Water area ecological service value evaluation method based on long-time sequence multi-source remote sensing data

Abstract

The invention discloses a water area ecological service value evaluation method based on long-time sequence multi-source remote sensing data, and relates to the technical field of ecological environment evaluation, and the method comprises the following steps: S1, obtaining multi-source remote sensing data and social economic and ecological data of a research area for many years, and building a comprehensive database; s3, counting food crop data over the years, and calculating grain production service prices of a unit area farmland ecosystem; S4, arranging water area ecological service value data over the years, and cleaning; and S5, importing various data into a model, and generating a water area ESV spatial-temporal change chart and an economic development coordination degree spatial-temporal change chart. The method has the advantages that the problems that in the prior art, water body extraction precision and efficiency are unbalanced, long-time sequence data support is lacked, and single-subject evaluation is limited are solved, more accurate, comprehensive and dynamic evaluation is achieved, and a powerful basis is provided for water area ecological protection and management.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological environment assessment, and focuses on an innovative method for accurately evaluating the ecological service value of water areas by using technical means such as long-term multi-source remote sensing data, Geographic Information System (GIS), and deep learning. Background Art

[0002] With the increasingly severe challenges faced by global environmental change and regional sustainable development, the importance of water ecosystems has become more prominent. However, the current water ecosystems are facing many problems such as pollution, ecological degradation, and biodiversity reduction. Accurately evaluating the ecological service value of water areas is crucial for the rational planning and management of water resources and the coordinated development of economy and ecology.

[0003] Existing assessment methods have many deficiencies. Some water body extraction technologies are difficult to balance between accuracy, computational resource requirements, and processing efficiency, resulting in inaccurate assessment results and poor timeliness. Traditional methods often lack the support of long-term time series data and are difficult to comprehensively reflect the spatio-temporal variation laws of the ecological service value of water areas. In addition, the assessment methods of a single discipline cannot fully consider the complexity and diversity of water ecosystems, limiting the scientificity and comprehensiveness of assessment results. Summary of the Invention

[0004] In view of the many defects existing in the current assessment methods for the ecological service value of water areas, such as the imbalance between water body extraction accuracy and efficiency, the lack of long-term time series data support, and the limitations of single-discipline assessment, etc., the present invention aims to provide an assessment method for the ecological service value of water areas based on long-term multi-source remote sensing data to achieve more accurate, comprehensive, and dynamic assessment, and provide a strong basis for the protection and management of water ecosystems.

[0005] An assessment method for the ecological service value of water areas based on long-term multi-source remote sensing data, comprising the following steps:

[0006] S1: Data collection. Cooperate with satellite data providers to obtain long-term (e.g., 2000 - 2022) multi-source remote sensing data (such as Landsat, GF-1 images), and at the same time collect socioeconomic data such as food crop information, GDP, and economic coordination from local government departments to construct a basic database.

[0007] S2: Data preprocessing and water body extraction. For the given multi-source remote sensing data, after preprocessing, use the NSWI combined with the Otsu threshold method to extract water bodies. Calculate the NSWI and then use the Otsu method combined with the local adaptive strategy to determine the threshold. Randomly select a large number of sample points, and obtain the true categories through on-site or high-resolution image interpretation, and evaluate the accuracy with a confusion matrix.

[0008] S3: Calculation of ecological service value. Statistically calculate the total price and sown area of food crops, and substitute them into Equation (1) to calculate the service price of food production in the farmland ecosystem per unit area. Obtain the authoritative equivalent value of the ecological service value of the water area, and combine it with the water area data to calculate the ecological service value of the water area using Equation (2).

[0009]

[0010] S4: Construction and prediction of the LSTM model. Import the data into Excel, clean and transform it, select the key columns, normalize them using Formula (3), create a time series dataset according to Formula (4), divide the training set and test set at a ratio of 70% and 30%, construct a two-layer LSTM model (select appropriate activation functions such as, optimizers such as, and an initial learning rate of 0.001), train it using the learning rate scheduler and early stopping callback function, predict the future values, restore them using the inverse scaling Formula (4), and evaluate the performance using Equation (5).

[0011]

[0012] original_value = scaled_value × (max - min) + min(5)

[0013]

[0014] S5: Comprehensive analysis and decision support. Import the data into the model, and use technologies such as spatial interpolation and overlay analysis to generate the spatio-temporal variation map of the ESV of the water area and the spatio-temporal variation map of the coordination degree with economic development. According to the analysis results, formulate ecological protection and management strategies for the region, and this method can be adjusted and applied to other regions.

