Mangrove forest height estimation method, device, equipment and medium
By using multi-source satellite remote sensing data and a random forest regression model, the problem of mangrove tree height measurement being limited by the growth environment was solved, and accurate tree height estimation was achieved in shallow sea areas.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-18
- Publication Date
- 2026-03-10
AI Technical Summary
In existing technologies, the measurement of mangrove tree height is limited by the fact that they grow in shallow sea areas such as estuaries and bays, making it inconvenient to conduct manual on-site measurements.
Using multi-source satellite remote sensing data, including spectral bands, topographic features, and radar backscattering characteristics, the mean tree height of mangroves was estimated using a random forest regression model. Tree height was estimated using data from the ESA Sentinel-2 multispectral imager and synthetic aperture radar, combined with the random forest regression model.
It enables accurate estimation of mangrove tree height in shallow sea areas, improving the accuracy of tree height estimation and is not limited by the mangrove growth environment.
Smart Images
Figure CN121632045A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of tree height statistics technology, specifically to a method, apparatus, equipment, and medium for estimating the height of mangrove trees. Background Technology
[0002] Mangroves, as important coastal ecosystems, play a crucial role in biodiversity conservation, coastal protection, and carbon sequestration. Accurately determining mangrove tree height is essential for assessing their ecological functions, monitoring their health, and supporting sustainable management. Current technologies rely on manual field measurements to determine mangrove tree height; however, mangroves mostly grow in shallow sea areas such as estuaries and bays, which are not suitable for manual field measurements.
[0003] In conclusion, existing methods for measuring mangrove tree height are limited by the mangrove's growth environment.
[0004] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention
[0005] To address the aforementioned technical problems, this invention provides a method, apparatus, equipment, and medium for estimating mangrove tree height, which solves the problem that existing methods for measuring mangrove tree height are limited by the mangrove growth environment.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for estimating the height of mangrove trees, comprising: Determine the boundary area where the mangroves are located; Collect multi-source satellite remote sensing data of the boundary region, including spectral bands, topographic features, and radar backscattering features; Based on the spectral bands, the topographic features, and the radar backscattering characteristics, the average tree height of the mangroves in the boundary area is estimated.
[0007] In one implementation, determining the boundary area where the mangrove forest is located includes: Traverse the boundary points of the mangrove planting area; Determine the minimum convex polygon constructed from the boundary points; Extend a set distance outward from the smallest convex polygon to obtain the boundary area where the mangrove forest is located.
[0008] In one implementation, estimating the average tree height of the mangroves in the boundary region based on the spectral band, the terrain features, and the radar backscattering characteristics includes: Based on the spectral bands, remote sensing indices are determined, including NDVI; Based on the NDVI, determine the ecological structural indicators of the boundary area; The average tree height of mangroves in the boundary area is estimated based on the spectral bands, remote sensing indices, ecological structural indicators, topographic features, and radar backscattering characteristics.
[0009] In one implementation, determining the remote sensing index based on the spectral band includes: Identify the blue, green, red, and near-infrared bands within the aforementioned spectral bands. Shortwave infrared band; NDVI is determined based on the near-infrared band and the red band. Based on the green light band and shortwave infrared band, the MNDWI is determined; The NDBI is determined based on the near-infrared band and the short-wave infrared band. The EVI is determined based on the blue light band, the short-wave infrared band, the red light band, and the near-infrared band. BSI is determined based on the blue light band, the red light band, the near-infrared band, and the short-wave infrared band; and NDVI, MNDWI, NDBI, EVI, and BSI are used as remote sensing indices.
[0010] In one implementation, the average tree height of the mangroves in the boundary region is estimated based on the spectral band, the remote sensing index, the ecological structural index, the topographic features, and the radar backscattering characteristics, including: A random forest regression model is applied to the spectral bands, the remote sensing indices, the ecological structural indicators, the topographic features, and the radar backscattering features to estimate the mean tree height of the mangroves in the boundary area.
[0011] In one implementation, a global mangrove observation dataset is obtained, which includes spectral sample bands, topographic sample features, and radar backscatter sample features collected in the mangrove observation area, as well as tree height labels in the mangrove observation area. The original values of the radar backscatter sample features are standardized to obtain the dB values of the radar backscatter sample features. Based on the numerical value of the tree height label, invalid labels are deleted from the tree height label to obtain the preprocessed tree height label; Remove terrain sample features that are greater than a set value to obtain preprocessed terrain sample features; The spectral sample bands, the dB values, the preprocessed tree height labels, and the preprocessed terrain sample features are transformed into the same coordinate system to obtain training samples.
