A method and system for evaluating degraded grassland restoration based on aerial photography and satellite remote sensing

Abstract

The present invention belongs to but is not limited to the field of satellite remote sensing technology, and in particular to a method and system for evaluating degraded grassland restoration based on aerial photography and satellite remote sensing. The degraded grassland restoration in the study area is observed by aerial photography based on the FragMap UAV aerial photography analysis system to obtain the coverage information of refined grassland vegetation. Based on the GEE Google Engine cloud computing platform, the STARFM high spatiotemporal data fusion algorithm is used to obtain a long-term series (2000-2022) of high spatiotemporal resolution NDVI data sets (with a time resolution of 16d and a spatial resolution of 30m), and the spatiotemporal change process of grassland coverage during the restoration of degraded grassland is carried out, thereby providing a theoretical basis for the adjustment of degraded grassland restoration strategies. The present invention establishes a remote sensing estimation model for grassland coverage, and qualitatively and quantitatively analyzes the degree of influence of degradation restoration measures on grassland coverage by analyzing the annual maximum grassland coverage in 2023 and the average annual maximum coverage from 2000 to 2022.

Description

Technical Field

[0001] The present invention belongs to but is not limited to the field of satellite remote sensing technology, and in particular relates to a degraded grassland restoration evaluation method and system based on aerial photography and satellite remote sensing. Background Art

[0002] Ground observations can provide high-resolution data, can capture subtle vegetation and soil characteristics, and can simultaneously monitor multiple key parameters such as vegetation, soil, hydrology, etc., provide comprehensive assessments, and obtain more accurate data. However, they are also limited in major vegetation surveys and field sampling. Ground observations can usually only cover limited areas, making it difficult to conduct large-scale assessments. Large-scale fixed-point repeated observations are difficult to achieve in actual research. To obtain long-term series data, a large amount of manpower and material resources are required for field investigations and long-term monitoring. Remote sensing technology has become an important means of evaluating grassland degradation due to its advantages of fast, efficient, large-scale, low-cost, and long-term series. Remote sensing can provide large-scale, continuous data. Compared with ground observations, satellite remote sensing is more cost-effective and can obtain a large amount of data in a shorter period of time, making it suitable for a wide range of regional assessments. Traditional medium- and low-resolution remote sensing satellite images have high temporal resolution and spatial coverage, but their spatial resolution is low. In comparison, although high-resolution satellite images have high spatial resolution, their temporal resolution is low. In addition, the growing season of grasslands on the Qinghai-Tibet Plateau is easily affected by weather conditions such as clouds and rain. Clouds and haze will affect the quality of satellite remote sensing data, and it is difficult to obtain large-scale and long-term series data sets.

[0003] In recent years, drone aerial photography technology has developed rapidly. Its aerial photography is efficient and fast, and can quickly cover a large area. Compared with traditional ground surveys, it saves a lot of time and labor costs, can provide high-resolution image data, and can take aerial photos at different heights and angles, accurately capture subtle surface features and vegetation information, which is helpful for refined evaluation and monitoring, and is conducive to a comprehensive understanding of grassland conditions. In addition, drone aerial photography data can be fed back to the ground in real time, and can be adjusted and processed in time. At the same time, Yi et al. also developed a drone aerial photography system based on the climate and environmental characteristics of my country's fragile ecological belt, which can realize the positioning and repeated observation of a large number of monitoring points, and can provide massive fixed-point monitoring data for grassland resource monitoring. In addition, with the development of remote sensing data fusion algorithms, multi-source remote sensing data fusion products have been widely used in vegetation remote sensing monitoring, especially the spatial and temporal adaptive reflection fusion model (STARFM). The STARFM algorithm uses the information of neighboring similar pixels as a reference to predict the pixel values ​​of high-resolution images. That is, with the predicted pixel as the center, the similarity of the pixel values, spatial distance, and temporal distance are used as the selection basis to find the similar points of the predicted pixels in the sliding window, and all similar pixels are given certain weights. Finally, the value of the intermediate predicted pixel is obtained, and a convolution operation is performed on the entire image to obtain the predicted value of the entire image. Combined with the STARFM high-temporal and spatial data fusion algorithm, a high-temporal and spatial resolution NDVI data set can be obtained, providing a remote sensing data basis for the spatiotemporal dynamic changes of grassland biomass and coverage during grassland restoration in the study area.

[0004] In view of the above analysis, the technical problems that urgently need to be solved in the existing technology are: in the existing evaluation methods, ground observation is time-consuming and labor-intensive, and it is difficult to monitor for a long time; satellite remote sensing is limited to limited spatial and temporal resolution, making it difficult to obtain detailed vegetation information. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a method and system for evaluating the restoration of degraded grassland based on aerial photography and satellite remote sensing.

[0006] The present invention is implemented as follows: a method for evaluating the restoration of degraded grassland based on aerial photography and satellite remote sensing, comprising:

[0007] S1, ground survey / UAV aerial photography: Aerial observation of the degraded and restored grassland in the study area was carried out based on the FragMap UAV aerial photography analysis system to monitor the coverage and biomass information of grassland vegetation;

[0008] S2, satellite remote sensing data acquisition and preprocessing: using the GEE Google Engine cloud computing platform, obtain the MOD13Q1 and Landsat NDVI datasets before restoration;

[0009] S3, preprocessing of drone aerial photos: based on the work point and waypoint route information uploaded to the server by the field observation tablet, use DJI Locator software to add field observation records to the database, and select routes of different flight modes to locate aerial photos; according to the latitude and longitude information of the waypoint records and the location information of the drone aerial photos, the drone aerial photos are matched with the waypoints one by one, and numbered from 1 to 16 in sequence according to the waypoint sequence number, and each photo entering the database is assigned a unique DJIPT ID number for later data call and maintenance;

[0010] S4, aerial photo coverage extraction: The coverage of aerial photos is obtained by using the vegetation patch analysis software Pixel Based Manual Classifier. The software calculates the greenness index based on the threshold method to achieve binary processing of vegetation and bare soil, thereby calculating the vegetation part as the grassland coverage value;

[0011] S5, EGI threshold division of aerial photos: UAV images are stored in JPEG format, and the grayscale value range of the three bands R, G, and B is 0-255, corresponding to the EGI index range of -500 to 500. In order to improve processing efficiency, referring to the previous research results, the EGI index of alpine vegetation on the Qinghai-Tibet Plateau is generally between 60 and 140. Therefore, the initial threshold is set within this range;

[0012] S6, construction of remote sensing estimation model and accuracy test: Based on the Pixel Based Manual Classifier software, the coverage of all aerial photos was obtained, and the fused NDVI vegetation index value corresponding to each aerial photo was extracted in turn; with NDVI as the independent variable and the grassland coverage obtained from drone aerial photos as the dependent variable, linear, exponential, logarithmic and power remote sensing estimation models of grassland coverage were constructed respectively, combined with the leave-one-out cross-validation method, the correlation coefficient and root mean square error between the validation set and the measured data were used to evaluate, and the optimal grassland biomass estimation model for alpine grassland was screened;

[0013] S7, grassland coverage inversion before restoration: Based on the selected optimal grassland coverage remote sensing inversion model, combined with the NDVI fusion data set of the year before restoration, the grassland coverage data before restoration was inverted, and the annual maximum grassland coverage before restoration was calculated using the MVC maximum synthesis algorithm. The average value of the maximum coverage in the year before restoration was calculated, and the spatial differentiation characteristics of the maximum coverage value in the year before restoration were analyzed.

