A method for extracting and restoring lake area based on remote sensing images

Abstract

The present invention belongs to the technical field of lake dynamic monitoring, and specifically discloses a method for extracting and restoring lake area based on remote sensing images. Compared with the conventional division method of natural lakes and reservoirs based on a single data set, the present invention combines multiple widely used reservoir and dam data sets, and supplements the information of newly built reservoirs, thereby achieving a more accurate and comprehensive division of natural lakes and reservoirs. Compared with the traditional water body classification and reconstruction method based on water occurrence frequency threshold, the present invention further considers the difference between water body occurrence frequency and actual water body category, has higher classification and reconstruction effect and accuracy, and can still accurately restore the missing area of ​​the lake when the lake area is missing, and displays the spatial effect of the lake after the area is restored. The method of the present invention is of great significance for separately monitoring the area of ​​natural lakes and reservoirs, and studying their dynamics and spatial changes.

Description

Technical Field

[0001] The invention belongs to the technical field of lake dynamic monitoring, and relates to a lake area extraction and restoration method based on remote sensing images. Background Art

[0002] Lakes include natural lakes and artificial reservoirs. Natural lakes and artificial reservoirs (hereinafter referred to as reservoirs) play the role of regulators, integrators and sentinels in climate change, forming a widely distributed ecosystem network that can provide valuable information on the patterns and mechanisms of how terrestrial and aquatic ecosystems respond to climate change.

[0003] Natural lakes and reservoirs differ in several characteristics, including four main types of physical and chemical properties: (1) Catchment characteristics: The catchment area and surface area ratio of natural lakes and reservoirs are quite different. In terms of morphology, natural lakes are more rounded, while reservoirs are longer, narrower, and more complex in shape. (2) Water body characteristics: Reservoirs are generally shallower than lakes and have higher bottom water temperatures. (3) Systematic management: Reservoirs are more systematically managed than natural lakes. (4) Geographic distribution: Natural lakes and reservoirs also differ greatly in their geographic distribution. These differences may affect the response of ecosystems to climate change. Therefore, it is important to monitor the area of ​​natural lakes and reservoirs separately and study their dynamic changes.

[0004] Optical remote sensing data has become the main data source for automatic dynamic monitoring of surface water in large areas due to its easy acquisition, simple processing and high water extraction accuracy. Landsat series images, with their more than 40-year operating cycle and relatively high spatial resolution (30 meters for Landsat TM / ETM+ / OLI), can reflect the long-term changes of lakes and capture small-scale lakes, making them one of the most widely used datasets and forming lake datasets such as GLAKERS.

[0005] However, during the satellite observation process, about one-third of the pixels in the GSW dataset were marked as "no data" (masked pixels) due to the influence of clouds, cloud shadows, terrain shadows, and the failure of the Scan Line Corrector (SLC) of Landsat ETM+. If these masked pixels are not taken into account, the estimation of lake area will be seriously underestimated, which greatly limits the monitoring of long-term changes in lake range. In view of this situation, it is a better choice to reconstruct the masked pixels to restore the lake area.

[0006] In the existing technology, the division of natural lakes and reservoirs is usually based on a single data set, and there is a lack of updates on newly built reservoirs, resulting in insufficient accuracy and comprehensiveness in the division of the two, which not only affects the estimation of natural lakes and reservoirs, but may also lead to misjudgment of their dynamic changes and long-term trends. The method based on the water body frequency threshold is the most widely used water body classification and reconstruction method. In most of the existing methods based on the water body frequency threshold, a threshold is set and the data-free pixels whose water body frequency value exceeds the threshold are corrected to water pixels. However, in reality, there is also a certain proportion of non-water pixels within the range exceeding the water body frequency threshold, which leads to deviations in the estimated lake area. Summary of the invention

[0007] The purpose of the present invention is to propose a lake area extraction and restoration method based on remote sensing images, so as to achieve accurate and comprehensive division of natural lakes and reservoirs, and at the same time, to accurately restore the missing area when the lake area is missing.

