A method for evaluating wetland ecological drought, an electronic device and a storage medium

Abstract

A kind of wetland ecological drought condition evaluation method, electronic equipment and storage medium belong to wetland ecological evaluation technical field.To solve the problem of accurate evaluation wetland ecological drought condition.This application collects image data and meteorological data of wetland, constructs image data set of wetland, meteorological data set of wetland;Wetland ecological factor is constructed, including humidity, rainfall and groundwater level of wetland;Deep learning network model is constructed, image data of wetland in image data set of wetland is input into deep learning network model to extract wetland water body separation result graph;The change trend of wetland ecological factor is solved, and the change trend data set of wetland ecological factor is obtained;Wetland drought condition index model is constructed;The obtained wetland drought condition index model is trained using random forest regression algorithm, and the optimized wetland drought condition index model is obtained to evaluate the wetland ecological drought condition.This application provides scientific basis for wetland protection and management.

Description

Technical Field

[0001] This invention belongs to the field of wetland ecological assessment technology, specifically relating to a method for assessing wetland ecological drought conditions, electronic equipment, and storage medium. Background Technology

[0002] Wetlands are one of the most important ecosystems on Earth, possessing significant ecological and economic value. However, due to factors such as climate change and human activities, wetland ecosystems face the threat of drought. Therefore, developing a method to accurately assess the drought situation in wetland ecosystems is of great importance for wetland conservation and ecological environment management.

[0003] Establishing a wetland ecological drought index to reveal the spatiotemporal characteristics and development patterns of ecological drought is a key technical means for assessing wetland ecological drought conditions. How to scientifically identify and assess wetland ecological drought and its severity is a critical technical issue that urgently needs to be addressed in wetland hydrological regulation and water resource management in response to climate change. However, wetland ecological drought assessment studies based on single factors neglect the unique characteristics of wetland ecosystems and cannot meet the urgent need for accurate assessment of current wetland ecological drought conditions. Currently, there is an urgent need to establish theories and methods for analyzing wetland drought characteristics to analyze the spatiotemporal variations of wetland ecological drought conditions. Summary of the Invention

[0004] The problem to be solved by this invention is to accurately assess the ecological drought situation of wetlands, and to propose a method, electronic device and storage medium for assessing the ecological drought situation of wetlands.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for assessing the ecological drought status of wetlands includes the following steps:

[0007] S1. Collect image data and meteorological data of wetlands, and construct image datasets and meteorological datasets of wetlands;

[0008] S2. Construct wetland ecological factors, including wetland humidity, rainfall, and groundwater level;

[0009] S3. Construct a deep learning network model and input the wetland image data from the wetland image dataset obtained in step S1 into the deep learning network model to extract the water body distribution map of the wetland.

[0010] S4. Based on the water body separation result map of the wetland obtained in step S3, obtain the water body distribution map of the wetland. Combine the meteorological data of the wetland in the meteorological dataset of the wetland obtained in step S1 to solve the changing trend of wetland ecological factors and obtain the changing trend dataset of wetland ecological factors.

[0011] S5. Construct a wetland drought index model;

[0012] S6. Based on the wetland ecological factor change trend dataset obtained in step S4, the wetland drought index model obtained in step S5 is trained using the random forest regression algorithm to obtain an optimized wetland drought index model for assessing the wetland ecological drought situation.

[0013] Furthermore, the wetland image data mentioned in step S1 includes one of satellite remote sensing data, UAV remote sensing data, and ground remote sensing data.

[0014] Furthermore, the specific implementation method of step S3 includes the following steps:

[0015] S3.1 Constructing the spectral branches of the deep learning network model, wherein the spectral branches include three... Convolutional layers To focus on spectral dimensional information, a batch normalization layer and a rectified linear unit (BN+ReLU) are added after each convolutional layer to improve the network's nonlinear representation capability. Then, a max-pooling layer (MaxPool) is added before the second and third convolutional layers to reduce the network's parameters and computational complexity. The wetland image data obtained in step S1 is input into the spectral branch to extract spectral features from the wetland image data. Its calculation expression is:

[0016]

[0017]

[0018] in, Represents a 1×1 convolutional layer. Represents the spectral branch; The first branch in the spectral representation Output feature maps of convolutional layers; The first branch in the spectral representation The weights of the convolutional layers; The first branch in the spectral representation Bias of convolutional layers; Representative activation function ; Represents the max pooling layer;

[0019] S3.2 Construct the spatial branches of the deep learning network model, wherein the spatial branches include three Convolutional layers This is used to focus on semantic location information; a batch normalization layer and an activation function are added after each convolutional layer. A rectified linear unit (BN+ReLU) is used to improve the network's nonlinear representation capability. Then, a max pooling layer (MaxPool) is added before the second and third convolutional layers to reduce the network's parameters and computational complexity. The wetland image data obtained in step S1 is input into the spatial branch to extract spatial features from the wetland image data. Its calculation expression is:

[0020]

[0021]

[0022] in, Represents a 3×3 convolutional layer. Represents spatial branches; The first in the spatial branch Output feature maps of convolutional layers; The first in the spatial branch The weights of the convolutional layers; The first in the spatial branch Bias of convolutional layers; Representative activation function ; Represents the max pooling layer;

[0023] S3.3 Construct a global spectral spatial attention module to aggregate the spectral features obtained in step S3.1. The space obtained in step S3.2 is based on features Get integrated features The calculation expression is:

[0024]

[0025] in, It integrates spectral information and spatial semantic information, among which , , These represent the height, width, and number of bands of the integrated feature map, respectively.