[0014] S8. Analysis of grassland coverage changes before and after restoration: Based on the average maximum coverage in the year before restoration and the maximum grassland coverage in 2023, the grassland coverage deviation after restoration in the study area was calculated. Combined with the boundary data of the implementation range of different degradation restoration measures in the study area, the grassland coverage before and after the treatment of different degraded restoration grasslands was extracted, and the impact of degradation restoration measures on grassland coverage was qualitatively and quantitatively evaluated.

[0015] Furthermore, the sample plot layout mainly follows the following rules:

[0016] 1) Sample plots are set up according to the scope and terrain conditions of the study area;

[0017] 2) Set up a 250m×250m GRID flight route, and the route setting evenly covers the entire study area;

[0018] 3) The GRID flight route has a flight altitude of 20m, the photo resolution is better than 1cm, and the photo coverage area is 26m×35m;

[0019] 4) 16 waypoints are evenly distributed within the route range, and an aerial photo is taken vertically downward at each waypoint to obtain data. The data obtained includes the grassland vegetation coverage and biomass information monitored by drone aerial photos.

[0020] Furthermore, satellite remote sensing data acquisition and preprocessing specifically include:

[0021] The STARFM algorithm is used to fuse multiple remote sensing images with different spatial resolutions and temporal resolutions into an image with high spatial resolution and high temporal resolution. The relationship between low-resolution images and high-resolution images at different time points is analyzed, and this relationship is used to derive the spatial distribution and temporal variation information of the high-resolution image. A linear or nonlinear method is used to establish a spatial reflectance model to describe the relationship between images at different spatial resolutions. The time series analysis statistical model is then used to describe the law of image variation over time. The weight adjustment mechanism of both space and time is combined to adaptively adjust the weights according to different scenes and characteristics to achieve high quality and high precision in the image fusion process.

[0022] Furthermore, in the extraction of aerial photo coverage, the greenness index calculation formula is as follows:

[0023] EGI=2G-RB (1)

[0024] Where: EGI is the greenness index, R, G, and B represent the red, green, and blue bands of the drone image, respectively;

[0025] Calculate the greenness index of each pixel in the drone image in turn, set the initial threshold of the greenness index, and compare it with the greenness index value of the image. If the greenness index of the pixel is greater than the set threshold, the pixel is regarded as a vegetation pixel, otherwise it is regarded as a bare soil pixel; compare the classified image with the original image for visual judgment. If the classification result is not ideal, repeat the above process until the vegetation area in the classified image is consistent with the vegetation area in the original image.

[0026] Furthermore, the construction and accuracy test of remote sensing estimation model specifically include:

[0027] The leave-one-out cross validation (LOOCV) method is used to select one group of sample plot data from n groups of sample plot data as the validation data, and the remaining n-1 groups of sample plot data are fitted into an NDVI estimation model, and the selected validation data is used to evaluate the accuracy of the estimation model. This is repeated n times and the average of the n evaluation indicators is taken as the final evaluation indicator; the prediction ability of the model is determined by the correlation coefficient R and RMSE (root mean square error, RMSE) between the model evaluation results and the observed values, where R is often used to measure the accuracy of the model, and RMSE is often used to quantify the accuracy of the model, that is, the closer R is to 1, the lower the RMSE value is, and the higher the model accuracy is;

[0028] The calculation formula of R is:

[0029]

[0030] The calculation formula of RMSE is:

[0031]

[0032] In the formula, y i is the grassland biomass value measured in the field, is the average value of the measured grassland biomass, y′ i The values ​​of grassland biomass estimated by the model.

[0033] Another object of the present invention is to provide a degraded grassland restoration evaluation system based on aerial photography and satellite remote sensing, comprising:

[0034] Ground survey / UAV aerial photography module: Based on the FragMap UAV aerial photography analysis system, aerial observation of the degraded and restored grassland in the study area is carried out to monitor the coverage and biomass information of grassland vegetation;

[0035] Satellite remote sensing data acquisition and preprocessing module: using GEE Google Engine cloud computing platform to obtain MOD13Q1 and Landsat NDVI datasets before restoration;

[0036] UAV aerial photo preprocessing module: Based on the work point and waypoint route information uploaded to the server by the field observation tablet, the DJI Locator software is used to add field observation records to the database, and routes of different flight modes are selected to locate aerial photos; according to the latitude and longitude information of the waypoint records and the location information of the drone aerial photo attributes, the drone aerial photos are matched with the waypoints one by one, and are numbered from 1 to 16 in sequence according to the waypoint sequence number. At the same time, each photo entering the database is assigned a unique DJIPT ID number for later data call and maintenance;

[0037] Aerial photo coverage extraction module: The coverage of aerial photos is obtained by using the vegetation patch analysis software Pixel Based Manual Classifier. The software calculates the greenness index based on the threshold method to achieve binary processing of vegetation and bare soil, thereby calculating the vegetation part as the grassland coverage value;

[0038] Aerial photo EGI threshold division module: UAV images are stored in JPEG format, and the grayscale value range of the three bands R, G, and B is 0-255, corresponding to the EGI index range of -500 to 500. In order to improve processing efficiency, referring to the previous research results, the EGI index of alpine vegetation on the Qinghai-Tibet Plateau is generally between 60 and 140. Therefore, the initial threshold is set within this range;

[0039] Remote sensing estimation model construction and accuracy test module: Based on the PixelBasedManual Classifier software, the coverage of all aerial photos is obtained, and the fused NDVI vegetation index value corresponding to each aerial photo is extracted in turn; with NDVI as the independent variable and the grassland coverage obtained from drone aerial photos as the dependent variable, linear, exponential, logarithmic and power remote sensing estimation models of grassland coverage are constructed respectively, combined with the leave-one-out cross-validation method, the correlation coefficient and root mean square error between the validation set and the measured data are used for evaluation to screen the optimal grassland biomass estimation model for alpine grasslands;

[0040] Grassland coverage inversion module before restoration: Based on the selected optimal grassland coverage remote sensing inversion model, combined with the NDVI fusion data set of the year before restoration, the grassland coverage data before restoration was inverted, and the annual maximum grassland coverage before restoration was calculated using the MVC maximum synthesis algorithm. The average value of the maximum coverage in the year before restoration was calculated, and the spatial differentiation characteristics of the maximum coverage value in the year before restoration were analyzed.

[0041] Analysis module of grassland coverage changes before and after restoration: Based on the average maximum coverage in the year before restoration and the maximum grassland coverage in 2023, the grassland coverage deviation after restoration in the study area is calculated, and combined with the boundary data of the implementation range of different degradation restoration measures in the study area, the grassland coverage before and after the treatment of different degraded restoration grasslands is extracted, and the impact of degradation restoration measures on grassland coverage is qualitatively and quantitatively evaluated.

