A lake data processing system based on remote sensing images

Abstract

The application provides a lake data processing system based on remote sensing images, and relates to the technical field of lake management. The method comprises the following steps: using a YOLO-R model to capture and process images and DEM elevation thresholds of a lake area in real time; transmitting the processed images to a data processing layer; processing the images to obtain edge inference results; obtaining a lake boundary map, a boundary change map and a reclamation expansion rate through the DEM elevation thresholds and the edge inference results; and producing a lake evolution animation through the lake boundary map, the boundary change map and the reclamation expansion rate to complete water body boundary processing of the lake remote sensing images. The technical scheme of the application realizes a high-precision lake boundary extraction function.

Description

Technical Field

[0001] This application relates to the field of lake management technology, and in particular to a lake data processing system based on remote sensing images. Background Technology

[0002] With the intensification of global climate change and human activities, lakes, as important freshwater resources and ecosystems, have received widespread attention for their water quantity changes, water quality, and ecological health. Remote sensing technology, with its large-scale, periodic observation capabilities, can acquire real-time information on lake area, water level, water optical characteristics, and surrounding environment through multispectral, hyperspectral, and radar data sources, providing crucial data support for lake resource management, ecological protection, and disaster early warning.

[0003] Traditional remote sensing image processing systems use the NDWI thresholding method to extract water body boundaries. When a lake is partially covered by thin clouds, and the cloud shadows are projected onto the bay area, shallow water areas are easily misidentified as land, or narrow bays are missed. Therefore, a lake data processing system based on remote sensing images is proposed. Summary of the Invention

[0004] The purpose of this invention is to address the problem of low accuracy in lake boundary extraction in existing lake remote sensing processing methods, and to provide a lake data processing system based on remote sensing images.

[0005] The above-mentioned objective of this application is achieved through the following technical solution:

[0006] Data acquisition layer: Using the YOLO-R model, images of the lake area and DEM elevation thresholds are captured and processed in real time;

[0007] Data transmission layer: Transmits the processed image to the data processing layer;

[0008] Data processing layer: processes the images to obtain edge inference results;

[0009] Data analysis layer: By using DEM elevation thresholds and edge inference results, lake boundary maps, boundary change maps, and reclamation expansion rates are obtained;

[0010] Data application layer: By using lake boundary maps, boundary change maps, and reclamation expansion rates, lake evolution animations are created to complete the water body boundary processing of lake remote sensing images.

[0011] Optionally, the data acquisition layer utilizes the YOLO-R model to capture images and DEM elevation thresholds of the lake area in real time, including the following steps:

[0012] Real-time imagery and DEM elevation thresholds of the lake area are acquired using multi-source satellites.

[0013] The RPC model is used to orthogonally correct the images and unify them to the WGS84 coordinate system.

[0014] Optionally, the transmission steps of the data transmission layer include: compressing the image using the JPEG2000 compression algorithm; and transmitting the compressed image data stream to the data processing layer via a 5G network or fiber optic leased line.

[0015] Optionally, the data processing layer is used to extract dynamic texture features by batch processing images;

[0016] The patch shape index includes aspect ratio and compactness;

[0017] The data processing layer is also used to fuse patch shape indices and texture trajectories of dynamic texture features through a graph convolutional network (GCN) to perform deep coupling of multi-source features and obtain edge inference results. Specific steps include:

[0018] Construct a dynamic texture feature library as follows:

[0019] Extract the gray-level co-occurrence matrix of candidate patches in the image to obtain dynamic texture features;

[0020] Dynamic texture features include: GLCM contrast, entropy value, and angular second moment;

[0021] Dynamic texture features stored quarterly to construct texture trajectories ;

[0022] Constructing a shape-texture coupling model, which is the edge inference result, involves the following steps:

[0023] Each candidate patch is treated as a node, and its feature vector is represented as follows:

[0024]

[0025] in Indicates the first Feature vectors of each node; Aspect ratio, For firmness, The slope of the quarterly contrast change; This represents the slope of the quarterly entropy change. This represents the slope of the second moment of the quarter angle.

[0026] Connect spatially adjacent candidate patches;

[0027] Using texture similarity as edge weights, denoted as...