[0015] Compared with the existing technologies, the advantages of the present invention are as follows: It provides a method for evaluating the ecological service value of water areas based on long-time series multi-source remote sensing data, solves the problems of the imbalance between the accuracy and efficiency of water body extraction, the lack of long-time series data support, and the limitations of single-discipline evaluation in the existing technologies, realizes a more accurate, comprehensive and dynamic evaluation, and provides a strong basis for the ecological protection and management of water areas. Description of the Drawings

[0016] Figure 1 It is a flow chart of the method for evaluating the ecological service value of water areas based on long-time series multi-source remote sensing data.

[0017] Figure 2 It is a variation map of the ecological service value (ESV) of water areas.

[0018] Figure 3 Spatial distribution map of the coordination between the ecological services of water areas and economic development

[0019] Figure 4 Loss curve diagram during the training process of the prediction model

[0020] Figure 5 The ESV prediction map of area 5 obtained by the prediction model. Detailed implementation manners

[0021] Referring to Figures 1-5 , a method for evaluating the ecological service value of water areas based on multi-source remote sensing data of long time series, comprising the following steps:

[0022] S1: Obtain multi-source remote sensing data of a specific research area during 2000-2022, including Landsat series satellite images, GF-1 images, etc. Meanwhile, collect local main food crop information, economic data, and ecological related data, and establish a comprehensive database.

[0023] S2: Preprocess the given multi-source remote sensing data. Use NSWI combined with the Otsu threshold method to extract water bodies, calculate NSWI and then use the Otsu threshold method combined with an adaptive strategy to segment the image. Randomly select a large number of sample points, obtain the true categories through on-site or high-resolution image interpretation, calculate the confusion matrix, and ensure the water body extraction accuracy.

[0024] S3: Statistically analyze the food crop data over the years, calculate the total price by integrating multi-channel price information, combine GIS and on-site investigation and measurement of the sown area, calculate the food production service price of the farmland ecosystem per unit area, and calculate the food production service price of the farmland ecosystem per unit area through Equation 1;

[0025]

[0026] wherein, Ea is the food production service price of the farmland ecosystem per unit area (yuan·hm -2 );

[0027] Ni is the total price of food crops in the i-th year (yuan);

[0028] Mi is the sown area of food crops in the i-th year (hm 2 );

[0029] Thus, obtain the authoritative value equivalent of the ecological service value of the water area, combine the water area data, and calculate the ecological service value of the water area using Equation 3:

[0030]

[0031] wherein, ESV is the total ecological service value of the water area in the research area (yuan);

[0032] Ek is the value equivalent of the k-th ecological service function of the water area;

[0033] Eij is the area of the j-th type of water area in the i-th region (hm 2 ).

[0034] Then, calculate the ecological service value of water areas based on the water area data, distinguishing different water body types and ecological service functions.

[0035] S4: Organize the historical ecological service value data of water areas, clean it, and map the data to a specific range using the normalization method. Use Equation 3 for normalization:

[0036]

[0037] where value is the original data value, min is the minimum value in the dataset, max is the maximum value in the dataset, and scaled_value is the normalized data value.

[0038] Then, create a time series dataset according to Equation 4:

[0039]

[0040] where data is the normalized data sequence, Xi is the i-th input sequence, including the data of three consecutive time steps, Yi is the i-th target value, that is, the data at the next time step after the input sequence. Divide the training set and the test set at 70% and 30%, construct a two-layer LSTM model, select appropriate parameters and compile it. During training, use the learning rate scheduler and the early stopping callback function for training, predict the future values, and then restore them according to the following formula:

[0041] original_value = scaled_value × (max - min) + min(5)

[0042] After training, predict the training set and the test set, reverse scale and restore the predicted values, calculate indicators such as the mean square error to evaluate the performance, and use Equation 6 to evaluate the performance:

[0043]

[0044] where n is the number of samples, Yi is the actual value, predi is the predicted value, and based on the model, predict the ESV values for the next 5 years, and conduct visual display and uncertainty analysis.

[0045] S5: Import various types of data into the model, use technologies such as spatial interpolation and overlay analysis to generate the spatio-temporal change map of the water area ESV and the spatio-temporal change map of the coordination degree with economic development. According to the analysis results, formulate ecological protection and management strategies for the study area.