[0042] The present invention provides a computer device, which includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of the degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing.

[0043] The present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing.

[0044] The present invention provides an information data processing terminal, which includes the degraded grassland restoration evaluation system based on aerial photography and satellite remote sensing.

[0045] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0046] First, the present invention combines Landsat and MODIS NDVI data, uses the STARFM spatiotemporal data fusion algorithm and the GEE cloud computing platform to generate a NDVI data set with a temporal resolution of 16d and a spatial resolution of 30m in the Yeniugou study area of ​​Qilian County, Haibei Prefecture from 2000 to 2023. On this basis, a grassland coverage remote sensing estimation model was established, and the impact of degradation restoration measures on grassland coverage was qualitatively and quantitatively analyzed by analyzing the annual maximum grassland coverage in 2023 and the average annual maximum coverage from 2000 to 2022. The main conclusions are as follows: 1) Among the four types of remote sensing estimation models based on high temporal and spatial resolution NDVI datasets, the linear model has the highest accuracy, with R and RMSE of 0.82 and 12.02 respectively; 2) From 2000 to 2023, the grassland coverage in the Yeniugou study area has increased significantly as a whole, with an overall average coverage of 42.15%, an increase of 63.63% over the average coverage of the previous 23 years (25.77%); 3) According to the deviation results of the grassland coverage in 2023 from the average grassland coverage of the previous 23 years, it can be clearly observed that the coverage increment of Plot1-1 and Plot1-2 is large, indicating that the restoration measures in this area have a good restoration effect. Therefore, it is of great significance to understand the effectiveness of restoration measures, guide the planning and implementation of future restoration activities, and maintain and enhance the functions and services of alpine grassland ecosystems.

[0047] Second, as auxiliary evidence of the inventiveness of the claims of the present invention, it is also reflected in the following important aspects:

[0048] (1) The expected benefits and commercial value of the technical solution of the present invention after transformation are as follows: The degraded grassland restoration evaluation method and system based on aerial photography and satellite remote sensing proposed in the present invention, combined with drone aerial photography and satellite remote sensing technology, can achieve a refined assessment of the spatiotemporal dynamic changes of grassland biomass and coverage during large-scale grassland restoration. This method is of great significance for the monitoring, management and restoration of grassland ecosystems, and is expected to provide a scientific basis for grassland resource protection and sustainable land use, and is expected to bring significant economic and social benefits.

[0049] (2) The technical solution of the present invention fills the technical gap in the industry at home and abroad: In the field of grassland restoration evaluation, traditional ground observation and satellite remote sensing have limitations in temporal resolution, spatial resolution and data acquisition. The comprehensive evaluation method and system based on aerial photography and satellite remote sensing of the present invention fill the technical gap in this field, effectively combining the advantages of aerial photography and satellite remote sensing, and providing an efficient, accurate and comprehensive evaluation solution.

[0050] (3) The technical solution of the present invention solves a technical problem that people have long been eager to solve but have never been able to solve: traditional grassland restoration evaluation methods are limited by data acquisition and analysis methods, making it difficult to achieve refined monitoring and evaluation of large-scale grassland restoration processes. The evaluation method and system based on aerial photography and satellite remote sensing provided by the present invention overcome this difficulty and effectively solve the technical problem of evaluating the spatiotemporal dynamic changes in grassland biomass and coverage.

[0051] (4) The technical solution of the present invention overcomes technical bias: In the field of grassland restoration evaluation, people generally prefer traditional ground observation methods or single satellite remote sensing data analysis. This traditional evaluation method has many limitations, such as time-consuming and labor-intensive ground observation and low resolution of satellite remote sensing data. The present invention overcomes this technical bias and proposes an evaluation method that comprehensively utilizes aerial photography and satellite remote sensing data, making full use of their respective advantages, making up for the shortcomings of a single data acquisition method, and giving full play to the advantages of various data sources. It provides a new perspective and method for grassland restoration evaluation and realizes comprehensive and efficient monitoring and evaluation of grassland conditions.

[0052] Third, existing technologies have many problems in evaluating the effects of degraded grassland restoration. First, traditional ground survey methods are time-consuming and labor-intensive, with limited coverage, and cannot timely reflect the dynamic changes of large-scale grasslands. Second, existing satellite remote sensing data processing technologies have problems such as low resolution and large time intervals, making it difficult to accurately monitor subtle changes in grassland vegetation. In addition, data acquisition and processing lack standardization and automation, resulting in low data processing efficiency and inaccurate and incomplete results. These problems have seriously affected the scientific evaluation and effective management of the effects of degraded grassland restoration.

[0053] The present invention solves many problems existing in the prior art by combining drone aerial photography and satellite remote sensing technology. First, by using drone aerial photography technology, the coverage and biomass information of grassland vegetation can be obtained in real time with high resolution, realizing large-scale and high-precision grassland monitoring. Secondly, by using cloud computing platform to process satellite remote sensing data, by fusing MOD13Q1 and Landsat NDVI data sets, the spatial and temporal resolution of the data is significantly improved, which can more accurately reflect the dynamic changes of grassland vegetation. In addition, the present invention improves the efficiency and accuracy of data processing through standardized data preprocessing procedures and automated data processing algorithms.

[0054] The present invention also proposes a vegetation coverage extraction method based on the greenness index and a remote sensing estimation model construction and accuracy verification method. Through the calculation of the greenness index and the threshold division, vegetation and bare soil can be accurately distinguished, and the automatic extraction of grassland coverage can be realized. The remote sensing estimation model is constructed and verified by the leave-one-out cross-validation method, and the accuracy of the model is evaluated by the correlation coefficient and the root mean square error, ensuring the reliability and accuracy of the estimation results. These innovative methods significantly improve the evaluation quality of the restoration effect of degraded grassland.

[0055] In summary, the present invention has made significant technical progress in the evaluation of the restoration effect of degraded grassland. By combining drone aerial photography, satellite remote sensing, cloud computing and automated data processing technology, high-resolution, real-time and accurate large-scale grassland monitoring is achieved. Standardized data preprocessing procedures and accurate estimation models ensure the efficiency of data processing and the reliability of results. The above technological advances provide strong technical support for the scientific evaluation of the restoration effect of degraded grassland, and have important application value and promotion prospects.

[0056] Fourth, existing technologies face problems such as incomplete data acquisition, inaccurate processing methods, and insufficient model prediction capabilities in the evaluation of degraded grassland restoration effects. Specifically, ground surveys are time-consuming and labor-intensive and cannot reflect the dynamic changes of large-scale grasslands in a timely manner; satellite remote sensing data has low resolution and large time intervals, making it difficult to accurately monitor subtle changes in grassland vegetation; and there is a lack of effective mathematical models to comprehensively analyze and predict grassland restoration effects. These problems make it difficult for existing evaluation methods to provide scientific and comprehensive decision support.