[0028]

[0029] in Indicates the first The node and the first The weights of the edges between nodes; Represents the texture feature of the i-th node; Represents the texture feature of the j-th node; Indicates the standard deviation;

[0030] The graph convolutional layers of the graph convolutional network use the Chebyshev multinomial approximation, and the propagation rule is expressed as follows:

[0031]

[0032] in This represents the output of the l-th graph convolutional layer; This represents the order of the Chebyshev polynomial; These are learnable parameters; For Chebyshev polynomials; For the normalized Laplace matrix, This represents the weight matrix of the l-th layer;

[0033] The loss function of a graph convolutional network is to jointly optimize texture trajectory and shape regularity, expressed as:

[0034]

[0035] in, Indicates the weighting parameter; The mean square error of the texture trajectory; Mean square error representing shape regularity; This represents the regularization term.

[0036] Optional, uncertainty is addressed by using graph convolutional edge inference results to analyze cloud cover uncertainty, combining DEM elevation thresholds, and refining reclamation / sand mining boundaries using graph convolutional networks to distinguish between natural changes and reclamation, thereby generating lake boundary maps.

[0037] Optionally, the data analysis layer is also used to extract the lake extent, compare the lake boundaries during the high-water season and the low-water season, identify changes in lake area, and generate a boundary change map.

[0038] Optionally, the data analysis layer is also used to identify abrupt change points in sand mining areas based on a time-series texture feature library using the isolated forest algorithm, and to predict the rate of land reclamation expansion by combining a linear regression and LSTM hybrid model.

[0039] Optionally, the data application layer uses the TimeManager tool to create lake evolution animations by combining lake boundary maps, boundary change maps, and reclamation expansion rates.

[0040] An electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to enable the electronic device to perform a lake data processing system based on remote sensing images.

[0041] A computer-readable storage medium storing instructions that, when executed, perform a lake data processing system based on remotely sensed images.

[0042] The beneficial effects of the technical solution provided in this application are:

[0043] Combining DEM elevation thresholds and edge inference results with historical cloud cover probability distribution, multiple hypothetical water body boundaries are generated. By statistically analyzing high-frequency areas, the final lake boundary location is optimized. In areas with severe cloud cover during the dry season, traditional thresholding methods result in jagged boundaries due to misjudgment of mixed pixels. Monte Carlo simulation, through probabilistic fusion, outputs smooth water body contours that conform to terrain logic. Simultaneously, the automated Monte Carlo process reduces manual workload. In the time-series texture feature library, isolated forests automatically learn normal fluctuation patterns, identify abrupt change points, and, combined with LSTM-predicted normal texture trajectories, locate the spatiotemporal coordinates of abrupt changes to generate a probabilistic heatmap. Attached Figure Description

[0044] The present application will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0045] Figure 1 This is a system structure diagram in an embodiment of this application;

[0046] Figure 2 This is a schematic diagram of the electronic device structure in the embodiments of this application. Detailed Implementation

[0047] To provide a clearer understanding of the technical features, objectives, and effects of this application, the specific embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0048] Embodiments of this application provide a lake data processing system based on remote sensing images.

[0049] Please refer to Figure 1 , Figure 1 This is a system architecture diagram of a lake data processing system based on remote sensing images, as described in an embodiment of this application. The system includes:

[0050] Data acquisition layer: Using the YOLO-R model, images of the lake area and DEM elevation thresholds are captured and processed in real time;

[0051] In one specific embodiment of this application, Landsat and high-resolution satellite images of the target lake are collected, covering both the high-water and low-water seasons. Combined with DEM and land use data, and utilizing a cropped version of the YOLO-R model deployed by ground receiving stations, commercial APIs, and edge terminals, images and auxiliary data of the lake area are captured in real time.

[0052] Data transmission layer: Transmits the processed image to the data processing layer;

[0053] In one specific embodiment of this application, the data transmission layer compresses satellite images via an edge terminal and transmits them to the cloud via 5G / fiber optics, using Kafka or MQTT for communication between the edge and the cloud;

[0054] Data processing layer: processes the images to obtain edge inference results;

[0055] In one specific embodiment of this application, the data processing layer uses a lightweight inference model to perform real-time target detection at the edge, and extracts dynamic texture features through batch processing in the cloud, and uses a graph convolutional network to fuse patch shape index and texture trajectory to perform deep coupling of multi-source features.

