Satellite remote sensing water area change information extraction and carbon footprint analysis method and system

Abstract

The application discloses a satellite remote sensing water area change information extraction and carbon footprint analysis method and system. The data preprocessing and fusion module fuses and enhances the satellite remote sensing data after preprocessing and the processed OSM water system data in the research area, and generates fused water body data for input of water body analysis. The water body analysis module based on U-Net takes the fused water body data as input, trains the U-Net network, and generates water body interannual analysis, significant water body annual analysis and landscape fragmentation analysis in the research period, and water body classification information. The land use carbon emission interannual analysis module is an artificial intelligence algorithm and water body classification, which performs land classification and carbon emission coefficient in the research area to estimate the land use carbon emission in the research area interannually, and finally fuses the water body interannual analysis, significant water body annual analysis and landscape fragmentation analysis result to produce a comprehensive analysis of the influence factors of water body change on carbon emission.

Description

Technical Field

[0001] This invention relates to a method and system for extracting water change information and analyzing carbon footprint from satellite remote sensing data, applicable to the field of aerospace information. Background Technology

[0002] Urban water bodies, as an important component of natural water bodies, include rivers, springs, lakes, streams, reservoirs, and artificial lakes within cities. They are crucial for maintaining public health and the living and working environment, and are an important part of the urban environment. Therefore, monitoring and trend analysis of the spatiotemporal evolution characteristics of water bodies are particularly important.

[0003] Traditional field surveys and monitoring methods are time-consuming and labor-intensive, while rapidly developing satellite remote sensing technology enables rapid qualitative and quantitative surveys and monitoring of large areas of the Earth's surface. In recent years, high-resolution satellite remote sensing technology has provided strong data support for refined ground observation. The continuous supply of high-resolution Earth observation images is beneficial for long-term Earth observation and the study of land use change and its trends. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for extracting satellite remote sensing water area change information and analyzing carbon footprints, which can help us understand some past water area changes and contribute to the management and protection of water resources.

[0005] A method and system for extracting water body change information and analyzing carbon footprint from satellite remote sensing data is disclosed. The system includes a data preprocessing and fusion module, a U-Net-based water body analysis module, and an interannual land use carbon emission analysis module. By using input satellite remote sensing data and OSM (Optical System Management) water system data, a comprehensive analysis of the impact of water body changes on carbon emissions in the study area is achieved.

[0006] The data preprocessing and fusion module preprocesses satellite remote sensing data and then fuses and enhances it with processed OSM (OpenStreetMap) water system data within the study area, generating fused water body data for input to water body analysis.

[0007] The U-Net-based water analysis module takes the fused water data as input, trains the U-Net network, and generates interannual water analysis, significant annual water analysis, and landscape fragmentation analysis for the study period by integrating the results of water extraction methods such as MNDWI, AWEI, and SVM, as well as water classification information to support interannual land use carbon emission analysis.

[0008] The land use carbon emission interannual analysis module uses artificial intelligence algorithms and water body classification to classify land and carbon emission coefficients within the study area to estimate land use carbon emissions within the study area. Finally, it integrates the results of water body interannual analysis, significant water body annual analysis, and landscape fragmentation analysis to produce a comprehensive analysis of the factors affecting water body changes on carbon emissions.

[0009] The label creation method used in the U-Net-based water analysis module employs a fusion approach: direct image labeling, modification of water body data extracted using MNDWI, and modification of data downloaded from the OSM platform, all to obtain a high-quality sample database. Direct image labeling involves visually interpreting remote sensing images and generating corresponding labels; then, water body data is extracted using MNDWI, and the interpreted label data is modified; finally, a final correction is made based on the water system data converted from the OSM platform.

[0010] The water analysis module U-Net model, as described above, employs a combined loss function that integrates Focal loss and Dice loss to train the U-Net model. This approach combines the advantages of both loss functions, effectively enhancing the model's ability to extract water bodies that are too small.

[0011] A satellite remote sensing-based system for extracting water body change information and analyzing carbon footprints employs the aforementioned data preprocessing and fusion module, a U-Net-based water body analysis module, and a land use carbon emission interannual analysis module. By using input satellite remote sensing data and OSM water system data, it achieves a comprehensive analysis of the impact of water body changes on carbon emissions in the study area.

[0012] The beneficial effects of this invention are:

[0013] (1) By integrating remote sensing data with OSM data, a set of label libraries for training intelligent models was constructed, realizing the refined extraction of water bodies and the impact of water body changes on carbon footprint.

[0014] (2) It provides a new factor influencing carbon change. Attached Figure Description

[0015] Figure 1 This is a block diagram of the method and system structure for extracting information on changes in water bodies from satellite remote sensing and analyzing carbon footprints.

[0016] Figure 2 This is an administrative division map of Hangzhou.