[0008] In order to achieve the above object, the present invention adopts the following technical scheme:

[0009] A lake area extraction and restoration method based on remote sensing images includes the following steps:

[0010] Step 1. First, divide the natural lakes and artificial reservoirs and add the information of newly built reservoirs;

[0011] Step 2. Create an annual water classification composite map;

[0012] Step 3. Use the annual water classification composite map and water body frequency map to restore the lake area;

[0013] Step 4. Create a water body occurrence frequency summary map as a base map for displaying spatial effects, find the water body occurrence frequency threshold that best matches the number of water pixels in the lake after area restoration, and generate a lake water body spatial distribution map.

[0014] The present invention has the following advantages:

[0015] As described above, the present invention relates to a method for extracting and restoring lake area based on remote sensing images. First, natural lakes and artificial reservoirs are divided and the information of newly built reservoirs is supplemented. Then, an annual water classification composite map is made. The lake area is restored using the annual water classification composite map and the water body frequency map. Then, a summary map of water body frequency for many years is made as a base map for spatial effect display. Finally, a frequency threshold matching it is found to generate a spatial map of lake water after area restoration. Compared with the conventional division method of natural lakes and reservoirs based on a single data set, the present invention combines multiple widely used reservoir and dam data sets and supplements the information of newly built reservoirs, thereby achieving a more accurate and comprehensive division of natural lakes and reservoirs. In addition, compared with the traditional water body classification and reconstruction method based on water frequency threshold, the present invention further considers the difference between water body frequency and actual water body category, has higher classification and reconstruction effect and accuracy, and can accurately restore the missing area of ​​the lake when the lake area is missing. In addition, the present invention also displays the spatial effect of the lake after area restoration. The present invention is of great significance for monitoring the area of ​​natural lakes and reservoirs separately and studying their dynamic and spatial changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Flow chart of a method for extracting and restoring lake area based on remote sensing images in an embodiment of the present invention;

[0017] Figure 2 It is a schematic diagram of the time series of the surface water area finally estimated for the lake example in an embodiment of the present invention; Figure 2 (a), (b), (c), and (d) are the estimated surface water area time series for Reservoir A, Reservoir B, Lake C, and Lake D, respectively;

[0018] Figure 3 This is a visualization of the spatial map of the restored example lake area using the GEE platform; Figure 3 (a), (b), (c), and (d) are spatial map visualizations of Reservoir A, Reservoir B, Lake C, and Lake D after the area restoration. DETAILED DESCRIPTION

[0019] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0020] In order to solve the problem that a large amount of pixel data is missing in remote sensing images, resulting in a serious underestimation of lake area estimation, thereby limiting the long-term change monitoring of lake extent, this embodiment proposes a method for restoring the area of ​​natural lakes and artificial reservoirs so that the missing area of ​​lakes can be accurately restored even when the lake area is missing.

[0021] This embodiment takes Chinese lakes as an example, based on the vector boundaries of Chinese lakes in the GLAKERS dataset, combined with a variety of reservoir and dam datasets around the world and in China, to accurately and comprehensively classify natural lakes and reservoirs through sophisticated spatial analysis and image interpretation. The JRC monthly water history dataset is used to produce an annual water classification composite map, and each pixel is classified into one of three categories: water, non-water, or no data. In addition, the present invention also proposes a method based on water occurrence thresholds to restore the area of ​​each natural lake and reservoir on the annual water classification composite map and the JRC water body occurrence frequency map. Finally, a water body occurrence frequency summary map is produced, and a frequency threshold is determined by calculating the minimum difference between the total number of lake water pixels after restoration and the total number of pixels in the frequency value interval of each water body occurrence, which is used to display the spatial effect of the restored lake.

[0022] like Figure 1 As shown, the lake area extraction and restoration method based on remote sensing images in this embodiment includes the following steps:

[0023] Step 1. First, divide the natural lakes and artificial reservoirs and add the information of newly built reservoirs.

[0024] This step 1 includes three sub-steps, namely, a preliminary screening and classification sub-step of lakes, a further classification sub-step based on dam data, and a sub-step of supplementing information on newly built reservoirs. These three sub-steps are described in detail below.

[0025] Step 1.1. Perform preliminary screening and classification of lakes based on lake and reservoir datasets.

[0026] From the GLAKERS lake dataset, lakes with an area of ​​more than 1 km in China were selected. 2 Lake vector border.

[0027] Obtain Chinese reservoir datasets with high data quality reliability, such as China-LDRL and GeoDAR datasets, and select those that meet the requirements within China, i.e., those with an area of ​​more than 1km2. 2 Reservoir polygon.