[0057] The present invention solves the problem of incomplete data acquisition by introducing drone aerial photography and satellite remote sensing technology. The drone aerial photography system can obtain grassland vegetation coverage and biomass information in real time with high resolution, and combine it with satellite remote sensing data for efficient processing through a cloud computing platform. The main parameters include: resolution, coverage, flight altitude, shooting angle of drone aerial photos, and NDVI value of satellite remote sensing data. The acquisition and processing of these parameters realize the comprehensiveness and accuracy of grassland vegetation monitoring.

[0058] The present invention solves the problems of inaccurate processing methods and insufficient model prediction capabilities in the prior art by constructing a mathematical model. Specifically, by calculating the greenness index (EGI) and dividing the threshold, accurate distinction between vegetation and bare soil is achieved, and grassland coverage values ​​are extracted. Based on the obtained aerial photo coverage values ​​and NDVI vegetation index values, linear, exponential, logarithmic and power remote sensing estimation models of grassland coverage are constructed, and the leave-one-out cross-validation method is used to test the model accuracy. These mathematical models make the estimation of grassland coverage and biomass more scientific and accurate.

[0059] The innovative method and mathematical model of the present invention significantly improve the quality and efficiency of the evaluation of the restoration effect of degraded grassland. First, by using high-resolution aerial photographs and high-temporal-resolution satellite remote sensing data, combined with automated data processing algorithms, real-time monitoring and accurate evaluation of grassland restoration effects are achieved. Secondly, standardized data preprocessing procedures and precise mathematical models ensure the efficiency of data processing and the reliability of results. Finally, based on the inversion analysis of the remote sensing estimation model, the changes in grassland coverage before and after restoration can be quantitatively evaluated, providing a scientific basis for the optimization of degraded grassland restoration measures.

[0060] In summary, the present invention solves the problems of incomplete data acquisition, inaccurate processing methods, and insufficient model prediction capabilities in the prior art by introducing efficient data acquisition means and accurate mathematical models, and achieves significant technical progress in the evaluation of degraded grassland restoration effects. These technological advances not only improve the scientificity and accuracy of grassland monitoring and restoration effect evaluation, but also provide important technical support for grassland management and protection, and have broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a flow chart of a degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing provided by an embodiment of the present invention;

[0062] Figure 2 It is a structural diagram of a degraded grassland restoration evaluation system based on aerial photography and satellite remote sensing provided by an embodiment of the present invention;

[0063] Figure 3 is a schematic diagram of the location of the study area and the distribution of sampling points provided by an embodiment of the present invention;

[0064] Figure 4 : is a schematic diagram of data fusion of the study area provided by an embodiment of the present invention; wherein (a) is the original Landsat image (affected by cloud shadow); (b) is the original MODIS image (coarse resolution); (c) is the fused image;

[0065] Figure 5 is a schematic diagram of MODIS and Landsat image data provided by an embodiment of the present invention;

[0066] Figure 6 This is a schematic diagram of positioning of drone aerial photos provided by an embodiment of the present invention;

[0067] Figure 7 is a flowchart of EGI threshold division of aerial photos provided by an embodiment of the present invention;

[0068] Figure 8 Schematic diagram of EGI division software for aerial photos provided by an embodiment of the present invention;

[0069] Fig. 9 It is a schematic diagram of a grassland coverage remote sensing estimation model based on fusion NDVI provided in an embodiment of the present invention;

[0070] Fig.10 : is a schematic diagram of grassland coverage provided by an embodiment of the present invention; wherein, (a) the annual maximum average value of grassland coverage from 2000 to 2023, (b) the spatial distribution of the annual maximum value of grassland coverage in 2023;

[0071] Fig.11 Schematic diagram of grassland coverage difference analysis (p<0.05) in different experimental sample areas provided by an embodiment of the present invention;

[0072] Fig.12 It is a deviation map of grassland coverage in 2023 and the average grassland coverage in the previous 23 years provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0073] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0074] like Figure 1 As shown, the degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing provided by the embodiment of the present invention includes:

[0075] S1, ground survey / UAV aerial photography: Aerial observation of the degraded and restored grassland in the study area was carried out based on the FragMap UAV aerial photography analysis system to monitor the coverage and biomass information of grassland vegetation;

[0076] S2, satellite remote sensing data acquisition and preprocessing: using the GEE Google Engine cloud computing platform, obtain the MOD13Q1 and Landsat NDVI datasets before restoration;

[0077] S3, preprocessing of drone aerial photos: based on the work point and waypoint route information uploaded to the server by the field observation tablet, use DJI Locator software to add field observation records to the database, and select routes of different flight modes to locate aerial photos; according to the latitude and longitude information of the waypoint records and the location information of the drone aerial photos, the drone aerial photos are matched with the waypoints one by one, and numbered from 1 to 16 in sequence according to the waypoint sequence number, and each photo entering the database is assigned a unique DJIPT ID number for later data call and maintenance;

[0078] S4, aerial photo coverage extraction: The coverage of aerial photos is obtained by using the vegetation patch analysis software Pixel Based Manual Classifier. The software calculates the greenness index based on the threshold method to achieve binary processing of vegetation and bare soil, thereby calculating the vegetation part as the grassland coverage value;

[0079] S5, EGI threshold division of aerial photos: UAV images are stored in JPEG format, and the grayscale value range of the three bands R, G, and B is 0-255, corresponding to the EGI index range of -500 to 500. In order to improve processing efficiency, referring to the previous research results, the EGI index of alpine vegetation on the Qinghai-Tibet Plateau is generally between 60 and 140. Therefore, the initial threshold is set within this range;

[0080] S6, construction of remote sensing estimation model and accuracy test: Based on the Pixel Based Manual Classifier software, the coverage of all aerial photos was obtained, and the fused NDVI vegetation index value corresponding to each aerial photo was extracted in turn; with NDVI as the independent variable and the grassland coverage obtained from drone aerial photos as the dependent variable, linear, exponential, logarithmic and power remote sensing estimation models of grassland coverage were constructed respectively, combined with the leave-one-out cross-validation method, the correlation coefficient and root mean square error between the validation set and the measured data were used to evaluate, and the optimal grassland biomass estimation model for alpine grassland was screened;

[0081] S7, grassland coverage inversion before restoration: Based on the selected optimal grassland coverage remote sensing inversion model, combined with the NDVI fusion data set of the year before restoration, the grassland coverage data before restoration was inverted, and the annual maximum grassland coverage before restoration was calculated using the MVC maximum synthesis algorithm. The average value of the maximum coverage in the year before restoration was calculated, and the spatial differentiation characteristics of the maximum coverage value in the year before restoration were analyzed.

[0082] S8. Analysis of grassland coverage changes before and after restoration: Based on the average maximum coverage in the year before restoration and the maximum grassland coverage in 2023, the grassland coverage deviation after restoration in the study area was calculated. Combined with the boundary data of the implementation range of different degradation restoration measures in the study area, the grassland coverage before and after the treatment of different degraded restoration grasslands was extracted, and the impact of degradation restoration measures on grassland coverage was qualitatively and quantitatively evaluated.