[0017] Figure 3 This is a comparative analysis of interannual data from 1986 to 2008 on the eastern side of the Qiantang River. Detailed Implementation

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] like Figure 1 As shown, a method and system for extracting water change information and analyzing carbon footprint from satellite remote sensing data is presented. It includes a data preprocessing and fusion module, a U-Net-based water body analysis module, and an interannual land use carbon emission analysis module. By inputting satellite remote sensing data and OSM water system data, it achieves a comprehensive analysis of the impact of water changes on carbon emissions in the study area.

[0020] The data preprocessing and fusion module preprocesses satellite remote sensing data and then fuses and enhances it with processed OSM (OpenStreetMap) water system data within the study area to generate fused water body data for input to water body analysis. The remote sensing image preprocessing includes steps such as radiometric calibration, geometric correction, orthorectification, image mosaicking and fusion, and image cropping, aiming to process multi-satellite remote sensing data to a level suitable for data interpretation. The U-Net-based water body analysis module uses the fused water body data as input to train the U-Net network, then fuses the results of MNDWI, AWEI, and SVM water body extraction methods to generate interannual water body analysis, significant annual water body analysis, landscape fragmentation analysis, and water body classification information to support interannual land use carbon emission analysis within the study period.

[0021] The land use carbon emission interannual analysis module uses artificial intelligence algorithms and water body classification to classify land and carbon emission coefficients within the study area, and then estimates land use carbon emissions within the study area on an interannual basis. Finally, it integrates the results of water body interannual analysis, significant water body annual analysis, and landscape fragmentation analysis to produce a comprehensive analysis of the factors affecting water body changes on carbon emissions.

[0022] The labeling method used in the U-Net-based water analysis module employs a fusion approach: direct image labeling, modification of water body data extracted using MNDWI, and modification of data downloaded from the OSM platform. This method aims to obtain a high-quality sample database. Direct image labeling involves visually interpreting remote sensing images and generating corresponding labels. Then, water body data is extracted using MNDWI, and the interpreted label data is modified. Finally, a final correction is made based on the water system data converted from the OSM platform.

[0023] The water analysis module U-Net model, as described above, employs a combined loss function that integrates Focal loss and Dice loss to train the U-Net model. This approach combines the advantages of both loss functions, effectively enhancing the model's ability to extract water bodies that are too small. Example

[0024] Using Landsat time-series data as input, and taking the main urban area of ​​Hangzhou as the study area, the administrative divisions of Hangzhou are as follows: Figure 2 As shown.

[0025] Hangzhou is located in the eastern plains of China, with a network of rivers and lakes of varying sizes. Rivers, lakes, and reservoirs account for approximately 8% of the city's water system. The Qiantang River, famous for its tidal bore, flows through the city, which also features the beautiful Xixi Wetland and West Lake, as well as the Grand Canal that runs north-south through the city.

[0026] In terms of data processing, a system for extracting water body change information and analyzing carbon footprint in the main urban area of ​​Hangzhou City based on Landsat time-series data was first constructed. This system includes a data preprocessing and fusion module, a U-Net-based water body analysis module, and a land use carbon emission interannual analysis module. By inputting satellite remote sensing data and OSM (Open Street Map) water system data, a comprehensive analysis of the impact of water body changes on carbon emissions in the study area is achieved. The data preprocessing and fusion module preprocesses the satellite remote sensing data and then fuses and enhances it with processed OSM data from the study area, generating fused water body data for input to the water body analysis. Then, the U-Net-based water body analysis module trains the U-Net network using the fused water body data as input. By integrating the results of water body extraction methods such as MNDWI, AWEI, and SVM, interannual water body analysis, significant water body annual analysis, and landscape fragmentation analysis are generated for the study period, along with water body classification information to support the interannual analysis of land use carbon emissions. Finally, the land use carbon emission interannual analysis module uses artificial intelligence algorithms and water body classification to estimate land use carbon emissions interannually within the study area based on land classification and carbon emission coefficients. It then integrates the results of water body interannual analysis, significant water body annual analysis, and landscape fragmentation analysis to produce a comprehensive analysis of the factors influencing water body changes on carbon emissions. In the construction of this system, the labeling method used in the U-Net-based water body analysis module employs a fusion approach: direct image labeling, modification after water body extraction using MNDWI, and modification of data downloaded from the OSM platform, in order to obtain a high-quality sample database. Direct image labeling involves visually interpreting remote sensing images and generating corresponding labels; then, water bodies are extracted using MNDWI, and the interpreted label data is modified; finally, a final correction is made based on water system data converted from the OSM platform. The U-Net model training uses a combined loss function that integrates Focal loss and Dice loss, combining the advantages of both loss functions to effectively enhance the model's ability to extract small water bodies. A comparative analysis of the interannual changes in the eastern section of the Qiantang River within the main urban area of ​​Hangzhou from 1986 to 2010 reveals three main phases: 1986-1988, 1992-1996, and 2004-2008. During all three phases, the area of ​​the Qiantang River decreased. Figure 3 As shown.