[0026] Then, two convolutional layers are used to... Divided into those with spectral sensitivity characteristics and features of spatial sensitivity The calculation expression is:

[0027]

[0028]

[0029] in, and The kernel size of all convolutions is ;

[0030] Then the global characteristics of spectral sensitivity Global features with spatial sensitivity The spectral attention map and spatial attention map are obtained respectively using the spectral and spatial attention modules, and the calculation expression is as follows:

[0031]

[0032]

[0033] in, and These represent spectral attention maps and spatial attention maps, respectively. represent Spectral attention mechanism represent Spatial attention mechanisms; This represents element-wise multiplication;

[0034] S3.4. Combine the spectral attention map and spatial attention map obtained in step S3.3 with the spatial features, respectively. and spectral characteristics Adding them together further highlights important classification information and suppresses irrelevant information. The calculation expression is:

[0035]

[0036] in, and These represent the spectral feature map after spectral attention enhancement and the spatial feature map after spatial attention enhancement, respectively;

[0037] The final spectral feature map and spatial feature map are concatenated along the channel dimension to obtain a joint spatial-spectral feature map. Then, the Flatten operation is used to reshape the integrated joint spatial-spectral feature map into a one-dimensional vector, which is then input into a fully connected layer for feature fusion. Finally, it is input into a Softmax layer for classification, calculating the probability that each pixel belongs to the water body. The calculation expression is as follows:

[0038]

[0039] in, , These represent the i-th and j-th elements in the input feature vector, respectively.

[0040] Using cross-entropy loss function The parameters of the deep learning network model are updated using the following calculation expression:

[0041]

[0042] in, This represents the total number of wetland types;

[0043] S3.5. Using the classification probability of each pixel belonging to water obtained in step S3.4, classify each pixel to obtain the final wetland water body separation probability map.

[0044] Furthermore, in step S3.4 Spectral attention mechanism The specific implementation method is as follows: use global average pooling respectively and global maximum pooling The global features of the input spectral sensitivity are pooled, and then the resulting features are fed into a multilayer perceptron with hidden layers. The size of the hidden layer vector is set to... Where r represents the attenuation coefficient, the element-wise summation of the multilayer perceptron output is used to obtain the spectral attention weights, calculated as follows:

[0045]

[0046]

[0047]

[0048]

[0049] in, Global features representing the spectral sensitivity of global average pooling. Global characteristics representing the spectral sensitivity of global max pooling. represent Activation function; Represents a multilayer perceptron. This represents the spectral attention weight.

[0050] Furthermore, in step S3.4 Spatial attention mechanism The specific implementation method is as follows: Global average pooling and global maximum pooling are used to pool the spatially sensitive global features of the input. Then, the feature maps of the two channels are processed by standard convolution to obtain the spatial attention map. The calculation expression is:

[0051]

[0052]

[0053]

[0054] in, The size of the convolution kernel is represented by , This represents spatial attention weights.

[0055] Furthermore, the specific implementation method of step S4 includes the following steps:

[0056] S4.1 The water area of ​​the wetland is calculated by obtaining the number of pixels classified as water bodies through statistical analysis of the probability of each pixel belonging to a water body. The calculation expression is as follows:

[0057]

[0058] in, Represents the water area of ​​wetlands. This represents the total number of water body pixels in the water body separation result image of the wetland. This represents the resolution of the input remote sensing data;

[0059] S4.2. Based on the determination of wetland water body classification, the water level is calculated using satellite altimetry data within the water body area. The expression for calculating the water level is:

[0060]

[0061] in, The EGM96 reference water level represents the actual surface water level. This represents the distance between the altimeter and the ground. Represents the distance between the ground level and the Earth's ellipsoid. Represents the distance between the satellite and the water surface. Represents the correction parameter;

[0062] The calculation expression is:

[0063]

[0064] in, , These represent dry and wet tropospheric corrections, respectively. Represents ionospheric correction. , These represent solid tides and polar tide corrections, respectively.

[0065] By combining the wetland water separation results obtained in S3.4 with the water area obtained in S4.1 and the water level information obtained in S4.2, the final wetland water distribution map is obtained. Water distribution map It contains information about wetland water bodies, including the area and water level of those water bodies.

[0066] S4.3. Based on the wetland water distribution map obtained in step S4.2, collect the wetland water distribution map from the wetland meteorological dataset obtained in step S1. Long-term data on humidity, rainfall, and groundwater level in the region were used to analyze the changing trends of wetland ecological factors, which are represented by y.