[0083] like Figure 2 As shown, the degraded grassland restoration evaluation system based on aerial photography and satellite remote sensing provided by the embodiment of the present invention includes:

[0084] Ground survey / UAV aerial photography module: Based on the FragMap UAV aerial photography analysis system, aerial observation of the degraded and restored grassland in the study area is carried out to monitor the coverage and biomass information of grassland vegetation;

[0085] Satellite remote sensing data acquisition and preprocessing module: using GEE Google Engine cloud computing platform to obtain MOD13Q1 and Landsat NDVI datasets before restoration;

[0086] UAV aerial photo preprocessing module: Based on the work point and waypoint route information uploaded to the server by the field observation tablet, the DJI Locator software is used to add field observation records to the database, and routes of different flight modes are selected to locate aerial photos; according to the latitude and longitude information of the waypoint records and the location information of the drone aerial photo attributes, the drone aerial photos are matched with the waypoints one by one, and are numbered from 1 to 16 in sequence according to the waypoint sequence number. At the same time, each photo entering the database is assigned a unique DJIPT ID number for later data call and maintenance;

[0087] Aerial photo coverage extraction module: The coverage of aerial photos is obtained by using the vegetation patch analysis software Pixel Based Manual Classifier. The software calculates the greenness index based on the threshold method to achieve binary processing of vegetation and bare soil, thereby calculating the vegetation part as the grassland coverage value;

[0088] Aerial photo EGI threshold division module: UAV images are stored in JPEG format, and the grayscale value range of the three bands R, G, and B is 0-255, corresponding to the EGI index range of -500 to 500. In order to improve processing efficiency, referring to the previous research results, the EGI index of alpine vegetation on the Qinghai-Tibet Plateau is generally between 60 and 140. Therefore, the initial threshold is set within this range;

[0089] Remote sensing estimation model construction and accuracy test module: Based on the Pixel Based Manual Classifier software, the coverage of all aerial photos is obtained, and the fused NDVI vegetation index value corresponding to each aerial photo is extracted in turn; with NDVI as the independent variable and the grassland coverage obtained from drone aerial photos as the dependent variable, linear, exponential, logarithmic and power remote sensing estimation models of grassland coverage are constructed respectively, combined with the leave-one-out cross-validation method, the correlation coefficient and root mean square error between the validation set and the measured data are used for evaluation to screen the optimal grassland biomass estimation model for alpine grasslands;

[0090] Grassland coverage inversion module before restoration: Based on the selected optimal grassland coverage remote sensing inversion model, combined with the NDVI fusion data set of the year before restoration, the grassland coverage data before restoration was inverted, and the annual maximum grassland coverage before restoration was calculated using the MVC maximum synthesis algorithm. The average value of the maximum coverage in the year before restoration was calculated, and the spatial differentiation characteristics of the maximum coverage value in the year before restoration were analyzed.

[0091] Analysis module of grassland coverage changes before and after restoration: Based on the average maximum coverage in the year before restoration and the maximum grassland coverage in 2023, the grassland coverage deviation after restoration in the study area is calculated, and combined with the boundary data of the implementation range of different degradation restoration measures in the study area, the grassland coverage before and after the treatment of different degraded restoration grasslands is extracted, and the impact of degradation restoration measures on grassland coverage is qualitatively and quantitatively evaluated.

[0092] 1. Data and Methods

[0093] 1.1 Overview of the study area

[0094] The Yeniugou area is located in the northwest of Qilian County, Haibei Tibetan Autonomous Prefecture, Qinghai Province ( Figure 3 ). The alpine grassland in the study area is seriously degraded due to factors such as permafrost melting and overgrazing, with low grassland productivity, single grassland species and fragile ecosystem. If degraded grassland restoration measures are not implemented in time, it is very easy to lead to grassland desertification.

[0095] 1.2 Ground survey / UAV aerial photography

[0096] Based on the FragMap UAV aerial photography analysis system, aerial observations of the degraded and restored grassland in the study area were carried out to monitor the coverage and biomass information of grassland vegetation. The observation time was July 2023. The sample plot layout mainly follows the following rules: 1) The sample plot layout is set according to the scope and terrain conditions of the study area; 2) A 250m×250m GRID flight route is set, and the route setting evenly covers the entire study area; 3) The GRID flight route is 20m high, the photo resolution is better than 1cm, and the photo coverage range is 26m×35m; 4) 16 waypoints are evenly arranged within the route range, and 1 aerial photo is taken vertically downward at each waypoint to obtain data. The data obtained includes monitoring grassland vegetation coverage and biomass information with the help of drone aerial photos. In July 2023, field drone aerial photography was carried out in the study area, and a total of 176 GRID drone aerial photos were completed.

[0097] 1.3 Satellite remote sensing data acquisition and preprocessing

[0098] Google Earth Engine (GEE) is a powerful cloud-based geographic information processing platform. The present invention uses the GEE Google Engine cloud computing platform to obtain MOD13Q1 and Landsat NDVI data sets from 2000 to 2023. The STARFM algorithm is an algorithm for remote sensing image fusion, and its full name is Spatial and Temporal Adaptive Reflectance Fusion Model. The algorithm is mainly used to fuse multiple remote sensing images with different spatial resolutions and temporal resolutions into an image with high spatial resolution and high temporal resolution. The core idea of ​​the STARFM algorithm is to analyze the relationship between low-resolution images and high-resolution images at different time points, and use this relationship to derive the spatial distribution and temporal variation information of high-resolution images. The algorithm usually uses linear or nonlinear methods to establish a spatial reflectance model to describe the relationship between images at different spatial resolutions, and then uses a time series analysis statistical model to describe the law of image change over time, combining the weight adjustment mechanism of both space and time, and adaptively adjusting the weight according to different scenes and characteristics to achieve high quality and high precision in the image fusion process.

[0099] Therefore, this project combined the STARFM high temporal and spatial data fusion algorithm to obtain a high temporal and spatial resolution NDVI dataset with a spatial resolution of 30m and a temporal resolution of 16d from 2000 to 2023 ( Figure 4 , Figure 5 ).

[0100] 1.4 Construction of grassland vegetation coverage model and its accuracy test

[0101] 1.4.1 Preprocessing of UAV aerial photos

[0102] Based on the work point and waypoint route information uploaded to the server by the field observation tablet, the field observation records are added to the database using the DJILocator software independently developed by the team ( Figure 6 a) Select different flight modes (GRID, BELT) to locate the aerial photos. According to the latitude and longitude information recorded in the waypoints and the location information of the drone aerial photos, the drone aerial photos are matched with the waypoints one by one, and numbered from 1 to 16 according to the waypoint sequence. At the same time, each photo entering the database is assigned a unique DJIPT ID number for later data call and maintenance ( Figure 6 b and 6c).

[0103] 1.4.2 Aerial Photo Coverage Extraction

[0104] The coverage of aerial photos was obtained using the vegetation patch analysis software Pixel Based Manual Classifier independently developed by the Institute of Fragile Ecological Environments team. The software calculates the greenness index based on the threshold method to achieve binary processing of vegetation and bare soil, thereby calculating the vegetation part as the grassland coverage value. The greenness index calculation formula is as follows:

[0105] EGI=2G-RB (1)

[0106] Where: EGI is the greenness index, R, G, and B represent the red, green, and blue bands of the drone image, respectively.