[0056] Data analysis layer: By using DEM elevation thresholds and edge inference results, lake boundary maps, boundary change maps, and reclamation expansion rates are obtained;

[0057] In one specific embodiment of this application, by combining DEM elevation thresholds and edge inference results, the uncertainty of cloud cover is analyzed through Monte Carlo simulation to optimize the accuracy of lake water body boundaries. By combining land use data or spectral features, the GCN model is used to refine the reclamation / sand mining boundaries, distinguish between natural changes and reclamation, generate lake boundary maps, extract lake ranges, compare lake boundaries during the wet and dry seasons, identify area changes, and identify abrupt changes in sand mining areas (contrast > 50%, sharp drop in second moment of angle) based on a time series texture feature library. By combining a linear regression and LSTM hybrid model, the rate of reclamation expansion is predicted, revealing the long-term evolution law and short-term abnormal disturbances of lake ecology.

[0058] Data application layer: By using lake boundary maps, boundary change maps, and reclamation expansion rates, lake evolution animations are created to complete the water body boundary processing of lake remote sensing images.

[0059] In one specific embodiment of this application, a user interface is provided to create a lake evolution animation based on data from the data analysis layer using TimeManager, overlay a quarterly GLCM index change curve, highlight abnormal areas, display the spatial distribution of reclamation and sand mining probabilities, push abrupt events (coordinates, area, confidence level) in real time, and generate a PDF / HTML report.

[0060] The data acquisition layer utilizes the YOLO-R model to capture images and DEM elevation thresholds of the lake area in real time, including the following steps:

[0061] Real-time imagery and DEM elevation thresholds of the lake area are acquired using multi-source satellites.

[0062] The RPC model is used to orthogonally correct the images and unify them to the WGS84 coordinate system.

[0063] In one specific embodiment of this application, multi-source satellite coordination is employed: Landsat 8 provides continuous observations, while GF-1 / WFV supplements high-frequency monitoring; radiometric calibration involves converting raw digital values ​​into surface reflectance, expressed by the following formula: ;

[0064] Geometric correction: Orthogonal correction is performed using the RPC model to unify the imagery to the WGS84 coordinate system;

[0065] Hydrological API integration: Real-time acquisition of lake water level and flow data for correcting inundation area;

[0066] Meteorological data fusion: Precipitation and evaporation are obtained through OpenWeatherMap, and input parameters for hydrological models are constructed to predict the transition nodes between wet and dry seasons;

[0067] Historical reclamation database: It calls upon the Google Earth Engine land use dataset to provide prior reclamation area boundaries to assist in model training and post-processing;

[0068] Lightweight inference for edge devices: Depthwise separable convolutions are used instead of standard convolutions, where the number of parameters in the depthwise separable convolution is expressed as follows: The standard convolution parameter quantity is expressed as: Where K is the kernel size, C is the number of channels, and H / W is the feature map size, key features are preserved through knowledge distillation, which is expressed as: .

[0069] The transmission steps of the data transmission layer include: compressing the image using the JPEG2000 compression algorithm; and transmitting the compressed image data stream to the data processing layer via a 5G network or fiber optic leased line.

[0070] In one embodiment, the JPEG2000 compression algorithm is used for compression (this algorithm uses discrete wavelet transform and multi-resolution coding technology to reduce the original data volume to 1 / 10-1 / 5 of the original size while maintaining key texture information), and then the compressed data stream is transmitted to the cloud via a 5G network or fiber optic leased line.