[0012] In one implementation, the radar backscatter sample features are data acquired by a synthetic aperture radar located on an ascending orbit.
[0013] Secondly, embodiments of the present invention also provide a mangrove tree height estimation device, wherein the device comprises the following components: The boundary delineation module is used to determine the boundary area where the mangroves are located; The data acquisition module is used to acquire multi-source satellite remote sensing data of the boundary area, including spectral bands, topographic features, and radar backscattering features. The tree height estimation module is used to estimate the average tree height of the mangroves in the boundary area based on the spectral band, the terrain features, and the radar backscattering features.
[0014] Thirdly, embodiments of the present invention also provide a terminal device, wherein the terminal device includes a memory, a processor, and a mangrove tree height estimation program stored in the memory and executable on the processor, wherein when the processor executes the mangrove tree height estimation program, it implements the steps of the mangrove tree height estimation method described above.
[0015] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a mangrove tree height estimation program, wherein when the mangrove tree height estimation program is executed by a processor, the steps of the mangrove tree height estimation method described above are implemented.
[0016] Beneficial Effects: This invention collects spectral bands, topographic features, and radar backscattering characteristics of the boundary area where mangroves are located, and estimates the average tree height of mangroves within that boundary area using these data. Since the topographic features and radar backscattering are obtained from satellite remote sensing data acquired by radar, and the spectral bands are collected from a spectral imager sensor, and radar can acquire satellite remote sensing data for any boundary area where mangroves are located, even if the boundary area is in shallow sea, this invention's method for estimating mangrove tree height is not limited by the mangrove's environment. Furthermore, this invention uses multi-source satellite remote sensing data to estimate tree height, which improves the accuracy of the estimated tree height. Attached Figure Description
[0017] Figure 1 This is the overall process of the present invention; Figure 2 This is a structural diagram of the mangrove tree height estimation device provided by the present invention; Figure 3 This is a block diagram illustrating the internal structure of a terminal device provided in an embodiment of the present invention. Detailed Implementation
[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0019] Research has shown that mangroves, as important coastal ecosystems, play a crucial role in biodiversity conservation, coastal protection, and carbon sequestration. Accurately determining mangrove tree height is essential for assessing their ecological functions, monitoring their health, and supporting sustainable management. Current technologies rely on manual field measurements to determine mangrove tree height; however, mangroves mostly grow in shallow sea areas such as estuaries and bays, which are not suitable for manual field measurements.
[0020] To address the aforementioned technical problems, this invention provides a method, apparatus, equipment, and medium for estimating mangrove tree height, which solves the problem that existing methods for measuring mangrove tree height are limited by the mangrove growth environment.
[0021] Example 1: The mangrove tree height estimation method of this example can be applied to a terminal device, which can be a terminal product with data processing capabilities, such as a computer. In this example, as... Figure 1 As shown, the mangrove tree height estimation method specifically includes the following steps: S100, defines the boundary area where the mangroves are located; S200, Collect multi-source satellite remote sensing data of the boundary area, including spectral bands, topographic features and radar backscattering features; S300, based on the spectral band, the terrain features, and the radar backscattering features, estimate the average tree height of the mangroves in the boundary area.
[0022] In this embodiment, determining the boundary area where the mangrove forest is located in step S100 includes the following specific steps S101, S102, and S103: S101 traverses the boundary points of mangrove planting areas.
[0023] Based on the vector boundary of the mangrove area, S102, determine the smallest convex polygon constructed from the boundary points.
[0024] Based on the vector boundary of the mangrove region, the smallest convex polygon containing all boundary points is found by traversing all boundary points.
[0025] Since the planting area of mangroves may not be a regular shape, it is necessary to collect images of the planting area and identify the mangroves using image recognition technology. The mangroves near the edge of the planting area are used as boundary points. The closed area formed by several boundary points can be of any shape. The reason why this embodiment selects the smallest convex polygon that includes all boundary points is that the smallest convex polygon can not only cover all the growth areas of mangroves, but also ensure that there are no extra invalid areas in the covered growth areas. For example, if the growth area of mangroves is a square, there can be multiple convex polygons of different sizes outside the square. Only the convex polygon with the smallest area is selected.