[0067] S4.3.1 Arrange the wetland ecological factor data in chronological order to construct data pairs of independent variable time and dependent variable wetland ecological factor, where day represents time; assuming a linear relationship between wetland ecological factors and time day, the calculation expression is as follows:

[0068]

[0069] in, and They represent coefficients, Represents the error term;

[0070] S4.3.2 Calculate the regression coefficients using the least squares method and The value of , the average of the calculation time. and the mean of wetland ecological factors The calculation expression is:

[0071]

[0072]

[0073] coefficient The calculation expression is:

[0074]

[0075] coefficient The calculation expression is:

[0076]

[0077] Where n represents the number of data pairs;

[0078] S4.3.3 Calculate the predicted value of the regression equation The calculation expression is:

[0079] ;

[0080] S4.3.4 Calculate the coefficient of determination To evaluate the goodness of fit of the model, The value of is between 0 and 1, with a value closer to 1 indicating a better model fit. The calculation expression is:

[0081] ;

[0082] By combining the meteorological data of the wetland obtained in step S1, the changing trends of wetland ecological factors are solved, and the wetland ecological drought situation is identified.

[0083] Furthermore, the specific implementation method of step S5 includes the following steps:

[0084] S5.1. Standardize the wetland ecological factors in the wetland ecological factor change trend dataset obtained in step S4. The calculation expression is as follows:

[0085]

[0086]

[0087]

[0088] in, Indicates the normalized humidity value. Indicates the wetland research area in Humidity at any given moment; and This represents the maximum and minimum humidity values ​​in the historical data of the wetland study area; and This represents the standardized values ​​of rainfall and groundwater level in the wetland study area; and This represents the rainfall and groundwater level in the wetland study area at time l. , , , These represent the maximum and minimum values ​​of rainfall and groundwater level in the historical data of the wetland study area, respectively.

[0089] S5.2. Based on the obtained standardized values ​​of humidity, rainfall, and groundwater level, a wetland drought index model is constructed, and the calculation expression is as follows:

[0090]

[0091] in, , , These represent the weights of humidity, rainfall, and groundwater level, respectively, and are used to adjust the influence of each ecological factor on the drought index.

[0092] Furthermore, the specific implementation method of step S6 includes the following steps:

[0093] S6.1 Based on the wetland ecological factor change trend dataset obtained in step S4, the wetland drought index model obtained in step S5 is trained using the random forest regression algorithm to find the optimal regression coefficients. , , and The calculation expression is:

[0094]

[0095] in, This represents the optimization correction coefficient of the optimized wetland drought index model;

[0096] S6.2. Select one of the traditional drought indices, SPI, SPEI, or SRI, and compare it with the wetland drought index constructed in step S6.1 using Pearson correlation analysis on historical wetland ecological data. The calculation expression is as follows:

[0097]

[0098] in, This represents the Pearson correlation coefficient. This represents the constructed wetland ecological drought index. This represents the traditional drought index; The mean wetland ecological drought index was constructed for the selected study area. The average traditional drought index for the selected study area;

[0099] The Pearson correlation coefficient ranges from 1 to -1. The larger the absolute value of the correlation coefficient, the stronger the correlation between the data. If the correlation coefficient is 0, it means that there is no correlation or a non-linear correlation.

[0100] S6.3. Based on historical data, the comprehensive ecological drought index of wetlands in the selected study area over the years was calculated. The least squares method was used to analyze the changing trend of the comprehensive ecological drought index of wetlands in the selected study area in recent years. The calculation expression of the least squares method is as follows:

[0101]

[0102] in, Represents the actual value; This represents the predicted value obtained through the assumed function;

[0103] S6.4. Based on the trend of the comprehensive drought index obtained in step S6.3, the trend of ecological status change in the study area over a long period is determined. Then, the collected relevant ecological factors are analyzed and predicted using the least squares method. 0.4 is set as severe drought; 0.4 <HRE 0.6 is set as moderate drought; 0.6 <HRE 0.8 was set as mild drought; HRE>0.8 was set as normal without drought, verifying the consistency and rationality with the changing trends of wetland ecological factors.

[0104] An electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the wetland ecological drought assessment method.

[0105] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned method for assessing drought conditions in wetland ecosystems.

[0106] The beneficial effects of this invention are:

[0107] The present invention discloses a method for assessing wetland ecological drought, which uses image processing algorithms to process satellite image data and analyzes it in conjunction with meteorological data to extract wetland ecological factors and analyze their variation patterns, thereby providing support for the subsequent establishment and assessment of wetland drought index models.

[0108] The present invention discloses a method for assessing wetland ecological drought. The establishment of the wetland ecological drought index model includes weighted combination of wetland ecological factors and setting corresponding weight coefficients to comprehensively consider the influence of each factor on wetland drought, thereby obtaining a more accurate wetland ecological drought index.

[0109] This invention discloses a method for assessing wetland ecological drought, which uses statistical analysis to evaluate the changing trends of wetland ecological factors. By employing statistical analysis, the changing trends of wetland ecological factors can be better understood, and the degree of wetland drought can be assessed accordingly, thereby improving the accuracy and reliability of the wetland ecological drought index.