[0071] The data processing layer is used to extract dynamic texture features by batch processing images;

[0072] The patch shape index includes aspect ratio and compactness;

[0073] The data processing layer is also used to fuse patch shape indices and texture trajectories of dynamic texture features through a graph convolutional network (GCN) to perform deep coupling of multi-source features and obtain edge inference results. Specific steps include:

[0074] Construct a dynamic texture feature library as follows:

[0075] Extract the gray-level co-occurrence matrix of candidate patches in the image to obtain dynamic texture features;

[0076] Dynamic texture features include: GLCM contrast, entropy value, and angular second moment;

[0077] Dynamic texture features stored quarterly to construct texture trajectories ;

[0078] Constructing a shape-texture coupling model, which is the edge inference result, involves the following steps:

[0079] Each candidate patch is treated as a node, and its feature vector is represented as follows:

[0080]

[0081] in Indicates the first Feature vectors of each node; Aspect ratio, For firmness, The slope of the quarterly contrast change; This represents the slope of the quarterly entropy change. This represents the slope of the second moment of the quarter angle.

[0082] Connect spatially adjacent candidate patches;

[0083] Using texture similarity as edge weights, denoted as...

[0084]

[0085] in Indicates the first The node and the first The weights of the edges between nodes; Represents the texture feature of the i-th node; Represents the texture feature of the j-th node; Indicates the standard deviation;

[0086] The graph convolutional layers of the graph convolutional network use the Chebyshev multinomial approximation, and the propagation rule is expressed as follows:

[0087]

[0088] in This represents the output of the l-th graph convolutional layer; This represents the order of the Chebyshev polynomial; These are learnable parameters; For Chebyshev polynomials; For the normalized Laplace matrix, This represents the weight matrix of the l-th layer;

[0089] The loss function of a graph convolutional network is to jointly optimize texture trajectory and shape regularity, expressed as:

[0090]

[0091] in, Indicates the weighting parameter; The mean square error of the texture trajectory; Mean square error representing shape regularity; This represents the regularization term.

[0092] To address uncertainty, the graph convolutional edge inference results are used to analyze cloud cover uncertainty. Combined with DEM elevation thresholds, the graph convolutional network is used to refine the reclamation / sand mining boundaries, distinguishing between natural changes and reclamation, and generating lake boundary maps.

[0093] In one specific embodiment of this application, the data analysis layer combines DEM elevation thresholds and edge inference results, analyzes cloud cover uncertainty through Monte Carlo simulation, optimizes the accuracy of lake water body boundaries, and refines reclamation / sand mining boundaries using a GCN model based on land use data or spectral features, distinguishing between natural changes and reclamation, and generating a lake boundary map, including the following steps:

[0094] Cloud coverage probability modeling: statistically analyze historical cloud cover distribution and define the cloud coverage probability. It follows a Beta distribution;

[0095] Random sampling simulation: Multiple random cloud masks are generated, and the water body boundary is re-extracted each time. The DEM elevation threshold and edge inference results are used as basic data or initial conditions in the simulation process to calculate the error distribution.

[0096] Graph structure: Nodes are candidate patches, and feature vectors include: GLCM contrast, entropy, aspect ratio, and compactness. Edges connect spatially adjacent patches, and edge weights are spectral similarity.

[0097] Generate the final lake boundary map: By combining the DEM elevation threshold, edge inference results, Monte Carlo confidence intervals, and graph convolutional network classification probabilities, a lake boundary map is generated. The final lake boundary is represented as DEM elevation threshold region ∩ graph convolutional network high probability region ∩ Monte Carlo confidence interval.

[0098] In one specific embodiment of this application, during the high-water season: the lake boundary includes a complete water surface and wetlands, and the misclassification rate of the reclaimed area is <3%. During the low-water season: it successfully distinguishes between sand mining traces (linear texture abrupt changes) and natural lake beaches (chaotic textures), with a GCN model accuracy of 92%.

[0099] The data analysis layer is also used to extract the lake's extent, compare the lake boundaries during the wet and dry seasons, identify changes in lake area, and generate a boundary change map.

[0100] In one specific embodiment of this application, the data analysis layer identifies area changes by extracting the lake's extent and comparing the lake's boundaries during the high-water and low-water seasons. Specific steps include:

[0101] Multi-temporal image preprocessing: Based on the dry season image, the wet season image is automatically registered (RMS error < 0.5 pixels) to eliminate displacement error. The histogram matching algorithm is used to adjust the reflectance distribution of the wet / dry season images to be consistent, reducing the impact of seasonal light differences.