[0027] Calculations show that between 1986 and 1988, the water area in this region decreased by approximately 36.99 km², exceeding the total reduction in the main urban area of ​​Hangzhou during that period. Between 1992 and 1996, the water area decreased by 20.69 km², about 80% of the reduction in the main urban area of ​​Hangzhou. Between 2004 and 2008, the water area decreased by approximately 13.13 km², about 50%. The area decreased in all three periods. From 1986 to 2010, the total area reduction on the east side of the Qiantang River in Hangzhou was approximately 78% of the reduction in the main urban area of ​​Hangzhou. However, the decreasing trend weakened during these three major periods. The period with the greatest change was from 1986 to 1988, when the water area reduction exceeded the combined reductions from 1992 to 1996 and from 2004 to 2008. This indicates that significant land use changes occurred in Hangzhou during the 1980s, leading to substantial changes in the management of the Qiantang River's water resources. Finally, based on the conversion area and carbon emission coefficient, the carbon emission changes caused by the changes in water area in each time period were estimated and an interannual analysis was conducted. The carbon emissions from the conversion of water area to other land use types are increasing. Since the carbon emission coefficient of building land is relatively large, the carbon emission increase brought about by the conversion of water area to building land is the largest. Among them, most of the land use in Xixi Wetland has been converted to building land. Therefore, the conversion of Xixi Wetland water area to building land is an important factor in the increase of carbon emissions.

[0028] The embodiments described above can be further implemented using different combinations or substitutions of satellite remote sensing data. These embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the concept and scope of the invention. Various changes and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the inventive concept are all within the protection scope of the present invention. The protection scope of the present invention is given by the appended claims and any equivalents.

Claims

1. A satellite remote sensing system for extracting water area change information and analyzing carbon footprint, characterized in that: It includes a data preprocessing and fusion module, a U-Net-based water body analysis module, and a land use carbon emission interannual analysis module. Through input satellite remote sensing data and OSM water system data, it can achieve a comprehensive analysis of the impact of water body changes on carbon emissions in the study area. The data preprocessing and fusion module preprocesses satellite remote sensing data and then fuses and enhances it with processed OSM water system data within the study area, generating fused water body data for input to water body analysis. The U-Net-based water analysis module takes the fused water data as input, trains the U-Net network, and then fuses the results of MNDWI, AWEI and SVM water extraction methods to generate interannual water analysis, annual analysis of significant water bodies and landscape fragmentation analysis within the study period, as well as water classification information to support interannual analysis of land use carbon emissions. The land use carbon emission interannual analysis module uses artificial intelligence algorithms and water body classification to classify land and carbon emission coefficients within the study area and estimate land use carbon emissions within the study area interannually. Finally, it integrates the results of water body interannual analysis, significant water body annual analysis and landscape fragmentation analysis to produce a comprehensive analysis of the factors affecting water body changes on carbon emissions. The label creation method used in the U-Net-based water analysis module employs a fusion approach: direct image labeling, modification of water body data extracted using MNDWI, and modification of data downloaded from the OSM platform, in order to obtain a high-quality sample database. Direct image labeling involves visually interpreting remote sensing images and generating corresponding labels. Then, water body data is extracted using MNDWI, and the interpreted label data is modified. Finally, a final correction is made based on the water system data converted from the OSM platform. The water analysis module based on U-Net uses a combined loss function that combines Focal loss and Dice loss to train the U-Net model.

2. A method for extracting water area change information and analyzing carbon footprint based on satellite remote sensing, characterized in that: This method performs the following steps: The satellite remote sensing data is preprocessed and then fused and enhanced with the processed OSM water system data in the study area to generate fused water body data for input to water body analysis. Using the fused water body data as input, the U-Net network is trained, and then the results of MNDWI, AWEI and SVM water body extraction methods are fused to generate interannual water body analysis, annual analysis of significant water bodies and landscape fragmentation analysis within the study period, as well as water body classification information to support interannual analysis of land use carbon emissions. Artificial intelligence algorithms and water body classification are used to classify land and carbon emission coefficients within the study area to estimate land use carbon emissions within the study area from year to year. Finally, the results of interannual water body analysis, significant annual water body analysis and landscape fragmentation analysis are integrated to produce a comprehensive analysis of the factors affecting water body changes on carbon emissions. The U-Net network employs a fusion method for label creation, which involves directly labeling against imagery, modifying water body data extracted using MNDWI, and modifying data downloaded from the OSM platform, in order to obtain a high-quality sample database. Direct image labeling involves visually interpreting remote sensing images and generating corresponding labels. Then, water body data is extracted using MNDWI, and the interpreted label data is modified. Finally, a final correction is made based on water system data converted from the OSM platform. The U-Net model is trained using a combined loss function that integrates Focal loss and Dice loss. By combining the advantages of both loss functions, the model's ability to extract small water areas is effectively enhanced.

Images