.

[0082] Evaluation Unit Loss of soil nutrient retention and ecological functions The calculation formula is:

[0083] In the formula, The unit is This loss quantifies the loss of value of nitrogen, phosphorus, and organic matter nutrients carried by the lost sediment, reflecting the indirect economic loss of soil fertility capital due to erosion.

[0084] Evaluation Unit loss of complex ecological functions The calculation formula is:

[0085] In the formula, The unit is This composite loss includes both the loss of soil parent material resources and the loss of soil nutrients, and is a comprehensive measure of the loss of soil and water conservation function caused by the damage to terraces.

[0086] The total functional loss is obtained by summing the combined ecological function losses of all assessment units. The calculation formula is:

[0087] In the formula, The unit is .

[0088] The functional loss assessment results are determined based on the total functional loss, including: The formula for calculating the functional loss intensity per unit area of ​​the study area is as follows:

[0089] In the formula, The functional loss intensity per unit area, This represents the total area of ​​terraced fields in the study area. This indicator is used to eliminate the impact of area differences on the assessment of total loss, facilitating horizontal comparisons between areas of different sizes and longitudinal tracking between different years.

[0090] The combined ecological function loss of three types of terraced fields—stone-embankment terraced fields, earthen-embankment terraced fields, and horizontal terraced fields—was calculated separately and denoted as follows: , and Calculate the proportion of each type of loss to the total loss. The calculation formula is:

[0091]

[0092] In the formula, Terraced fields The contribution of the loss; through Identifying the types of terraced fields that contribute the most to the losses provides a basis for targeted deployment of governance measures.

[0093] The combined ecological function loss of each assessment unit Sort by size from largest to smallest, calculate the cumulative loss curve, and sum the losses to the total functional loss. of The evaluation units corresponding to the cumulative range are identified as functional loss hotspot regions. The formula for identifying hotspot regions is as follows:

[0094] In the formula, Sort by size from largest to smallest The combined ecological function loss of each assessment unit To accumulate to The number of assessment units required for total loss. This is a set of assessment units for hotspot areas. Assessment units within hotspot areas are severely damaged and have concentrated ecological losses, and should be the preferred implementation areas for near-term soil and water conservation projects.

[0095] Desired damage intensity Increased soil erosion and loss of complex ecological functions Perform range standardization separately to obtain standardized values. , and The governance priority index is obtained by weighted summation of the three standardized values. The calculation formula is:

[0096] In the formula, As the weight of the expected damage intensity, To add weight to soil erosion, As a weight for the loss of complex ecological functions, . The range of values ​​is , The higher the assessment value, the greater the severity of the damage, the greater the harm caused by soil erosion, and the stronger the urgency of remediation. Remediation measures should be prioritized in these cases.

[0097] Total functional loss Unit area loss intensity Contribution of losses to different types of terraced fields In addition, the results of spatial hotspot distribution are summarized, combined with the governance priority index of each assessment unit. This results in a comprehensive assessment of functional loss. The assessment includes the following: a total loss statistics table, a distribution map of loss intensity per unit area, a spatial distribution map of functional loss hotspots, a list of priority areas for each assessment unit, and tiered management recommendations. This assessment directly guides the timing, funding allocation, and selection of optimal management measures for the terraced ecological restoration project in the study area.

[0098] Reference Figure 2 This invention provides a terraced field damage and functional loss assessment device 20, used to implement a method for assessing terraced field damage and functional loss. The device includes: The evaluation unit acquisition module 21 is used to acquire the spatial distribution data of terraced fields in the target area, and to divide the target area into terraced patches based on the spatial distribution data of terraced fields to obtain evaluation units. The standardized feature vector acquisition module 22 is used to acquire the initial multidimensional driving factor data of the evaluation unit, and to perform physical feature expansion and factor screening based on the initial multidimensional driving factor data to obtain the standardized feature vector. The expected damage intensity acquisition module 23 is used to input the standardized feature vector into the multi-task neural network for damage prediction, obtain the predicted damage probability and the predicted damage area ratio, and multiply the predicted damage probability and the predicted damage area ratio to obtain the expected damage intensity. A new soil erosion acquisition module 24 is added to acquire the terrace type and erosion factor set of the assessment unit. Based on the terrace type, expected damage intensity and erosion factor set, dynamic correction of engineering measures factors and soil loss analysis are performed to obtain the new soil erosion amount. The functional loss assessment result acquisition module 25 is used to acquire the sediment transport ratio and soil data set of the assessment unit, and to perform functional loss analysis based on the newly added soil erosion, sediment transport ratio and soil data set to obtain functional loss assessment results.