[0026] S103, extend a set distance outward from the smallest convex polygon to obtain the boundary area where the mangrove forest is located.
[0027] In this embodiment, the set distance is 10 kilometers, which means that the minimum convex polygon and the area 10 kilometers away from the minimum convex polygon are taken as the boundary area of the mangrove forest.
[0028] The boundary region, or Region of Interest (ROI), serves to ensure that subsequent remote sensing data extraction covers the entire mangrove vegetation area and external environmental information that may affect its growth by generating an area that includes the mangroves and their surrounding influence zones. This provides the model with more comprehensive environmental variable information.
[0029] Step S200 involves acquiring spectral bands using the ESA Sentinel-2 multispectral imager, including twelve spectral bands from B1 to B12. Topographic features and radar backscattering characteristics of the boundary area are then acquired using lidar.
[0030] In this embodiment, step S300 includes the following specific steps S301 to S308: S301, determine the blue light band (blue light band is B2 in B1 to B12), green light band (green light band is B3 in B1 to B12), red light band (red light band is B4 in B1 to B12), near-infrared band (near-infrared band is B8 in B1 to B12), and short-wave infrared band (short-wave infrared band is B11 in B1 to B12) in the spectral band.
[0031] The ESA Sentinel-2 multispectral imager can collect twelve spectral bands from B1 to B12 for the boundary area where mangroves are located. In this embodiment, B3, B4, B8, and B11 of the twelve spectral bands from B1 to B12 are used to calculate the remote sensing index.
[0032] S302, determine NDVI based on the near-infrared band B8 and the red band B4: NDVI=(B8-B4) / (B8+B4).
[0033] S303, based on the green light band B3 and the short-wave infrared band B11, determine the MNDWI: MNDWI=(B3-B11) / (B3+B11).
[0034] S304, Based on the near-infrared band B8 and the short-wave infrared band B11, determine NDBI: NDBI=(B11-B8) / (B11+B8).
[0035] S305, determine EVI based on the blue light band B2, the short-wave infrared band B11, the red light band B4, and the near-infrared band B8: EVI=((B11+B4)-(B8+B2)) / ((B11+B4)+(B8+B2)).
[0036] S306, determine the BSI based on the blue light band, the red light band, the near-infrared band, and the short-wave infrared band; and use the NDVI, the MNDWI, the NDBI, the EVI, and the BSI as remote sensing indices.
[0037] NDVI represents the Normalized Difference Vegetation Index. NDVI is closely related to the growth status and density of mangroves and can indirectly reflect the height of mangrove trees. EVI represents the Enhanced Vegetation Index. EVI can reflect the vegetation growth status and biomass characteristics of mangroves. EVI has a certain correlation with the average height of mangroves, so it can be used as one of the remote sensing indicators for estimating the height of mangrove trees. MNDWI represents the Normalized Difference Water Index. The value of MNDWI reflects the humidity level of mangroves, and humidity affects the growth of mangroves. Therefore, the average height of trees in the mangrove area can be estimated through MNDWI. NDBI represents the Normalized Difference Building Index, and BSI represents the Bare Soil Index.
[0038] The NDBI and BSI work together to distinguish between bare land and urban areas, helping to accurately indicate mangrove areas in complex coastal environments.
[0039] S307, Based on the NDVI, determine the ecological structural indicators of the boundary area.
[0040] In this embodiment, mangrove coverage is used as an ecological structural indicator, namely FVC. Calculating mangrove coverage based on NDVI is an existing technology.
[0041] The 1% and 99th percentiles of NDVI values (representing the sparsest and densest vegetation cover in the region, respectively) are dynamically extracted from the boundary area as normalization parameters. These two dynamic thresholds are then used to linearly stretch the NDVI values, thus standardizing the uneven distribution of NDVI values caused by differences in vegetation types across regions to a unified FVC interval. To further ensure the physical rationality of the calculation results, this invention also establishes a three-level masking mechanism: when the calculation result is less than 0, it is forcibly set to 0; when the result is between 0 and 1, the original value is maintained; and when the result is greater than 1, it is forcibly set to 1. This method not only automatically adapts to the vegetation characteristics of different geographical regions, improving the accuracy and reliability of FVC calculation, but also significantly enhances the stability and prediction accuracy of the mangrove tree height estimation model by providing accurate vegetation canopy density information.