[0028] The reservoir polygons in the reservoir dataset with higher data quality reliability are spatially intersected with the lake polygons in the GLAKERS lake dataset to conduct preliminary screening and classification of the lakes.

[0029] GLAKERS lake polygons that intersect with any reservoir polygon in the reservoirs dataset are directly classified as reservoirs.

[0030] Step 1.2. Further classify lakes based on the dam dataset.

[0031] Since the basis for distinguishing natural lakes and reservoirs is the corresponding dams built on the reservoirs, the proportion of missing dams in the dataset is reduced by associating and referencing multiple widely used dam datasets; the associated multiple dam datasets are then spatially intersected with the lake polygons. In this process, visual interpretation is combined to improve the precision and accuracy of classification.

[0032] In this example, five dam datasets including GeoDAR, GooDD, GRanD, OSM, and China-LDRL were selected.

[0033] A 1 km buffer analysis was performed on the dam vector points in the five dam datasets, that is, a buffer with a radius of 1 km was generated around each dam point and a spatial intersection operation was performed with the remaining lake polygons in the GLAKERS lake dataset.

[0034] For lake polygons in the intersection area, further classification is performed based on the size of the area. 2 The polygons are combined with high-resolution imagery (e.g. Google Earth) for visual interpretation to confirm whether they are reservoirs.

[0035] For areas less than or equal to 10km 2 The polygon is directly classified as a reservoir because of its small area and the construction of a dam.

[0036] This step improves classification accuracy by combining spatial and image analysis, especially in the identification of larger reservoirs.

[0037] Step 1.3. Supplement the newly built reservoir information based on the newly built reservoir dataset to provide complete reservoir information for the division of lakes.

[0038] In order to make up for the deficiency of the existing reservoir dataset in the information of newly built reservoirs after 2000, this embodiment introduces the GREI-p2k newly built reservoir dataset as a supplement, which records the detailed information of newly built reservoirs in the world after 2000.

[0039] Remove the polygons that intersect the reservoirs identified in steps 1.1 and 1.2 from the newly created reservoir dataset, merge the unidentified new reservoirs in the newly created reservoir dataset with the results of steps 1.1 and 1.2 to further improve the reservoir information.

[0040] This step obtains newly built reservoirs around the world after 2000 by introducing the GREI-p2k newly built reservoir dataset, and selects newly built reservoirs in China that meet the requirements, providing more complete reservoir information for the division of natural lakes and artificial reservoirs.

[0041] Step 2. Produce an annual water classification composite map. Producing an annual national water classification composite map is a crucial step. The process of lake area restoration requires the restoration of missing pixels on the annual water classification composite map.

[0042] Specifically, the monthly water history maps of JRC within each year were aggregated together to form an annual layer, and then the number of times a certain pixel was identified as a water pixel within the annual layer was divided by the total number of valid observations to obtain the water body frequency.

[0043] Water frequency The calculation formula is as follows:

[0044] .

[0045] In the formula, is the number of times the pixel is identified as water; is the total number of valid observations in the annual layer.

[0046] Water pixels, non-water pixels and no-data pixels are determined according to the water body frequency. Specifically, if the water body frequency exceeds the frequency preset value, the pixel is classified as a water pixel, otherwise the pixel is classified as a non-water pixel.

[0047] During the pixel classification process, each pixel should have at least one valid observation of water or non-water body in the monthly layer image of the rainy month and the dry month of the corresponding year, otherwise the pixel will be classified as no data in the annual layer.

[0048] Typically, the rainy season is from May to October, and the dry season is from November to December and from January to April.

[0049] Finally, the annual water classification composite map is obtained. The values ​​of the annual water classification composite map include three types: water (pixel value is 2), non-water (pixel value is 1), and no data (pixel value is 0).

[0050] Step 3. Use the annual water classification composite map and water body frequency map to restore the lake area.

[0051] In this step, the annual water classification composite map and the JR water body frequency map are used to perform long-term restoration of the long-term surface water area of ​​the lake. The long-term water body detection information of all pixels in the same lake is used to restore the water pixels by calculating the distribution ratio of water pixels and non-water pixels for each water body frequency value (from the JRC water body frequency map).