[0110] This invention discloses a method for assessing wetland ecological drought. By comparing the wetland ecological drought index with historical wetland ecological data, the degree of drought is determined. This method provides a better assessment of the current drought level and, by comparing it with past conditions, helps evaluate the degree and trend of wetland drought. The establishment of the wetland ecological drought index also considers and corrects for different wetland types and geographical locations. Since different wetland types and geographical locations have different characteristics and varying sensitivities to drought, these factors need to be considered and corrected when establishing the wetland ecological drought index to improve its accuracy and applicability. Machine learning algorithms are used to train the wetland ecological data and optimize the index calculation model. By utilizing machine learning algorithms, large amounts of wetland ecological data can be analyzed and learned, optimizing the index calculation model to make it more accurate and efficient, thereby improving the ability to assess wetland drought.

[0111] This invention discloses a method for assessing wetland ecological drought, which classifies wetlands into different drought levels. By setting a threshold for the wetland ecological drought index, wetlands can be divided into different drought levels, thereby achieving an accurate assessment of the degree of wetland drought and providing a scientific basis for wetland protection and management. Attached Figure Description

[0112] Figure 1 This is a flowchart of a wetland ecological drought assessment method according to the present invention;

[0113] Figure 2 This is a schematic diagram of the deep learning network model structure of the wetland ecological drought assessment method described in this invention;

[0114] Figure 3 This is a schematic diagram of the global spectral spatial attention module structure in the deep learning network model of the wetland ecological drought assessment method described in this invention. Detailed Implementation

[0115] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0116] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0117] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 -Appendix Figure 3 Detailed explanation is as follows: Specific implementation method one:

[0119] A method for assessing drought conditions in wetland ecosystems, characterized by comprising the following steps:

[0120] S1. Collect image data and meteorological data of wetlands, and construct image datasets and meteorological datasets of wetlands;

[0121] Furthermore, the wetland image data mentioned in step S1 includes one of satellite remote sensing data, UAV remote sensing data, and ground remote sensing data;

[0122] S2. Construct wetland ecological factors, including wetland humidity, rainfall, and groundwater level;

[0123] S3. Construct a deep learning network model and input the wetland image data from the wetland image dataset obtained in step S1 into the deep learning network model to extract the water body separation result map of the wetland.

[0124] Furthermore, the specific implementation method of step S3 includes the following steps:

[0125] S3.1 Constructing the spectral branches of the deep learning network model, wherein the spectral branches include three... Convolutional layers To focus on spectral dimensional information, a batch normalization layer and a rectified linear unit (BN+ReLU) are added after each convolutional layer to improve the network's nonlinear representation capability. Then, a max-pooling layer (MaxPool) is added before the second and third convolutional layers to reduce the network's parameters and computational complexity. The wetland image data obtained in step S1 is input into the spectral branch to extract spectral features from the wetland image data. Its calculation expression is:

[0126]

[0127]

[0128] in, Represents a 1×1 convolutional layer. Represents the spectral branch; The first branch in the spectral representation Output feature maps of convolutional layers; The first branch in the spectral representation The weights of the convolutional layers; The first branch in the spectral representation Bias of convolutional layers; Representative activation function ; Represents the max pooling layer;

[0129] S3.2 Construct the spatial branches of the deep learning network model, wherein the spatial branches include three Convolutional layers This is used to focus on semantic location information; a batch normalization layer and an activation function are added after each convolutional layer. A rectified linear unit (BN+ReLU) is used to improve the network's nonlinear representation capability. Then, a max pooling layer (MaxPool) is added before the second and third convolutional layers to reduce the network's parameters and computational complexity. The wetland image data obtained in step S1 is input into the spatial branch to extract spatial features from the wetland image data. Its calculation expression is:

[0130]

[0131]

[0132] in, Represents a 3×3 convolutional layer. Represents spatial branches; The first in the spatial branch Output feature maps of convolutional layers; The first in the spatial branch The weights of the convolutional layers; The first in the spatial branch Bias of convolutional layers; Representative activation function ; Represents the max pooling layer;

[0133] S3.3 Construct a global spectral spatial attention module to aggregate the spectral features obtained in step S3.1. and the spatial features obtained in step S3.2 Get integrated features The calculation expression is:

[0134]

[0135] in, It integrates spectral information and spatial semantic information, among which , , These represent the height, width, and number of bands of the integrated feature map, respectively.