[0102] Dynamic water body extraction: Water body boundaries during the high-water season are extracted, and continuous water body regions are extracted using the improved MNDWI index, with the area denoted as... Extract the water body boundary during the dry season, and combine the DEM elevation threshold and dynamic NDWI threshold to extract the water body during the dry season. The area is denoted as [missing information]. ;

[0103] Boundary optimization and change detection: Morphological processing is performed on the extracted water body boundaries during the wet and dry seasons to remove noise points, and the absolute change is calculated. And the relative rate of change; vector superposition of the water body boundary during the wet season and the water body boundary during the dry season to identify the expansion zone, contraction zone and stable zone;

[0104] Statistical significance test: The Z-score is used to test whether the change in area is significant, and is expressed as... ,in and The mean and standard deviation of the changes over the same historical period;

[0105] Results output and verification: Different colors are used to mark the expansion region, contraction region and stable region to generate a boundary change map.

[0106] The data analysis layer is also used to identify abrupt change points in sand mining areas based on a time-series texture feature library and an isolated forest algorithm, and to predict the rate of land reclamation expansion by combining a linear regression and LSTM hybrid model.

[0107] In one specific embodiment of this application, the data analysis layer is further used to identify abrupt change points in sand mining areas based on a time-series texture feature library using the isolated forest algorithm, and to predict the rate of land reclamation expansion by combining a linear regression and LSTM hybrid model, including the following steps:

[0108] Calculate the feature deviation: Calculate the Z-score for each time series;

[0109] Isolation Forest Scoring: Train an isolation forest model and output anomaly scores. A higher score indicates a greater anomaly, where the path length of an isolated forest is... The abnormal score is ,in Let x be the path length of sample x;

[0110] Linear regression captures long-term trends: it fits the relationship between reclaimed area and time, expressed as... ,in This represents the annual expansion rate;

[0111] LSTM for seasonal fluctuations: Input seasonal texture features (contrast, ASM, entropy), pass through LSTM layers, fully connected layers, and linear activation, outputting the quarterly increase in reclaimed area. ;

[0112] Hybrid model integration: Linear regression and LSTM are integrated to obtain the predicted total area;

[0113] Ecological patterns and anomaly analysis: linear regression slope Reflecting the rate of land reclamation expansion, LSTM captures seasonal sand mining activities, and abrupt changes detected in isolated forests correspond to sand mining events, which are then verified by overlaying land use data.

[0114] The data application layer uses the TimeManager tool to create lake evolution animations by combining lake boundary maps, boundary change maps, and reclamation expansion rates.

[0115] In one specific embodiment of this application, the data application layer, through a user interface, uses TimeManager to create a lake evolution animation based on data from the data analysis layer, overlays a quarterly GLCM index change curve, highlights abnormal areas, and displays the spatial distribution of reclamation and sand mining probabilities, including the following steps:

[0116] Lake evolution animation production: Extract multi-period boundary vector files of the target lake from the database, with the time range covering the years of alternation between the high water and low water seasons. Set the time slider in the TimeManager tool, set the step size to quarter, bind the time field of the boundary file, generate dynamic GIF or MP4, and overlay the quarterly timestamp.

[0117] Overlaying GLCM indicator change curves: Obtain the GLCM indicator of the candidate region, embed a line chart in the sidebar of the animation interface, with the horizontal axis representing the quarter and the vertical axis representing the standardized indicator value, and mark anomalies.

[0118] Abnormal area highlighting: Abnormal patches are highlighted by identifying abrupt changes in the sand mining area using the isolated forest algorithm;

[0119] Probability heatmap of reclamation / sand mining: based on the output of the GCN model, the Viridis color scale is used for grading and color marking.

[0120] This application also discloses an electronic device. (See reference...) Figure 2 , Figure 2 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. The electronic device 500 may include: at least one processor 501, at least one network interface 504, a user interface 503, a memory 505, and at least one communication bus 502.

[0121] The communication bus 502 is used to enable communication between these components.

[0122] The user interface 503 may include a display screen, and optionally, the user interface 503 may also include a standard wired interface or a wireless interface.

[0123] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0124] This application also discloses a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to execute the aforementioned lake data processing system based on remote sensing images.