[0107] Calculate the greenness index of each pixel of the drone image in turn, set the initial threshold of the greenness index, and compare it with the greenness index value of the image. If the greenness index of the pixel is greater than the set threshold, the pixel is regarded as a vegetation pixel, otherwise it is regarded as a bare soil pixel. Compare the classified image with the original image for visual judgment. If the classification result is not ideal, repeat the above process until the vegetation area in the classified image is consistent with the vegetation area in the original image. The specific processing flow is as follows: Figure 7 and Figure 8 As shown:

[0108] The drone images are stored in JPEG format. The grayscale values ​​of the three bands R, G, and B are all in the range of 0-255, corresponding to the EGI index range of -500 to 500. To improve the processing efficiency, referring to the previous research results, the EGI index of the alpine vegetation on the Qinghai-Tibet Plateau is generally between 60 and 140. Therefore, the initial threshold is set within this range ( Figure 8 ).

[0109] 1.4.3 Remote sensing estimation model construction and accuracy test

[0110] Based on the Pixel Based Manual Classifier software, the coverage of all aerial photos was obtained, and the fused NDVI vegetation index value corresponding to each aerial photo was extracted in turn. With NDVI as the independent variable and the grassland coverage obtained from drone aerial photos as the dependent variable, the linear, exponential, logarithmic and power remote sensing estimation models of grassland coverage were constructed respectively. Combined with the leave-one-out cross validation method, the correlation coefficient and root mean square error between the validation set and the measured data were used to evaluate the optimal grassland biomass estimation model for alpine grassland. The leave-one-out cross validation (LOOCV) method selects one group of sample plot data from n groups of sample plot data as the validation data, and the remaining n-1 groups of sample plot data are fitted into an NDVI estimation model, and the selected validation data is used to evaluate the accuracy of the estimation model. This is repeated n times and the average value of the n evaluation indicators is taken as the final evaluation indicator. The predictive ability of the model is determined by the correlation coefficient R and RMSE (root mean square error, RMSE) between the model evaluation results and the observed values. R is often used to measure the accuracy of the model, and RMSE is often used to quantify the accuracy of the model. That is, the closer R is to 1, the lower the RMSE value is, and the higher the model accuracy is.

[0111] The calculation formula of R is:

[0112]

[0113] The calculation formula of RMSE is:

[0114]

[0115] In the formula, y i is the grassland biomass value measured in the field, is the average value of the measured grassland biomass, y′ i The values ​​of grassland biomass estimated by the model.

[0116] 1.5 Inversion of grassland coverage from 2000 to 2023

[0117] Based on the optimal grassland coverage remote sensing inversion model selected in 1.4.3, combined with the NDVI fusion dataset from 2000 to 2023, the grassland coverage data from 2000 to 2023 were inverted. The annual maximum grassland coverage from 2000 to 2023 was calculated using the MVC maximum synthesis algorithm. The average annual maximum coverage from 2000 to 2022 was calculated, and the spatial differentiation characteristics of the maximum coverage value in 2023 were analyzed.

[0118] 1.6 Analysis of grassland coverage changes before and after restoration

[0119] Based on the average annual maximum coverage from 2000 to 2022 and the maximum grassland coverage in 2023, the grassland coverage deviation in the study area in 2023 was calculated. Combined with the boundary data of the implementation range of different degradation restoration measures in the study area, the grassland coverage before and after different degradation restoration grassland treatments (the average annual maximum coverage from 2000 to 2022 and the maximum grassland coverage in 2023) was extracted to qualitatively and quantitatively evaluate the impact of degradation restoration measures on grassland coverage.

[0120] 2. Results and Analysis

[0121] 2.1 Statistical analysis of vegetation coverage

[0122] The statistical results of vegetation coverage analysis in the study area are shown in Table 1. There are large differences in vegetation coverage among the five sampling sites, with average vegetation coverage ranging from 35.08% to 58.51% and coefficient of variation (CV) ranging from 0.24 to 0.55. Among them, the maximum vegetation coverage is 92.09% and the minimum is 5.01%. The plot with the largest average vegetation coverage is Plot3-1, with an average vegetation coverage value and coefficient of variation of 58.51% and 0.30, respectively; the plot with the smallest average vegetation coverage is Plot3-2, with an average vegetation coverage value and coefficient of variation of 35.08% and 0.24, respectively. The average vegetation coverage of other areas is between 36.21% and 43.49%, with a coefficient of variation of 0.38 to 0.55. As for the study area, the average vegetation coverage and coefficient of variation are 42.90% and 0.40, respectively.

[0123] Table 1 Statistical analysis of vegetation coverage

[0124]

[0125] 2.2 Optimal grassland coverage estimation model

[0126] The accuracy of the model tested by the leave-one-out cross-validation method in 2.4.3 is shown in Table 2. The overall accuracy of the four models is not much different. The linear model has the smallest RMSEP of 12.02% and the largest R of 0.82. The exponential, logarithmic and power models are followed, with RMSEP ranging from 12.05% to 12.11 and R ranging from 0.79 to 0.81. The parameter estimates of the four models are as follows: Figure 6 As shown, the coefficients of determination of the linear, exponential, logarithmic and power models are 0.64, 0.59, 0.62 and 0.64 respectively. The optimal remote sensing estimation model calculation formula is y = 148.73x-12.029 ( Fig. 9 ).

[0127] Table 2 Accuracy test of grassland coverage remote sensing estimation model

[0128]

[0129] 2.3 Grassland coverage inversion

[0130] Based on the fused remote sensing dataset from 2000 to 2023 and the optimal grassland coverage remote sensing estimation model, the grassland coverage status of the experimental area from 2000 to 2023 was inverted, and the maximum grassland coverage of each year was calculated using the maximum synthesis method. The average annual maximum coverage from 2000 to 2022 and the spatial distribution of the maximum grassland coverage in 2023 are shown as follows: Fig.10 As shown. The average coverage value of the study area for many years ranges from 3.08% to 69.11, with an average coverage of 25.76%. Among them, the area with a coverage value of <10% accounts for 3.98% of the total area of ​​the study area, and the area with a coverage value between 10% and 20% accounts for 15.67% of the total area of ​​the study area; the grassland area with a coverage value between 20% and 30% accounts for the largest proportion, accounting for 55.58% of the total area of ​​the study area; the grassland area with a coverage value between 30% and 40% accounts for 20.41 of the total area of ​​the study area, and the area with a coverage greater than 40% only accounts for 4.36% of the total area of ​​the study area. In terms of the spatial heterogeneity of grassland coverage, the areas with higher grassland coverage over the past 23 years are mainly distributed in the southwest border area of ​​the study area, and the areas with lower coverage are mainly distributed around the middle runoff riverbed area of ​​the study area. The grassland coverage in other areas is relatively low, mostly less than 30%.