[0136] Then, two convolutional layers are used to divide the features into spectrally sensitive features. Global features with spatial sensitivity The calculation expression is:

[0137]

[0138]

[0139] in, and The kernel size of all convolutions is ;

[0140] Then the spectral sensitivity characteristics Features of spatial sensitivity The spectral attention map and spatial attention map are obtained using the spectral and spatial attention modules respectively, and the calculation expression is as follows:

[0141]

[0142]

[0143] in, and These represent spectral attention maps and spatial attention maps, respectively. represent Spectral attention mechanism represent Spatial attention mechanisms; This represents element-wise multiplication;

[0144] S3.4. Combine the spectral attention map and spatial attention map obtained in step S3.3 with the spatial features, respectively. and spectral characteristics Adding them together further highlights important classification information and suppresses irrelevant information. The calculation expression is:

[0145]

[0146] in, and These represent the spectral feature map after spectral attention enhancement and the spatial feature map after spatial attention enhancement, respectively;

[0147] The final spectral feature map and spatial feature map are concatenated along the channel dimension to obtain a combined spatial-spectral feature map. Then, the Flatten operation is used to reshape the combined spatial-spectral feature map into a one-dimensional vector, which is then input into a fully connected layer for feature fusion. Finally, it is input into a Softmax layer for classification, calculating the probability that each pixel belongs to the water body. The calculation expression is as follows:

[0148]

[0149] in, , These represent the i-th and j-th elements in the input feature vector, respectively.

[0150] Using cross-entropy loss function The parameters of the deep learning network model are updated using the following calculation expression:

[0151]

[0152] in, This represents the total number of wetland types.

[0153] Furthermore, in step S3.4 The specific implementation method of the spectral attention mechanism is as follows: global average pooling is used respectively. and global maximum pooling The global features of the input spectral sensitivity are pooled, and then the resulting features are fed into a multilayer perceptron with hidden layers. The size of the hidden layer vector is set to... Where r represents the attenuation coefficient, the element-wise summation of the multilayer perceptron output is used to obtain the spectral attention weights, calculated as follows:

[0154]

[0155]

[0156]

[0157]

[0158] in, Global features representing the spectral sensitivity of global average pooling. Global characteristics representing the spectral sensitivity of global max pooling. represent Activation function; Represents a multilayer perceptron. Represents spectral attention weights;

[0159] Furthermore, in step S3.4 The specific implementation method of the spatial attention mechanism is as follows: Global average pooling and global maximum pooling are used to pool the spatially sensitive global features of the input. Then, the feature maps of the two channels are passed through standard convolution to obtain the spatial attention map. The calculation expression is:

[0160]

[0161]

[0162]

[0163] in, The size of the convolution kernel is represented by , Represents spatial attention weights;

[0164] S3.5. Using the classification probability of each pixel belonging to water obtained in step S3.4, classify each pixel to obtain the final wetland water body separation probability map.

[0165] S4. Based on the water body separation result map of the wetland obtained in step S3, obtain the water body distribution map of the wetland. Combine the meteorological data of the wetland in the meteorological dataset of the wetland obtained in step S1 to solve the changing trend of wetland ecological factors and obtain the changing trend dataset of wetland ecological factors.

[0166] Furthermore, the specific implementation method of step S4 includes the following steps:

[0167] S4.1 The water area of ​​the wetland is calculated by obtaining the number of pixels classified as water bodies through statistical analysis of the probability of each pixel belonging to a water body. The calculation expression is as follows:

[0168]

[0169] in, Represents the water area of ​​wetlands. This represents the total number of water body pixels in the water body separation result image of the wetland. This represents the resolution of the input remote sensing data;

[0170] S4.2. Based on the determination of the wetland water body distribution, the water level is calculated using satellite altimetry data in the water body area to obtain the wetland water body distribution map. The calculation expression is:

[0171]

[0172] in, The EGM96 reference water level represents the actual surface water level. This represents the distance between the altimeter and the ground. Represents the distance between the ground level and the Earth's ellipsoid. Represents the distance between the satellite and the water surface. Represents the correction parameter;

[0173] The calculation expression is:

[0174]

[0175] in, , These represent dry and wet tropospheric corrections, respectively. Represents ionospheric correction. , These represent solid tides and polar tide corrections, respectively.

[0176] By combining the wetland water separation results obtained in S3.4 with the water area obtained in S4.1 and the water level information obtained in S4.2, the final wetland water distribution map is obtained. Water distribution map It contains information about wetland water bodies, including the area and water level of those water bodies.

[0177] S4.3. Based on the wetland water distribution map obtained in step S4.2, collect the wetland water distribution map from the wetland meteorological dataset obtained in step S1. Long-term data on humidity, rainfall, and groundwater level in the region were used to analyze the changing trends of wetland ecological factors, which are represented by y.

[0178] S4.3.1 Arrange the wetland ecological factor data in chronological order to construct data pairs of independent variable time and dependent variable wetland ecological factor, where day represents time; assuming a linear relationship between wetland ecological factors and time day, the calculation expression is as follows:

[0179]

[0180] in, and They represent coefficients, Represents the error term;

[0181] S4.3.2 Calculate the regression coefficients using the least squares method and The value of , the average of the calculation time. and the mean of wetland ecological factors The calculation expression is:

[0182]

[0183]

[0184] coefficient The calculation expression is:

[0185]

[0186] coefficient The calculation expression is:

[0187]

[0188] Where n represents the number of data pairs;

[0189] S4.3.3 Calculate the predicted value of the regression equation The calculation expression is:

[0190]

[0191] S4.3.4 Calculate the coefficient of determination To evaluate the goodness of fit of the model, The value of is between 0 and 1, with a value closer to 1 indicating a better model fit. The calculation expression is:

[0192] ;

[0193] By combining the meteorological data of the wetland obtained in step S1, the changing trends of wetland ecological factors are solved, and the wetland ecological drought situation is identified.