[0052] The recovery process is as follows:

[0053] Step 3.1. Calculate the water pixel proportion WP, non-water pixel proportion NWP and no-data pixel proportion NDP of each water body occurrence frequency value by dividing the number of water pixels WC, the number of non-water pixels NWC and the number of no-data pixels NDC by the total number of pixels TC.

[0054] The proportion of water body pixels for each water body occurrence frequency value , the proportion of non-water pixels , the percentage of pixels without data , can be calculated using the following formula:

[0055] .

[0056] In the formula, m, n, and t represent the index of lake, frequency value, and year, respectively. The superscript raw indicates the result that has not been processed by data recovery. The values ​​are WP, NWP and NDP. The possible values ​​are WC, NWC and NDC.

[0057] In this embodiment, no area restoration is performed on lakes with NDP exceeding 90%.

[0058] Step 3.2. Set the water pixel ratio threshold WP as 5% of the effective data pixel ratio (1-NDP). T .

[0059] The water pixel ratio threshold WP T As the water body occurrence frequency threshold WO T The basis for judgment is that if WP exceeds WP for the first time T , then the WP T The corresponding water body occurrence frequency value WO is used as the water body occurrence frequency threshold WO T .

[0060] Water pixel ratio threshold The calculation formula is as follows:

[0061] .

[0062] Step 3.3. In each water body occurrence frequency region that meets the conditions, when WO ≥ WO T When the water pixel ratio is restored.

[0063] Specifically, the percentage of water pixels after restoration is equal to the percentage of water pixels before restoration plus the percentage of no-data pixels allocated to water pixels according to the distribution ratio. The percentage of non-water pixels after restoration is equal to the percentage of non-water pixels before restoration plus the percentage of no-data pixels allocated to non-water pixels according to the ratio other than the distribution ratio.

[0064] The distribution ratio is determined by the proportion of water pixels and the proportion of non-water pixels. The distribution ratio is obtained by dividing the proportion of water pixels by the proportion of valid data pixels. The calculation formula is as follows:

[0065] .

[0066] Define the following coefficients , then the proportion of water pixels after restoration , the proportion of non-water pixels , the percentage of pixels without data The calculation formulas are as follows:

[0067] ;

[0068] ;

[0069] ;

[0070] .

[0071] In the formula, Represents the distribution ratio, and the superscript rec represents the result after recovery processing. Indicates the frequency of water body occurrence, parameter Represents a numerical symbol, in each water body occurrence frequency area that meets the conditions, that is, WO ≥ WO T ,parameter Assign a value of 1 and substitute it into the formula for calculation, otherwise the parameter Assign a value of 0 and substitute it into the formula for calculation.

[0072] In the water body frequency area that does not meet the conditions, that is, when WO<WO T No recovery will be performed.

[0073] Step 3.4. Calculate the area of ​​the lake after restoration by multiplying the total number of restored water pixels by the area of ​​the unit pixel (e.g. 30m × 30m). The final estimated surface water area time series of the lake is shown in the attached figure. Figure 2 shown.

[0074] Figure 2 (a), (b), (c), and (d) represent the estimated surface water area time series for Reservoir A, Reservoir B, Lake C, and Lake D, respectively. Reservoir A and Reservoir B are artificial reservoirs, while Lake C and Lake D are natural lakes.

[0075] Lake area after restoration The calculation formula is as follows:

[0076] ;

[0077] in, It represents the total number of pixels. The number 100 above the summation sign represents the maximum frequency of water bodies. The number 900 in the calculation formula represents the area of ​​the unit pixel.

[0078] Step 4. Create a water body occurrence frequency summary map as a base map for displaying spatial effects, find the water body occurrence frequency threshold that best matches the number of water pixels in the lake after area restoration, and generate a lake water body spatial distribution map.

[0079] In this embodiment, all images in the JRC monthly water history data for a total of 38 years from 1984 to 2021 are aggregated together, the number of times each pixel is identified as a water pixel in all images is calculated, and the number is divided by the total valid observation number to generate a 38-year water body occurrence frequency summary map, which is used to generate a base map of the lake spatial information after area restoration.

[0080] Using the Google Earth Engine (GEE) platform, the number of water pixels in the lake after area restoration was further matched with the total number of pixels in each frequency value range in the water body frequency summary map. That is, the minimum difference between the total number of water pixels after lake restoration and the total number of pixels in each frequency value range in the corresponding water body frequency summary map was calculated, and the corresponding frequency value was determined. This frequency value was set as the frequency threshold of the water body frequency summary map, and the area exceeding this threshold was classified as water body.