[0042] S308, apply a random forest regression model to the spectral bands, the remote sensing index, the ecological structural index, the topographic features, and the radar backscattering features to estimate the average tree height of the mangroves in the boundary area.
[0043] In other words, data is collected in the boundary area. The collected spectral bands, including the twelve bands B1 to B12, NDVI, MNDWI, NDBI, EVI, BSI, FVC, the elevation and slope features of the terrain, and four radar backscattering features, totaling twenty-four features, are input into the random forest regression model. The random forest regression model outputs the average tree height of the mangroves in the boundary area. In other words, the random forest regression model can estimate the impact of the above four data on the growth of mangroves based on the terrain, spectral bands, remote sensing index, and radar backscattering features of the boundary area, thereby estimating the average tree height reached by the mangroves under this influence.
[0044] The four radar backscattering features in this embodiment include Sentinel-1's VV, Sentinel-1's VH, ALOS PALSAR's HH, and ALOS PALSAR's HV. Sentinel-1's VV represents the signal wave formed vertically after a radar wave emitted vertically is reflected by mangroves; Sentinel-1's VH represents the signal wave formed horizontally after a radar wave emitted vertically is reflected by mangroves; ALOS PALSAR's HH represents the signal wave formed horizontally after a radar wave emitted horizontally is reflected by mangroves; and ALOS PALSAR's HV represents the signal wave formed vertically after a radar wave emitted horizontally is reflected by mangroves. These four radar backscattering features interact to reflect the growth environment provided by the boundary area where the mangroves are located. This growth environment, in turn, affects the overall tree height of the mangroves. Therefore, the average tree height of the mangroves can be estimated using these four radar backscattering features.
[0045] Example 2, based on Example 1, provides a training method for a random forest regression model. Specifically, it provides a pre-trained random forest regression model for Example 1. The training method includes constructing training samples, partitioning the training samples, and constructing a loss function. Constructing training samples includes the following specific steps S01, S02, S03, S04, and S05: S01, Obtain the global mangrove observation dataset, which includes spectral sample bands, topographic sample features, and radar backscatter sample features collected in the mangrove observation area, as well as tree height labels in the mangrove observation area.
[0046] The observation area differs from the boundary area mentioned above. The height of mangroves in the boundary area is unknown, while the height of mangroves in the observation area is known.
[0047] The spectral sample bands included are the same as those in Example 1, namely the twelve bands from B1 to B12. The terrain sample features also include elevation and slope, and the radar backscattering sample features also include Sentinel-1 VV, Sentinel-1 VH, ALOS PALSAR HH, and ALOS PALSAR HV.
[0048] The Sentinel-1 band and the PALSAR band have complementary penetration characteristics. That is, the Sentinel-1 band is sensitive to the top of the mangrove canopy, while the PALSAR band has stronger penetration capabilities and is more sensitive to the interior of the canopy and the trunk structure. The combination of the two can provide the model with more comprehensive information on the vertical structure of mangroves.
[0049] Tree height labels are RH98 data of mangroves in the observation area obtained through GEDI. RH98 data represents the canopy height of mangroves in the observation area.
[0050] GEDI represents the data of mangroves in the observation area obtained by lidar. The lidar emits lasers into the mangroves. After the laser pulses illuminate the ground, some of the energy is reflected back by the ground, canopy, and branches at different heights. The lidar receiver records the waveform of the returned energy changing over time and accumulates the waveform point by point from the ground upwards to obtain the cumulative energy curve. When the cumulative energy reaches 98% of the total energy, the corresponding height is RH98.
[0051] S02, the original values of the radar backscatter sample features are standardized and transformed to obtain the dB values of the radar backscatter sample features.
[0052] That is, the original values (DN) of Sentinel-1 VV, Sentinel-1 VH, ALOS PALSAR HH, and ALOS PALSAR HV contained in the radar backscatter sample features are standardized to obtain dB values.
[0053] dB = 10 * log10(DN²) - 83.0.
[0054] The Sentinel-1 in this embodiment is a Sentinel-1 that uses an ascent trajectory (ASCENDING). Data is used to reduce the impact of geometric distortion caused by orbital orientation.
[0055] S03, based on the numerical value of the tree height label, delete invalid labels from the tree height label to obtain the preprocessed tree height label.