[0194] S5. Construct a wetland drought index model;

[0195] Furthermore, the specific implementation method of step S5 includes the following steps:

[0196] S5.1. Standardize the wetland ecological factors in the wetland ecological factor change trend dataset obtained in step S4. The calculation expression is as follows:

[0197]

[0198]

[0199]

[0200] in, Indicates the normalized humidity value. Indicates the wetland research area in Humidity at any given moment; and This represents the maximum and minimum humidity values ​​in the historical data of the wetland study area; and This represents the standardized values ​​of rainfall and groundwater level in the wetland study area; and This represents the rainfall and groundwater level in the wetland study area at time l. , , , These represent the maximum and minimum values ​​of rainfall and groundwater level in the historical data of the wetland study area, respectively.

[0201] S5.2. Based on the obtained standardized values ​​of humidity, rainfall, and groundwater level, a wetland drought index model is constructed, and the calculation expression is as follows:

[0202]

[0203] in, , , These represent the weights of humidity, rainfall, and groundwater level, respectively, and are used to adjust the influence of each ecological factor on the drought index.

[0204] S6. Based on the wetland ecological factor change trend dataset obtained in step S4, the wetland drought index model obtained in step S5 is trained using the random forest regression algorithm to obtain an optimized wetland drought index model for assessing the wetland ecological drought situation.

[0205] Furthermore, the specific implementation method of step S6 includes the following steps:

[0206] S6.1 Based on the wetland ecological factor change trend dataset obtained in step S4, the wetland drought index model obtained in step S5 is trained using the random forest regression algorithm to find the optimal regression coefficients. , , and The calculation expression is:

[0207]

[0208] in, This represents the optimization correction coefficient of the optimized wetland drought index model;

[0209] S6.2. Select one of the traditional drought indices, SPI, SPEI, or SRI, and compare it with the wetland drought index constructed in step S6.1 using Pearson correlation analysis on historical wetland ecological data. The calculation expression is as follows:

[0210]

[0211] in, This represents the Pearson correlation coefficient. This represents the constructed wetland ecological drought index. This represents the traditional drought index; The mean wetland ecological drought index was constructed for the selected study area. The average traditional drought index for the selected study area;

[0212] The Pearson correlation coefficient ranges from 1 to -1. The larger the absolute value of the correlation coefficient, the stronger the correlation between the data. If the correlation coefficient is 0, it means that there is no correlation or a non-linear correlation.

[0213] S6.3. Based on historical data, the comprehensive ecological drought index of wetlands in the selected study area over the years was calculated. The least squares method was used to analyze the changing trend of the comprehensive ecological drought index of wetlands in the selected study area in recent years. The calculation expression of the least squares method is as follows:

[0214]

[0215] in, Represents the actual value; This represents the predicted value obtained through the assumed function;

[0216] S6.4. Based on the trend of the comprehensive drought index obtained in step S6.3, the trend of ecological status change in the study area over a long period is determined. Then, the collected relevant ecological factors are analyzed and predicted using the least squares method. 0.4 is set as severe drought; 0.4 <HRE 0.6 is set as moderate drought; 0.6 <HRE 0.8 was set as mild drought; HRE>0.8 was set as normal without drought, verifying the consistency and rationality with the changing trends of wetland ecological factors. Specific Implementation Method Two:

[0218] An electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the wetland ecological drought assessment method.

[0219] The computer device of the present invention may include a processor and a memory, such as a microcontroller containing a central processing unit. Furthermore, the processor executes the computer program stored in the memory to implement the steps of the aforementioned wetland ecological drought assessment method. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device. Specific implementation method three:

[0221] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned method for assessing drought conditions in wetland ecosystems.