[0131] Compared with the average grassland coverage of the previous 23 years, the grassland coverage in 2023 has increased significantly. The area with grassland coverage greater than 40% exceeds half of the total area of ​​the study area, accounting for 50.51%, and the area with grassland coverage between 30% and 40% accounts for 30.10% of the total study area; while the area with grassland coverage less than 30% only accounts for 19.38% of the total study area. Among them, the area with grassland coverage greater than 40% is mainly distributed in plot 1 in the northwest of the study area; while the area with lower grassland coverage (less than 30%) is mainly distributed along the riverbed in the middle of plots 1 and 2.

[0132] 2.4 Changes in grassland coverage in the experimental sample area

[0133] The results of pixel-by-pixel statistics of different degraded and restored grassland coverage in the study area are as follows: Fig.11As shown, except for sample plot No. 2 and mixed fertilization sample plot No. 3, whose grassland coverage in 2023 was slightly higher than the multi-year average but with no significant difference, the grassland coverage of other sample plots in 2023 was significantly higher than the average grassland coverage of the previous 23 years. Among all the degradation restoration plots, the grassland cover changes in Plot 1-1 and Plot 1-2 were the largest, with average covers of 51.98% and 41.37%, respectively, an increase of 125.36% and 2688% over the average covers of the previous 23 years (23.07% and 18.23%); followed by Plot 2-1 and Plot 3-1, with average covers of 38.96% and 38.32%, respectively, an increase of 45.17% and 40.53% over the average of the grasslands in the previous 23 years (26.84% and 27.27%); while the average covers of Plot 2-2 and Plot 3-2 were 31.65% and 28.96%, respectively, an increase of 5.31% and 8.94% over the average of the previous 23 years (30.05% and 26.59%). Overall, the grassland coverage in the study area in 2023 increased significantly compared with the average of the previous 23 years, with the overall average coverage being 42.15%, an increase of 63.63% compared with the average coverage of the previous 23 years (25.77%).

[0134] 2.5 Spatial heterogeneity of grassland coverage changes

[0135] The calculation results of the deviation of grassland coverage in 2023 from the average grassland coverage of the previous 23 years are as follows: Fig.12 As shown. The results show that compared with the average grassland coverage of the previous 23 years, the grassland coverage increase ranges from 0% to 56.73%, and the average coverage of the study area has increased by 16.39%. Among them, the areas with grassland coverage increase greater than 20% are mainly distributed in Plot 1-1 and Plot 1-2 in the northwest of the study area, accounting for 32.08% of the total study area; the areas with an increase of 10% to 20% are mainly distributed in the middle and northeast of Plot 2-1 and the central area of ​​Plot3-1, accounting for 35.30% of the total study area; the areas with an increase of less than 10% are mainly distributed in Plot 2-2, Plot 1-2 and the southwest of Plot 3-2, and the western area of ​​Plot 3-1, accounting for 32.60% of the total study area.

[0136] An application embodiment of the present invention provides a computer device, which includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of a degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing.

[0137] An application embodiment of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of a degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing.

[0138] An application embodiment of the present invention provides an information data processing terminal, which includes a degraded grassland restoration evaluation system based on aerial photography and satellite remote sensing.

[0139] The technical solution of the present invention can be widely used in many fields:

[0140] 1. Grassland resource management: This technical solution can provide grassland resource management departments with more accurate and comprehensive grassland assessment data, helping them monitor key indicators such as grassland coverage and biomass changes, so as to formulate scientific management strategies and improve the rational utilization efficiency of grassland resources.

[0141] 2. Ecological and environmental protection: In terms of ecological and environmental monitoring, this technical solution can be used to monitor changes in grassland habitats, assess grassland biodiversity and ecological stability, help to promptly identify ecological and environmental problems and take corresponding protection and restoration measures to maintain ecological balance.

[0142] 3. Animal Husbandry Production: The agricultural sector can use this technical solution to monitor and evaluate grassland growth, assist farmers in rationally arranging grazing, fertilization and other tasks, improve the quality of livestock forage, optimize farmland management methods, and improve agricultural production efficiency.

[0143] Relevant evidence of the technical effects achieved by the embodiments of the present invention.

[0144] 1. Data accuracy and reliability: The evaluation system based on aerial photography and satellite remote sensing technology used in the embodiment can achieve refined monitoring and evaluation of grassland biomass and coverage, and the data accuracy and reliability are high.

[0145] 2. Improved efficiency: Compared with traditional ground observation methods, this technical solution greatly improves the evaluation efficiency. Through aerial photography and satellite remote sensing technology, rapid evaluation of large-scale grasslands can be achieved, saving time and cost.

[0146] 3. Technological gaps and breakthroughs: The implementation examples fill the technical gap in the field of grassland restoration evaluation, successfully combine the advantages of aerial photography and satellite remote sensing, solve the limitations of traditional methods, and demonstrate the potential for technological breakthroughs.

[0147] 4. Application prospects: The successful application of the application embodiment of the present invention indicates that it has broad application prospects in the field of grassland ecosystem monitoring and restoration, and provides new ideas and methods for scientific research and practice in related industries.

[0148] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0149] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for evaluating degraded grassland restoration based on aerial photography and satellite remote sensing, characterized in that: include: S1, ground survey / UAV aerial photography: Aerial observation of the degraded and restored grassland in the study area was carried out based on the UAV aerial photography analysis system to monitor the coverage information of grassland vegetation; S2, satellite remote sensing data acquisition and preprocessing: using cloud computing platform, obtain MOD13Q1 and Landsat NDVI datasets before restoration; S3, preprocessing of drone aerial photos: based on the work point and waypoint route information uploaded to the server by the field observation tablet, use DJI Locator software to add field observation records to the database, and select routes of different flight modes to locate aerial photos; according to the latitude and longitude information of the waypoint records and the location information of the drone aerial photos, the drone aerial photos are matched with the waypoints one by one, and numbered from 1 to 16 in sequence according to the waypoint sequence number, and each photo entering the database is assigned a unique DJIPT ID number for later data call and maintenance; S4, aerial photo coverage extraction: The coverage of aerial photos is obtained by using vegetation patch analysis software, and the greenness index is calculated based on the threshold method to achieve binary processing of vegetation and bare soil, so as to calculate the vegetation part as the grassland coverage value; S5, EGI threshold division of aerial photos: UAV images are stored in JPEG format, and the grayscale value range of the three bands R, G, and B is 0-255, corresponding to the EGI index range of -500 to 500. The EGI index of alpine vegetation is between 60 and 140, so the initial threshold is set within this range; S6, remote sensing estimation model construction and accuracy test: based on obtaining the coverage of all aerial photos, and extracting the fused NDVI vegetation index value corresponding to each aerial photo in turn; taking NDVI as the independent variable and the grassland coverage obtained from drone aerial photos as the dependent variable, linear, exponential, logarithmic and power remote sensing estimation models of grassland coverage were constructed respectively, combined with the leave-one-out cross-validation method, using the correlation coefficient and root mean square error evaluation between the validation set and the measured data, to screen the optimal grassland coverage remote sensing inversion; S7, grassland coverage inversion before restoration: Based on the selected optimal grassland coverage remote sensing inversion model, combined with the NDVI fusion data set of the year before restoration, the grassland coverage data before restoration was inverted, and the annual maximum grassland coverage before restoration was calculated using the MVC maximum synthesis algorithm. The average value of the maximum coverage in the year before restoration was calculated, and the spatial differentiation characteristics of the maximum coverage value in the year before restoration were analyzed. S8, analysis of grassland coverage changes before and after restoration: based on the average maximum coverage in the year before restoration and the maximum grassland coverage after restoration, the deviation of grassland coverage after restoration in the study area was calculated, and combined with the boundary data of the implementation range of different degradation restoration measures in the study area, the grassland coverage before and after the treatment of different degradation restoration grasslands was extracted, and the impact of degradation restoration measures on grassland coverage was qualitatively and quantitatively evaluated.