[0056] In other words, the laser energy reflected back from the mangroves collected by GEDI is filtered based on GEDI's own quality label, retaining the laser energy with quality_flag equal to 1 and degrade_flag equal to 0. Based on the retained laser energy, RH98 is obtained. Then, the RH98 is checked for numerical range. For example, only RH98 values greater than 0 are retained to exclude invalid lidar measurements, so as to obtain the final RH98. The final RH98 is the tree height label after preprocessing.
[0057] S04, remove terrain sample features that are greater than the set value to obtain preprocessed terrain sample features.
[0058] This involves applying terrain-based quality control, for example, retaining only pixels with a slope of less than 30° to eliminate measurement errors in steep terrain areas.
[0059] S05, the spectral sample bands, the dB values, the preprocessed tree height labels, and the preprocessed terrain sample features are transformed into the same coordinate system to obtain training samples.
[0060] This embodiment also obtains the remote sensing sample index based on the spectral sample band. This process is the same as obtaining the remote sensing index from the spectral band in step S300, and will not be described in detail here.
[0061] This embodiment converts positive or negative infinity values appearing in the spectral sample bands into missing values (NaN), calculates the mean for each band, and uses this mean to fill in the missing positions through interpolation, resulting in the interpolated spectral sample bands. This method recovers some feature data lost due to remote sensing image quality issues without significantly introducing bias.
[0062] For tree height labels, all locations with missing or infinite values are directly removed to ensure that the target values used for training are derived from reliable measurement results and have clear physical meaning.
[0063] Finally, in this embodiment, the interpolated spectral sample bands, dB values, remote sensing sample indices, preprocessed tree height labels (after removing missing and infinite values), and preprocessed terrain sample features are all spatially registered under the same coordinate system (UTM) on the Google Earth Engine (GEE) platform. All data sources are resampled to 25-meter resolution using bilinear interpolation and uniformly converted to floating-point type to maintain data continuity and eliminate spatial differences between different data sources, thus obtaining training samples.
[0064] The integrity and stability of the training samples are significantly improved, while retaining the maximum number of usable samples. This effectively avoids interference from non-mangrove areas and remote sensing outliers on the model, ensuring the stability, interpretability, and generalization ability of the model training.
[0065] This embodiment uses training samples to train a random forest regression model. Before training, the training samples are divided into training and test sets in a 70%:30% ratio. The loss function used for model training is mean squared error (MSE). The training terminates when a preset number of iterations (i.e., the number of decision trees in the random forest) is reached. The model performance is comprehensively evaluated using four metrics: mean squared error (MSE), root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R²). After the model training is complete, it is saved in a general model format for use by the subsequent region prediction module.
[0066] In summary, this invention employs a random forest regression model with excellent nonlinear fitting capabilities for modeling. This model is combined with multi-source satellite remote sensing data and leverages the efficient data processing capabilities of Google Earth Engine (GEE). This combination allows the model to fully utilize the complementary advantages of different remote sensing data sources, effectively capturing the nonlinear relationship between mangrove tree height and complex environmental factors, thereby significantly improving prediction accuracy and generalization ability on a large scale.
[0067] This embodiment also provides a mangrove tree height estimation device, such as... Figure 2 As shown, the device comprises the following components: The boundary delineation module is used to determine the boundary area where the mangroves are located; The data acquisition module is used to acquire multi-source satellite remote sensing data of the boundary area, including spectral bands, topographic features, and radar backscattering features. The tree height estimation module is used to estimate the average tree height of the mangroves in the boundary area based on the spectral band, the terrain features, and the radar backscattering features.
[0068] Based on the above embodiments, the present invention also provides a terminal device, the principle block diagram of which can be as follows: Figure 3 As shown, the terminal device includes a processor, memory, network interface, and display screen connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for estimating the height of mangrove trees. The display screen can be an LCD screen or an e-ink screen.
[0069] Those skilled in the art will understand that Figure 3 The schematic diagram shown is only a partial structural diagram related to the present invention and does not constitute a limitation on the terminal device to which the present invention is applied. The specific terminal device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.
[0070] In one embodiment, a terminal device is provided, comprising a memory, a processor, and a mangrove tree height estimation program stored in the memory and executable on the processor. When the processor executes the mangrove tree height estimation program, it implements the following operation instructions: Determine the boundary area where the mangroves are located; Collect multi-source satellite remote sensing data of the boundary region, including spectral bands, topographic features, and radar backscattering features; Based on the spectral bands, the topographic features, and the radar backscattering characteristics, the average tree height of the mangroves in the boundary area is estimated.