[0081] The formula for finding the minimum difference is:

[0082] .

[0083] In the formula, The difference k in the number of pixels represents the corresponding frequency value, Represents the number of water pixels in lake i after area restoration, Represents the number of pixels within the corresponding frequency range in the water body occurrence frequency summary map.

[0084] Finally, the GEE platform was used to visualize the spatial map of the lake area after restoration. Figure 3 (a), (b), (c), and (d) are spatial map visualizations of Reservoir A, Reservoir B, Lake C, and Lake D after the area restoration.

[0085] Compared with traditional methods, the present invention achieves a more accurate and comprehensive division of natural lakes and reservoirs, and has a higher classification reconstruction effect and accuracy. In the case of missing lake area, the missing area of ​​the lake can still be accurately restored. The method of the present invention is of great significance for separately monitoring the area of ​​natural lakes and reservoirs and studying their dynamic and spatial changes.

[0086] Of course, the above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiments. It should be noted that all equivalent substitutions and obvious deformation forms made by any technician familiar with the field under the guidance of this specification fall within the essential scope of this specification and should be protected by the present invention.

Claims

1. A method for extracting and restoring lake area based on remote sensing images, characterized in that: The steps include: Step 1. First, divide the natural lakes and artificial reservoirs and add the information of newly built reservoirs; Step 2. Create an annual water classification composite map; The monthly water history maps within each year were aggregated together to form an annual layer, and then the number of times a certain pixel was identified as a water pixel in the annual layer was divided by the total number of valid observations to obtain the water body frequency; If the water body frequency exceeds the preset frequency value, the pixel is classified as a water pixel, otherwise it is classified as a non-water pixel; In the pixel classification process, each pixel should have at least one valid observation of water or non-water in the monthly layer image of the rainy season and the dry season of the corresponding year, otherwise the pixel will be classified as no data in the annual layer; Step 3. Use the annual water classification composite map and water body frequency map to restore the lake area; The step 3 is specifically as follows: Step 3.

1. Calculate the water pixel proportion WP, non-water pixel proportion NWP and no-data pixel proportion NDP of each water body occurrence frequency value by dividing the number of water pixels WC, the number of non-water pixels NWC and the number of no-data pixels NDC by the total number of pixels TC; lakes with NDP exceeding 90% will not be restored; The proportion of water body pixels for each water body occurrence frequency value The proportion of non-water pixels The percentage of pixels without data The calculation formula is as follows: In the formula, m, n, and t represent the index of lake, frequency value, and year, respectively. The superscript raw indicates the result that has not been processed by data recovery. The values ​​of P are WP, NWP, and NDP, and the values ​​of C are WC, NWC, and NDC. Step 3.

2. Set 5% of the effective data pixel ratio as the water pixel ratio threshold WP T ; The water pixel ratio threshold WP T As the water body occurrence frequency threshold WO T The basis for judgment is that if WP exceeds WP for the first time T , then the WP T The corresponding water body occurrence frequency value WO is used as the water body occurrence frequency threshold WO T ; Water pixel ratio threshold The calculation formula is as follows: Step 3.

3. In each water body occurrence frequency area that meets the conditions, when WO ≥ WO T When the water pixel ratio is restored, the restored water pixel ratio is equal to the water pixel ratio before restoration plus the no-data pixel ratio allocated to water pixel ratio according to the distribution ratio, and the restored non-water pixel ratio is equal to the non-water pixel ratio before restoration plus the no-data pixel ratio allocated to non-water pixel ratio according to the distribution ratio. The distribution ratio is determined by the proportion of water body pixels and the proportion of non-water body pixels. The distribution ratio is obtained by dividing the proportion of water body pixels by the proportion of valid data pixels. In the water body frequency area that does not meet the conditions, that is, when WO<WO T No restoration will be carried out when Distribution ratio DR m,n,t The calculation formula is as follows: Define the coefficient θ as follows m,n,t , then the proportion of water pixels after restoration The proportion of non-water pixels The percentage of pixels without data The calculation formulas are as follows: In the formula, DR represents the distribution ratio, the superscript rec represents the result after restoration, WO represents the frequency of water body occurrence, and the parameter θ represents a numerical symbol. In each water body occurrence frequency area that meets the conditions, WO ≥ WO T , the parameter θ is assigned a value of 1 and substituted into the formula for calculation, otherwise the parameter θ is assigned a value of 0 and substituted into the formula for calculation; Step 3.