[0046] As is known by common technical knowledge, the present invention can be implemented by other embodiments that do not depart from its spiritual essence or essential features. Therefore, the above-disclosed embodiments are illustrative in all respects and not exclusive. All changes within the scope of the present invention or within the scope equivalent to the present invention are encompassed by the present invention.

Claims

1. A method for assessing the value of water ecosystem services based on long-term series multi-source remote sensing data, characterized in that: The following steps are involved: S1: Data collection; Include multi-source remote sensing data and socio-economic data to build a basic database; S2: Preprocess the given data, then use NSWI combined with Otsu threshold method to extract water bodies, calculate NSWI, use Otsu method combined with local adaptive strategy to determine the threshold, and randomly select multiple sample points; S3: Collect data on grain crops over the years, calculate the total price based on price information from multiple channels, measure the sown area by combining GIS and field surveys, calculate the price of grain production services per unit area of ​​farmland ecosystems, obtain the equivalent value of water ecosystem services applicable to the study area, calculate the value of water ecosystem services based on water area data, and distinguish different water body types and ecological service functions; S4: Manage the historical water ecological service value data, clean it, use the normalization method to map the data to a specific range, construct a time series data set, divide it into training set and test set, build a two-layer LSTM model, select appropriate parameters and compile it, use the learning rate scheduler and early stopping callback function during training, predict the training set and test set after training, reverse scale and restore the predicted value, calculate the mean square error and other indicators to evaluate the performance, predict the ESV value of the next 5 years based on the model, and perform visualization and uncertainty analysis; S5: Import various data into the model to generate spatiotemporal changes in ESV of water areas and spatiotemporal changes in the degree of coordination with economic development. Based on the analysis results, formulate ecological protection and management strategies for the study area.

2. The method for evaluating the value of water ecosystem services based on long-term multi-source remote sensing data according to claim 1, characterized in that: In step S1, the multi-source remote sensing data collected includes the period from 2000 to 2022, including Landsat series satellite images and GF-1 images. At the same time, the collected socio-economic data includes local major food crop information, GDP, economic coordination and ecological related data.

3. The method for evaluating the value of water ecosystem services based on long-term series multi-source remote sensing data according to claim 1 is characterized by: In step S2, the true category is obtained by interpreting the field or high-resolution images, the confusion matrix is ​​calculated, and the accuracy is evaluated using the confusion matrix.

4. The method for evaluating the value of water ecosystem services based on long-term multi-source remote sensing data according to claim 1 is characterized by: The price of food production services per unit area of ​​farmland ecosystem is calculated by formula 1; Among them, Ea is the price of food production service per unit area of ​​farmland ecosystem (yuan·hm2 -2 ); Ni is the total price of grain crops in year i (yuan); Mi is the sown area of ​​grain crops in year i (hm 2 ); In this way, we can obtain the authoritative equivalent of water ecosystem service value, combine it with water area data, and use formula 3 to calculate the water ecosystem service value: Among them, ESV is the total ecosystem service value of the water area in the study area (yuan); E k is the value equivalent of the kth water ecosystem service function; Eij is the area of ​​the jth type of water in region i (hm 2 ).

5. The method for evaluating the value of water ecosystem services based on long-term series multi-source remote sensing data according to claim 1, characterized in that: In step S4, normalization is performed using formula 3: Among them, value is the original data value, min is the minimum value in the data set, max is the maximum value in the data set, and scaled_value is the normalized data value; Then create a time series dataset according to Formula 4: Among them, data is the normalized data sequence, Xi is the i-th input sequence, including data of three consecutive time steps, and Yi is the i-th target value, that is, the data of the next time step after the input sequence; Then divide the training set and test set, build a two-layer LSTM model, train with the learning rate scheduler and early stopping callback function, predict future values, and then restore according to the following formula: original_value=scaled_value×(max-min)+min(5) Finally, the performance is evaluated using Equation 6: Among them, n is the number of samples, Yi is the actual value, and predi is the predicted value.

6. The method for evaluating the value of water ecosystem services based on long-term multi-source remote sensing data according to claim 1, characterized in that: In step S5, spatial interpolation and overlay analysis techniques are used to generate spatiotemporal variation maps of the water ecosystem service value and the spatiotemporal variation maps of the degree of coordination with economic development.

Images