2. The degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing according to claim 1, characterized in that: The layout of the sample plot mainly follows the following rules: 1) Sample plots are set up according to the scope and terrain conditions of the study area; 2) Set up a 250m×250m GRID flight route, and the route setting evenly covers the entire study area; 3) The GRID flight route has a flight altitude of 20m, the photo resolution is better than 1cm, and the photo coverage area is 26m×35m; 4) 16 waypoints are evenly distributed within the route range, and an aerial photo is taken vertically downward at each waypoint to obtain data. The data obtained includes the grassland vegetation coverage information monitored by drone aerial photos.

3. The degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing according to claim 1, characterized in that: Satellite remote sensing data acquisition and preprocessing specifically include: The STARFM algorithm is used to fuse multiple remote sensing images with different spatial resolutions and temporal resolutions into an image with high spatial resolution and high temporal resolution. The relationship between low-resolution images and high-resolution images at different time points is analyzed, and this relationship is used to derive the spatial distribution and temporal variation information of the high-resolution image. A linear or nonlinear method is used to establish a spatial reflectance model to describe the relationship between images at different spatial resolutions. The time series analysis statistical model is then used to describe the law of image variation over time. The weight adjustment mechanism of both space and time is combined to adaptively adjust the weights according to different scenes and characteristics to achieve high quality and high precision in the image fusion process.

4. The degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing according to claim 1, characterized in that: In the extraction of aerial photo coverage, the greenness index calculation formula is as follows: EGI=2G-RB (1) Where: EGI is the greenness index, R, G, and B represent the red, green, and blue bands of the drone image, respectively; Calculate the greenness index of each pixel in the drone image in turn, set the initial threshold of the greenness index, and compare it with the greenness index value of the image. If the greenness index of the pixel is greater than the set threshold, the pixel is regarded as a vegetation pixel, otherwise it is regarded as a bare soil pixel; compare the classified image with the original image for visual judgment. If the classification result is not ideal, repeat the above process until the vegetation area in the classified image is consistent with the vegetation area in the original image.

5. The degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing according to claim 1, characterized in that: The construction and accuracy test of remote sensing estimation model specifically include: using the leave-one-out cross-validation method to select a group of sample plot data from n groups of sample plot data as verification data, fitting the remaining n-1 groups of sample plot data into an NDVI estimation model, and using the selected verification data to evaluate the accuracy of the estimation model, repeating this process n times and taking the average of the n evaluation indicators as the final evaluation indicator; the prediction ability of the model is determined by the correlation coefficient R and RMSE between the model evaluation results and the observed values, where R is often used to measure the accuracy of the model, and RMSE is often used to quantify the accuracy of the model, that is, the closer R is to 1, the lower the RMSE value is, and the higher the model accuracy is; The calculation formula of R is: The calculation formula of RMSE is: In the formula, y i is the value of grassland coverage measured on site, is the average value of the measured grassland coverage, y′ i is the grassland coverage value estimated by the model.

6. A degraded grassland restoration evaluation system based on aerial photography and satellite remote sensing using the degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing as claimed in any one of claims 1 to 5, characterized in that: include: Ground survey / UAV aerial photography module: Based on the UAV aerial photography analysis system, aerial observation of the degraded and restored grassland in the study area is carried out to monitor the coverage information of grassland vegetation; Satellite remote sensing data acquisition and preprocessing module: using GEE Google Engine cloud computing platform to obtain MOD13Q1 and Landsat NDVI datasets before restoration; UAV aerial photo preprocessing module: Based on the work point and waypoint route information uploaded to the server by the field observation tablet, add field observation records to the database, and select routes of different flight modes to locate aerial photos; according to the latitude and longitude information of the waypoint records and the location information of the drone aerial photo attributes, the drone aerial photos are matched with the waypoints one by one, and numbered from 1 to 16 in sequence according to the waypoint sequence number. At the same time, each photo entering the database is assigned a unique DJIPT ID number for later data call and maintenance; Aerial photo coverage extraction module: The coverage of aerial photos is obtained by using vegetation patch analysis software. The software calculates the greenness index based on the threshold method to achieve binary processing of vegetation and bare soil, thereby calculating the vegetation part as the grassland coverage value; Aerial photo EGI threshold division module: UAV images are stored in JPEG format. The grayscale value range of the three bands R, G, and B is 0-255, and the corresponding EGI index range is -500 to 500. The EGI index of alpine vegetation is between 60 and 140. Therefore, the initial threshold is set within this range; Remote sensing estimation model construction and accuracy test module: obtain the coverage of all aerial photos, and extract the fused NDVI vegetation index value corresponding to each aerial photo in turn; take NDVI as the independent variable and the grassland coverage obtained from drone aerial photos as the dependent variable, respectively construct grassland coverage linear, exponential, logarithmic and power remote sensing estimation models, combine the leave-one-out cross-validation method, use the correlation coefficient and root mean square error evaluation between the validation set and the measured data, and screen the optimal grassland coverage remote sensing inversion model; Grassland coverage inversion module before restoration: Based on the selected optimal grassland coverage remote sensing inversion model, combined with the NDVI fusion data set of the year before restoration, the grassland coverage data before restoration was inverted, and the annual maximum grassland coverage before restoration was calculated using the MVC maximum synthesis algorithm. The average value of the maximum coverage in the year before restoration was calculated, and the spatial differentiation characteristics of the maximum coverage value in the year before restoration were analyzed. Analysis module of grassland coverage changes before and after restoration: Based on the average maximum coverage in the year before restoration and the maximum grassland coverage in 2023, the grassland coverage deviation after restoration in the study area is calculated, and combined with the boundary data of the implementation range of different degradation restoration measures in the study area, the grassland coverage before and after the treatment of different degraded restoration grasslands is extracted, and the impact of degradation restoration measures on grassland coverage is qualitatively and quantitatively evaluated.

7. A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the degraded grassland restoration evaluation method based on aerial photography and satellite remote sensing as described in any one of claims 1 to 5.

9. An information data processing terminal, comprising the degraded grassland restoration evaluation system based on aerial photography and satellite remote sensing as claimed in claim 6.

Images