[0125] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure.

[0126] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A lake data processing system based on remote sensing images, characterized in that the system... include: Data acquisition layer: Using the YOLO-R model, images of the lake area and DEM elevation thresholds are captured and processed in real time; Data transmission layer: Transmits the processed image to the data processing layer; Data processing layer: processes the images to obtain edge inference results; The data processing layer is used to extract dynamic texture features by batch processing images; The patch shape index includes aspect ratio and compactness; The data processing layer is also used to fuse patch shape indices and texture trajectories of dynamic texture features through a graph convolutional network (GCN) to perform deep coupling of multi-source features and obtain edge inference results. Specific steps include: Construct a dynamic texture feature library as follows: Extract the gray-level co-occurrence matrix of candidate patches in the image to obtain dynamic texture features; Dynamic texture features include: GLCM contrast, entropy value, and angular second moment; Dynamic texture features stored quarterly to construct texture trajectories ; Constructing a shape-texture coupling model, which is the edge inference result, involves the following steps: Each candidate patch is treated as a node, and its feature vector is represented as follows: in Indicates the first The feature vector of each node; Aspect ratio, For firmness, The slope of the quarterly contrast change; This represents the slope of the quarterly entropy change. This represents the slope of the second moment of the quarter angle; Connect spatially adjacent candidate patches; Using texture similarity as edge weights, denoted as... in Indicates the first The node and the first The weights of the edges between nodes; Represents the texture feature of the i-th node; Indicates the first Texture features of each node; Indicates the standard deviation; The graph convolutional layers of the graph convolutional network use the Chebyshev multinomial approximation, and the propagation rule is expressed as follows: in This represents the output of the l-th graph convolutional layer; Indicates the order of the Chebyshev polynomial; These are learnable parameters; For Chebyshev polynomials; For the normalized Laplace matrix, This represents the weight matrix of the l-th layer; The loss function of a graph convolutional network is to jointly optimize texture trajectory and shape regularity, expressed as: in, Indicates the weighting parameter; The mean square error of the texture trajectory; Mean square error representing shape regularity; Represents the regularization term; Data analysis layer: By using DEM elevation thresholds and edge inference results, lake boundary maps, boundary change maps, and reclamation expansion rates are obtained; Data application layer: By using lake boundary maps, boundary change maps, and reclamation expansion rates, lake evolution animations are created to complete the water body boundary processing of lake remote sensing images.

2. The lake data processing system based on remote sensing images as described in claim 1, characterized in that, The data acquisition layer utilizes the YOLO-R model to capture images and DEM elevation thresholds of the lake area in real time, including the following steps: Real-time imagery and DEM elevation thresholds of the lake area are acquired using multi-source satellites. The RPC model is used to orthogonally correct the images and unify them to the WGS84 coordinate system.

3. The lake data processing system based on remote sensing images as described in claim 1, characterized in that, The transmission steps of the data transmission layer include: compressing the image using the JPEG2000 compression algorithm; and transmitting the compressed image data stream to the data processing layer via a 5G network or fiber optic leased line.

4. The lake data processing system based on remote sensing images as described in claim 1, characterized in that, To address uncertainty, we analyze cloud cover uncertainty using graph convolution edge inference results. Combined with DEM elevation thresholds, we refine reclamation / sand mining boundaries using graph convolutional networks to distinguish between natural changes and reclamation, generating lake boundary maps.

5. A lake data processing system based on remote sensing images as described in claim 1, characterized in that, The data analysis layer is also used to extract the lake's extent, compare the lake boundaries during the wet and dry seasons, identify changes in lake area, and generate a boundary change map.

6. The lake data processing system based on remote sensing images as described in claim 1, characterized in that, The data analysis layer is also used to identify abrupt change points in sand mining areas based on a time-series texture feature library and an isolated forest algorithm, and to predict the rate of land reclamation expansion by combining a linear regression and LSTM hybrid model.

7. A lake data processing system based on remote sensing images as described in claim 1, characterized in that, The data application layer uses the TimeManager tool to create lake evolution animations by combining lake boundary maps, boundary change maps, and reclamation expansion rates.

Images