[0099] This application provides an electronic device, including a processor and a memory; the memory stores a computer program, wherein the computer program, when executed by the processor, implements the terraced field damage and functional loss assessment method of any of the above schemes.

[0100] Specifically, the processor may include, for example, a general-purpose microprocessor, an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor may also include onboard memory for caching purposes. The processor may be a single processing unit or multiple processing units for performing different actions of the method flow according to embodiments of this application.

[0101] Memory can be any medium capable of containing, storing, transmitting, propagating, or transmitting instructions. For example, memory can include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, instruments, or propagation media. Specific examples of memory include: magnetic storage devices such as magnetic tape or hard disk drives (HDDs); optical storage devices such as optical discs (CD-ROMs); and also random access memory (RAM) or flash memory; and / or wired / wireless communication links.

[0102] This application also provides a computer-readable medium storing a computer program that, when executed by a processor, implements the terraced field damage and functional loss assessment method described above. This computer-readable medium may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into that device / apparatus / system. The aforementioned computer-readable medium carries one or more programs, which, when executed, implement the method as described in the embodiments of this application.

[0103] According to embodiments of this application, a computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wired, optical fiber, radio frequency signals, etc., or any suitable combination thereof.

[0104] Those skilled in the art will understand that the features described in the various embodiments and / or claims of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments and / or claims of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application. Therefore, the scope of this application should not be limited to the above embodiments, but should be defined not only by the appended claims, but also by their equivalents. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for assessing the damage and functional loss of terraced fields, characterized in that, include: Acquire the spatial distribution data of terraced fields in the target area, and divide the target area into terraced patches based on the spatial distribution data to obtain evaluation units; Obtain the initial multidimensional driving factor data of the evaluation unit, and perform physical feature expansion and factor screening based on the initial multidimensional driving factor data to obtain a standardized feature vector; The standardized feature vector is input into a multi-task neural network for damage prediction to obtain the predicted damage probability and the predicted damage area ratio. The predicted damage probability and the predicted damage area ratio are multiplied to obtain the expected damage intensity. Obtain the terrace type and erosion factor set of the assessment unit, and perform dynamic correction of engineering measures factors and soil loss analysis based on terrace type, expected damage intensity and erosion factor set to obtain the newly added soil erosion amount; Obtain the sediment transport ratio and soil data set of the assessment unit, and conduct functional loss analysis based on the newly added soil erosion, sediment transport ratio and soil data set to obtain the functional loss assessment results.

2. The method for assessing terraced field damage and functional loss according to claim 1, characterized in that, Based on the spatial distribution data of terraced fields, the target area is divided into terraced patches to obtain evaluation units, including: Acquire the digital elevation model data of terraced fields, slope classification standards, spectral characteristics of field boundaries, topographic texture characteristics, spatial continuity characteristics, and field ridge dividing lines from the spatial distribution data of terraced fields; Based on the slope grading standard, the spectral characteristics of field boundaries and the topographic texture characteristics, the terrace patch boundaries of the digital elevation model data of terraced fields are identified to obtain the vector data of terrace patch boundaries. Based on the spatial continuity characteristics and the field ridge dividing lines, the boundary vector data of the terraced fields are divided into independent fields to obtain multiple initial terraced field patch units; The boundaries of the initial terraced patch units are smoothed and topological consistency is screened to obtain evaluation units.

3. The method for assessing terraced field damage and functional loss according to claim 1, characterized in that, Obtain initial multidimensional driving factor data for the evaluation unit, and perform physical feature expansion and factor selection based on the initial multidimensional driving factor data to obtain a standardized feature vector, including: For each assessment unit, initial multidimensional driving factor data including natural environment, ecological characteristics and human activity disturbance are obtained, and initial physical characteristic extended variables including slope micro-topography and hydrological characteristics are calculated based on the initial multidimensional driving factor data. By using a composite screening index weighted by point-binary correlation coefficient and mutual information, factor screening is performed on the initial multidimensional driving factor data and the initial physical feature extended variables to obtain optimized multidimensional driving factor data and optimized physical feature extended variables. By optimizing multidimensional driving factor data and physical feature extension variables, a standardized feature vector for the evaluation unit is formed.