4. Multiply the total number of restored water pixels by the area of ​​the unit pixel to calculate the restored lake area; Lake area after restoration The calculation formula is as follows: Among them, TC m,n,t It represents the total number of pixels. The number 100 above the summation symbol represents the maximum value of the frequency of water body occurrence. The number 900 that appears in the calculation formula represents the area of ​​the unit pixel. Step 4. Create a water body occurrence frequency summary map as a base map for displaying spatial effects, find the water body occurrence frequency threshold that best matches the number of water pixels in the lake after area restoration, and generate a lake water body spatial distribution map.

2. The method for extracting and restoring lake area based on remote sensing images according to claim 1 is characterized in that: The step 1 is specifically as follows: Step 1.

1. Preliminary screening and classification of lakes based on lake and reservoir datasets; Step 1.

2. Further classify lakes based on the dam dataset; Step 1.

3. Supplement the newly built reservoir information based on the newly built reservoir dataset to provide complete reservoir information for the division of lakes.

3. The method for extracting and restoring lake area based on remote sensing images according to claim 2 is characterized in that: The step 1.1 is specifically as follows: Filter lakes with an area larger than 1km from the lake dataset 2 The lake vector boundary; Further filter out the reservoir data set that meets the requirements, that is, the area is larger than 1km 2 The reservoir polygons are spatially intersected with the lake polygons in the lake dataset to perform preliminary screening and classification of lakes.

4. The method for extracting and restoring lake area based on remote sensing images according to claim 2 is characterized in that: The step 1.2 is specifically as follows: A 1 km buffer analysis was performed on the dam vector points in the multiple dam datasets, that is, a buffer with a radius of 1 km was generated around each dam point, and a spatial intersection operation was performed with the remaining lake polygons in the lake dataset; For the lake polygons in the intersecting area, further classification is performed based on the size of the area; For areas larger than 10km 2 The polygons are combined with high-resolution images for visual interpretation to confirm whether they are reservoirs; for areas less than or equal to 10km 2 The polygon is directly classified as a reservoir because of its small area and the construction of a dam.

5. The method for extracting and restoring lake area based on remote sensing images according to claim 2 is characterized in that: The step 1.3 is specifically as follows: Remove the polygons that intersect the reservoirs identified in steps 1.1 and 1.2 from the newly created reservoir dataset, merge the unidentified new reservoirs in the newly created reservoir dataset with the results of steps 1.1 and 1.2 to further improve the reservoir information.

6. The method for extracting and restoring lake area based on remote sensing images according to claim 1, characterized in that: In step 2, the values ​​in the obtained annual water classification composite map include three types: 2, 1, and 0, which respectively represent water, non-water, and no data.

7. The method for extracting and restoring lake area based on remote sensing images according to claim 1 is characterized in that: In step 4, the process of making a water body occurrence frequency summary map is as follows: All images in the monthly water history data of consecutive years are aggregated together, the number of times each pixel is identified as a water pixel in all images is calculated, and the number is divided by the total number of valid observations to generate a summary map of the frequency of water body occurrence.

8. The method for extracting and restoring lake area based on remote sensing images according to claim 1, characterized in that: In step 4, the minimum difference between the total number of water pixels after the lake is restored and the total number of pixels in each frequency value range of the corresponding water body occurrence frequency summary map is calculated using the Google Earth Engine platform, and the corresponding frequency value is determined. The frequency value is set as the frequency threshold of the water body occurrence frequency summary map, and the area exceeding the frequency threshold is classified as a water body; The formula for finding the minimum difference is: ΔN=|C i -S i,k |; In the formula, ΔN is the difference in the number of pixels, k represents the corresponding frequency value, C i Represents the number of water pixels in lake i after area restoration, S i,k Represents the number of pixels in the corresponding frequency range in the summary map of water body occurrence frequency; Finally, the Google Earth Engine platform is used to visualize the spatial map of the restored lake area.

Images