[0222] The computer-readable storage medium of the present invention can be any form of storage medium that can be read by the processor of a computer device, including but not limited to non-volatile memory, volatile memory, ferroelectric memory, etc. The computer-readable storage medium stores a computer program. When the processor of the computer device reads and executes the computer program stored in the memory, the steps of the aforementioned wetland ecological drought assessment method can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0223] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0224] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for assessing the ecological drought situation of wetlands, characterized in that, Includes the following steps: S1. Collect image data and meteorological data of wetlands, and construct image datasets and meteorological datasets of wetlands; S2. Construct wetland ecological factors, including wetland humidity, rainfall, and groundwater level; S3. Construct a deep learning network model and input the wetland image data from the wetland image dataset obtained in step S1 into the deep learning network model to extract the water body separation result map of the wetland. The specific implementation method of step S3 includes the following steps: S3.1 Constructing the spectral branches of the deep learning network model, wherein the spectral branches include three... Convolutional layers To focus on spectral dimensional information, a batch normalization layer and a rectified linear unit (BN+ReLU) are added after each convolutional layer to improve the network's nonlinear representation capability. Then, a max-pooling layer (MaxPool) is added before the second and third convolutional layers to reduce the network's parameters and computational complexity. The wetland image data obtained in step S1 is input into the spectral branch to extract spectral features from the wetland image data. Its calculation expression is: in, Represents a 1×1 convolutional layer. Represents the spectral branch; The first branch in the spectral representation Output feature maps of convolutional layers; The first branch in the spectral representation The weights of the convolutional layers; The first branch in the spectral representation Bias of convolutional layers; Representative activation function ; Represents the max pooling layer; S3.2 Construct the spatial branches of the deep learning network model, wherein the spatial branches include three Convolutional layers This is used to focus on semantic location information; a batch normalization layer and an activation function are added after each convolutional layer. A rectified linear unit (BN+ReLU) is used to improve the network's nonlinear representation capability. Then, a max pooling layer (MaxPool) is added before the second and third convolutional layers to reduce the network's parameters and computational complexity. The wetland image data obtained in step S1 is input into the spatial branch to extract spatial features from the wetland image data. Its calculation expression is: in, Represents a 3×3 convolutional layer. Represents spatial branches; The first in the spatial branch Output feature maps of convolutional layers; The first in the spatial branch The weights of the convolutional layers; The first in the spatial branch Bias of convolutional layers; Representative activation function ; Represents the max pooling layer; S3.3 Construct a global spectral spatial attention module to aggregate the spectral features obtained in step S3.

1. and the spatial features obtained in step S3.2 Get integrated features The calculation expression is: in, It integrates spectral information and spatial semantic information, among which , , These represent the height, width, and number of bands of the integrated feature map, respectively. Then, two convolutional layers are used to... Divided into those with spectral sensitivity characteristics and features of spatial sensitivity The calculation expression is: in, and The kernel size of all convolutions is ; Then the spectral sensitivity characteristics Features of spatial sensitivity The spectral attention map and spatial attention map are obtained respectively using the spectral and spatial attention modules, and the calculation expression is as follows: in, and These represent spectral attention maps and spatial attention maps, respectively. represent Spectral attention mechanism represent Spatial attention mechanisms; This represents element-wise multiplication; S3.

4. Combine the spectral attention map and spatial attention map obtained in step S3.3 with the spatial features, respectively. and spectral characteristics Adding them together further highlights important classification information and suppresses irrelevant information. The calculation expression is: in, and These represent the spectral feature map after spectral attention enhancement and the spatial feature map after spatial attention enhancement, respectively; The final spectral and spatial feature maps are concatenated along the channel dimension to obtain an integrated spatial-spectral joint feature map. Then, the Flatten operation is used to reshape the integrated spatial-spectral joint feature map into a one-dimensional vector, which is then input into a fully connected layer for feature fusion. Finally, it is input into a Softmax layer for classification, calculating the probability that each pixel belongs to the water body. The calculation expression is as follows: in, , These represent the i-th and j-th elements in the input feature vector, respectively. Using cross-entropy loss function The parameters of the deep learning network model are updated using the following calculation expression: in, This represents the total number of wetland types; S3.

5. Using the classification probability of each pixel belonging to water body obtained in step S3.4, calculate the water body separation result map of the wetland; S4. Based on the water body separation result map of the wetland obtained in step S3, obtain the water body distribution map of the wetland. Combine the meteorological data of the wetland in the meteorological dataset of the wetland obtained in step S1 to solve the changing trend of wetland ecological factors and obtain the changing trend dataset of wetland ecological factors. S5. Construct a wetland drought index model; S6. Based on the wetland ecological factor change trend dataset obtained in step S4, the wetland drought index model obtained in step S5 is trained using the random forest regression algorithm to obtain an optimized wetland drought index model for assessing the wetland ecological drought situation.

2. The method for assessing wetland ecological drought conditions according to claim 1, characterized in that, The wetland image data mentioned in step S1 includes one of satellite remote sensing data, UAV remote sensing data, and ground remote sensing data.

3. The method for assessing wetland ecological drought conditions according to claim 2, characterized in that, In step S3.4 Spectral attention mechanism The specific implementation method is as follows: use global average pooling respectively and global maximum pooling The global features of the input spectral sensitivity are pooled, and then the resulting features are fed into a multilayer perceptron with hidden layers. The size of the hidden layer vector is set to... Where r represents the attenuation coefficient, the element-wise summation of the multilayer perceptron output is used to obtain the spectral attention weights, calculated as follows: in, Global features representing the spectral sensitivity of global average pooling. Global characteristics representing the spectral sensitivity of global max pooling. represent Activation function; Represents a multilayer perceptron. This represents the spectral attention weight.

4. The method for assessing wetland ecological drought conditions according to claim 3, characterized in that, In step S3.4 Spatial attention mechanism The specific implementation method is as follows: Global average pooling and global maximum pooling are used to pool the spatially sensitive global features of the input. Then, the feature maps of the two channels are processed by standard convolution to obtain the spatial attention map. The calculation expression is: in, The size of the convolution kernel is represented by , This represents spatial attention weights.