4. The method for assessing terraced field damage and functional loss according to claim 1, characterized in that, The multi-task neural network includes a shared feature extraction layer, a first fully connected layer, a second fully connected layer, a damage probability prediction layer, and a damage extent prediction layer. The process of obtaining the predicted damage probability and the predicted damage area ratio includes: The standardized feature vector is input into the shared feature extraction layer, and the standardized feature vector is recalibrated through a self-attention mechanism to obtain attention-enhanced features; The attention-enhanced features are sequentially input into the first fully connected layer and the second fully connected layer, and the output of the second fully connected layer is added to the input of the first fully connected layer through a residual connection to obtain the shared hidden features; The shared hidden features are input into the damage probability prediction layer to predict the damage probability. The shared hidden features are input into the damage prediction layer to predict the damaged area and obtain the predicted proportion of the damaged area.

5. The method for assessing terraced field damage and functional loss according to claim 1, characterized in that, Based on terrace type, expected damage intensity, and erosion factor set, dynamic correction of engineering measures and soil loss analysis are performed to obtain the newly added soil erosion, including: Based on the terrace type, determine the normal state engineering measures factor for the assessment unit under normal conditions, and the completely damaged state engineering measures factor and degradation response parameters under completely damaged conditions; A continuous nonlinear correction function is constructed based on the engineering measure factors under normal conditions, the engineering measure factors under completely damaged conditions, and the degradation response parameters. The expected damage intensity is substituted into the continuous nonlinear correction function as the independent variable to calculate the change in the engineering measure factors. Obtain the rainfall erosivity factor, soil erodibility factor, slope length factor, slope factor, biological measure factor, and tillage measure factor from the erosivity factor set. Substitute the changes in engineering measure factor, rainfall erosivity factor, soil erodibility factor, slope length factor, slope factor, biological measure factor, and tillage measure factor into the soil loss model to calculate the newly added soil erosion modulus of the assessment unit. Obtain the area of ​​the assessment unit, multiply the newly added soil erosion modulus by the area, and obtain the newly added soil erosion amount of the assessment unit.

6. The method for assessing terraced field damage and functional loss according to claim 1, characterized in that, Functional loss analysis was conducted based on the increase in soil erosion, sediment transport ratio, and soil dataset to obtain functional loss assessment results, including: The effective mass of sediment lost into river channels and valleys is calculated by multiplying the newly added soil erosion amount by the sediment transport ratio. Obtain the topsoil bulk density, topsoil price, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price from the soil dataset; The composite ecological function loss of the assessment unit is calculated based on the effective sediment mass, topsoil bulk density, topsoil price, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price. The sum of the combined ecological function losses of all assessment units is taken as the total functional loss, and the functional loss assessment result is determined based on the total functional loss.

7. The method for assessing terraced field damage and functional loss according to claim 6, characterized in that, The calculation process for the combined loss of ecological functions includes: The loss of topsoil resource substitution function is calculated based on the effective sediment mass, topsoil bulk density, and topsoil price. The loss of soil nutrient retention ecological function is calculated based on the effective sediment mass, total nitrogen mass fraction, total phosphorus mass fraction, organic matter mass fraction, total nitrogen fertilizer price, total phosphorus fertilizer price, and organic matter fertilizer price. The loss of topsoil resource substitution function and the loss of soil nutrient retention ecological function are added together to obtain the composite ecological function loss.

8. A device for assessing terraced field damage and functional loss, used to implement the method for assessing terraced field damage and functional loss as described in any one of claims 1 to 7, characterized in that, The device includes: The evaluation unit acquisition module is used to acquire the spatial distribution data of terraced fields in the target area, and to divide the target area into terraced patches based on the spatial distribution data of terraced fields to obtain evaluation units. The standardized feature vector acquisition module is used to acquire the initial multidimensional driving factor data of the evaluation unit, and to perform physical feature expansion and factor screening based on the initial multidimensional driving factor data to obtain the standardized feature vector. The expected damage intensity acquisition module is used to input the standardized feature vector into the multi-task neural network for damage prediction, obtain the predicted damage probability and the predicted damage area ratio, and multiply the predicted damage probability and the predicted damage area ratio to obtain the expected damage intensity. A new soil erosion acquisition module has been added to obtain the terrace type and erosion factor set of the assessment unit. Based on the terrace type, expected damage intensity and erosion factor set, dynamic correction of engineering measures factors and soil loss analysis are performed to obtain the new soil erosion amount. The module for obtaining functional loss assessment results is used to acquire the sediment transport ratio and soil data set of the assessment unit. Based on the newly added soil erosion, sediment transport ratio and soil data set, functional loss analysis is performed to obtain functional loss assessment results.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for assessing terraced field damage and functional loss as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for assessing terraced field damage and functional loss as described in any one of claims 1 to 7.

Images