[0071] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for estimating height of mangrove forest, characterized in that, The method comprises the following steps: determining a boundary area where the mangrove forest is located; collecting multi-source satellite remote sensing data of the boundary area, the multi-source satellite remote sensing data comprising spectral bands, terrain features and radar backscatter features; estimating the average tree height of the mangrove forest in the boundary area according to the spectral bands, the terrain features and the radar backscatter features.
2. The mangrove tree height estimation method of claim 1, wherein, The method for determining the boundary area where the mangrove forest is located comprises the following steps: traversing boundary points where the mangrove forest is located in a planting area of the mangrove forest; determining a minimum convex polygon constructed by the boundary points; extending a set distance to the outside of the minimum convex polygon to obtain the boundary area where the mangrove forest is located.
3. The mangrove tree height estimation method of claim 1, wherein, The method for estimating the average tree height of the mangrove forest in the boundary area according to the spectral bands, the terrain features and the radar backscatter features comprises the following steps: determining remote sensing indexes according to the spectral bands, the remote sensing indexes comprising NDVI; determining an ecological structural index of the boundary area according to the NDVI; estimating the average tree height of the mangrove forest in the boundary area according to the spectral bands, the remote sensing indexes, the ecological structural index, the terrain features and the radar backscatter features.
4. The mangrove tree height estimation method of claim 3, wherein, The method for determining remote sensing indexes according to the spectral bands comprises the following steps: determining blue light bands, green light bands, red light bands, near-infrared bands, short-wave infrared bands in the spectral bands; determining NDVI according to the near-infrared bands and the red light bands; determining MNDWI according to the green light bands and short-wave infrared bands; determining NDBI according to the near-infrared bands and short-wave infrared bands; determining EVI according to the blue light bands, the short-wave infrared bands and the red light bands and the near-infrared bands; determining BSI according to the blue light bands, the red light bands and the near-infrared bands and the short-wave infrared bands; and taking the NDVI, the MNDWI, the NDBI and the EVI and the BSI as the remote sensing indexes. The method for estimating the average tree height of the mangrove forest in the boundary area according to the spectral bands, the remote sensing indexes, the ecological structural index, the terrain features and the radar backscatter features comprises the following steps:
5. The mangrove tree height estimation method of claim 3, wherein, applying a random forest regression model to the spectral bands, the remote sensing indexes, the ecological structural index, the terrain features and the radar backscatter features to estimate the average tree height of the mangrove forest in the boundary area. The random forest regression model is a trained model, and the construction of training samples required for training the random forest regression model comprises the following steps:
6. The mangrove tree height estimation method of claim 5, wherein, obtaining a global mangrove forest observation data set, the global mangrove forest observation data set comprising spectral sample bands, terrain sample features and radar backscatter sample features collected in a mangrove forest observation area and tree height labels of the mangrove forest observation area; standardizing the original values of the radar backscatter sample features to obtain the dB values of the radar backscatter sample features; deleting invalid labels from the tree height labels according to the numerical values of the tree height labels to obtain preprocessed tree height labels; removing terrain sample features greater than a set value to obtain preprocessed terrain sample features; The spectral sample band, the dB value, the tree height label after preprocessing and the terrain sample feature after preprocessing are converted to the same coordinate system to obtain training samples.
7. The mangrove tree height estimation method of claim 6, wherein, The radar backscatter sample feature is data collected by a synthetic aperture radar located on an ascending orbit.
8. A mangrove tree height estimation device, characterized by, The device comprises the following components: A boundary division module is configured to determine a boundary area where the mangrove forest is located. A data acquisition module is configured to acquire multi-source satellite remote sensing data of the boundary area, the multi-source satellite remote sensing data comprising a spectral band, a terrain feature and a radar backscatter feature. A tree height estimation module is configured to estimate a tree height average of the mangrove forest in the boundary area according to the spectral band, the terrain feature and the radar backscatter feature.
9. A terminal device, comprising: The terminal device comprises a memory, a processor and a mangrove forest tree height estimation program stored in the memory and executable on the processor, and the processor executes the mangrove forest tree height estimation program to implement the steps of the mangrove forest tree height estimation method according to any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a mangrove forest tree height estimation program, and the mangrove forest tree height estimation program is executed by the processor to implement the steps of the mangrove forest tree height estimation method according to any one of claims 1-7.