5. The method for assessing wetland ecological drought conditions according to claim 4, characterized in that, The specific implementation method of step S4 includes the following steps: S4.1 The water area of ​​the wetland is calculated by obtaining the number of pixels classified as water bodies through statistical analysis of the probability of each pixel belonging to a water body. The calculation expression is as follows: in, Represents the water area of ​​wetlands. This represents the total number of water body pixels in the water body separation result image of the wetland. This represents the resolution of the input remote sensing data; S4.

2. Based on the determination of wetland water body classification, the water level is calculated in the water body area using satellite altimetry data. The expression for calculating the water level is: in, The EGM96 reference water level represents the actual surface water level. This represents the distance between the altimeter and the ground. Represents the distance between the ground level and the Earth's ellipsoid. Represents the distance between the satellite and the water surface. Represents the correction parameter; The calculation expression is: in, , These represent dry and wet tropospheric corrections, respectively. Represents ionospheric correction. , These represent solid tides and polar tide corrections, respectively. S4.3 Based on the wetland water distribution map obtained in step S4.2, collect long-term humidity, rainfall and groundwater level data of the wetland water distribution map area in the wetland meteorological dataset obtained in step S1, and conduct trend analysis of wetland ecological factors. The wetland ecological factors are represented by y. S4.3.1 Arrange the wetland ecological factor data in chronological order to construct data pairs of independent variable time and dependent variable wetland ecological factor, where day represents time; assuming a linear relationship between wetland ecological factors and time day, the calculation expression is as follows: in, and They represent coefficients, Represents the error term; S4.3.2 Calculate the regression coefficients using the least squares method and The value of , the average of the calculation time. and the mean of wetland ecological factors The calculation expression is: coefficient The calculation expression is: coefficient The calculation expression is: Where n represents the number of data pairs; S4.3.3 Calculate the predicted value of the regression equation The calculation expression is: ； S4.3.4 Calculate the coefficient of determination To evaluate the goodness of fit of the model, The value of is between 0 and 1, with a value closer to 1 indicating a better model fit. The calculation expression is: ； By combining the meteorological data of the wetland obtained in step S1, the changing trends of wetland ecological factors are solved, and the wetland ecological drought situation is identified.

6. The method for assessing wetland ecological drought conditions according to claim 5, characterized in that, The specific implementation method of step S5 includes the following steps: S5.

1. Standardize the wetland ecological factors in the wetland ecological factor change trend dataset obtained in step S4. The calculation expression is as follows: in, Indicates the normalized humidity value. Indicates the wetland research area in Humidity at any given moment; and This represents the maximum and minimum humidity values ​​in the historical data of the wetland study area; and This represents the standardized values ​​of rainfall and groundwater level in the wetland study area; and This represents the rainfall and groundwater level in the wetland study area at time l. , , , These represent the maximum and minimum values ​​of rainfall and groundwater level in the historical data of the wetland study area, respectively. S5.

2. Based on the obtained standardized values ​​of humidity, rainfall, and groundwater level, a wetland drought index model is constructed, and the calculation expression is as follows: in, , , These represent the weights of humidity, rainfall, and groundwater level, respectively, and are used to adjust the influence of each ecological factor on the drought index.

7. The method for assessing wetland ecological drought conditions according to claim 6, characterized in that, The specific implementation method of step S6 includes the following steps: S6.1 Based on the wetland ecological factor change trend dataset obtained in step S4, the wetland drought index model obtained in step S5 is trained using the random forest regression algorithm to find the optimal regression coefficients. , , and The calculation expression is: in, This represents the optimization correction coefficient of the optimized wetland drought index model; S6.

2. Select one of the traditional drought indices, SPI, SPEI, or SRI, and compare it with the wetland drought index constructed in step S6.1 using Pearson correlation analysis on historical wetland ecological data. The calculation expression is as follows: in, This represents the Pearson correlation coefficient. This represents the constructed wetland ecological drought index. This represents the traditional drought index; The mean wetland ecological drought index was constructed for the selected study area. The average traditional drought index for the selected study area; The Pearson correlation coefficient ranges from 1 to -1. The larger the absolute value of the correlation coefficient, the stronger the correlation between the data. If the correlation coefficient is 0, it means that there is no correlation or a non-linear correlation. S6.

3. Based on historical data, the comprehensive ecological drought index of wetlands in the selected study area over the years was calculated. The least squares method was used to analyze the changing trend of the comprehensive ecological drought index of wetlands in the selected study area in recent years. The calculation expression of the least squares method is as follows: in, Represents the actual value; This represents the predicted value obtained through the assumed function; S6.

4. Based on the trend of the comprehensive drought index obtained in step S6.3, the trend of ecological status change in the study area over a long period is determined. Then, the collected relevant ecological factors are analyzed and predicted using the least squares method. 0.4 is set as severe drought; 0.4 <HRE 0.6 is set as moderate drought; 0.6 <HRE 0.8 was set as mild drought; HRE>0.8 was set as normal without drought, verifying the consistency and rationality with the changing trends of wetland ecological factors.

8. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the wetland ecological drought assessment method according to any one of claims 1-7.

9. A 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 wetland ecological drought assessment method according to any one of claims 1-7.

Images