Method and system for detecting post-disaster recovery of mangrove forests based on remote sensing data

Abstract

The application provides a mangrove post-disaster recovery detection method and system based on remote sensing data, the method comprising: extracting a mangrove region in a remote sensing image of a first time series remote sensing image set of a target region through a sub-pixel analysis method to obtain a second time series remote sensing image set; constructing a continuous change detection model to identify and obtain a disturbance region in the mangrove region; based on the disturbance region in the mangrove region, performing image classification on the remote sensing image to obtain a pre-disaster mangrove remote sensing image, a post-disaster mangrove remote sensing image and a recovery period mangrove remote sensing image, and using the same to analyze the damage of the mangrove region; judging and analyzing the recovery years of the mangrove region whose damage degree exceeds a damage warning threshold to obtain a recovery year parameter of the mangrove region and performing recovery intervention on the mangrove region according to a preset recovery intervention standard. Thus, the application can improve the recovery detection, management and protection efficiency of the post-disaster mangrove region.

Description

Technical Field

[0001] This application relates to the field of remote sensing monitoring, and in particular to a method and system for detecting post-disaster recovery of mangroves based on remote sensing data. Background Technology

[0002] Mangroves are woody plant communities that grow in the intertidal zone of tropical and subtropical coasts. Understanding the stresses caused by extreme climates on mangrove growth is of great value for the scientific and rational ecological restoration and management of mangrove ecosystems.

[0003] However, existing methods for monitoring mangrove post-disaster recovery are unable to capture the progress of mangrove ecosystem recovery or decline, and they only focus on changes in mangrove cover classification without addressing changes in mangrove conditions. This makes it difficult for managers to implement reasonable recovery interventions. Therefore, they are limited and inefficient in monitoring, managing and protecting mangrove areas after natural disasters. Summary of the Invention

[0004] Therefore, the purpose of this application is to provide a method and system for detecting post-disaster recovery of mangroves based on remote sensing data, which can effectively improve the efficiency of detection, management and protection of mangrove areas after natural disasters.

[0005] The objective of this application can be achieved through the following technical solutions:

[0006] A method for detecting post-disaster recovery of mangrove forests based on remote sensing data includes the following steps: acquiring a first time-series remote sensing image set of the target area, wherein the first time-series remote sensing image set includes several time-series remote sensing images of the target area taken at different time nodes; extracting mangrove areas from each remote sensing image in the first time-series remote sensing image set using a sub-pixel analysis method to obtain a second time-series remote sensing image set; constructing a continuous change detection model to identify change areas in each remote sensing image in the second time-series remote sensing image set, obtaining disturbed areas and stable areas within the mangrove forest area, wherein the disturbed areas within the mangrove forest area are pixel change areas in the remote sensing images of the second time-series remote sensing image set; based on the disturbed areas within the mangrove forest area, using a random forest model to analyze the second time-series remote sensing image set... The remote sensing images in the inter-series remote sensing image set are classified to obtain pre-disaster mangrove remote sensing images, post-disaster mangrove remote sensing images, and recovery-period mangrove remote sensing images. The recovery-period mangrove remote sensing images are those of mangrove areas that have entered a natural recovery state after a natural disaster. Based on the pre-disaster and post-disaster mangrove remote sensing images, a disaster damage analysis is performed on the mangrove areas. Combined with preset disaster damage early warning standards, it is determined whether the degree of damage to the mangrove areas exceeds a preset damage early warning threshold. If so, based on the recovery-period mangrove remote sensing images and the post-disaster mangrove remote sensing images, a recovery years analysis is performed on the mangrove areas to obtain recovery years parameters. Based on the recovery years parameters and preset recovery intervention standards, recovery interventions are implemented for the mangrove areas.

[0007] A mangrove post-disaster recovery detection system based on remote sensing data, comprising: a first time-series remote sensing image acquisition unit, used to acquire a first time-series remote sensing image set of a target area, wherein the first time-series remote sensing image set includes several time-series remote sensing images of the target area taken at different time nodes; a second time-series remote sensing image acquisition unit, used to extract mangrove areas from each remote sensing image in the first time-series remote sensing image set using a sub-pixel analysis method, to obtain a second time-series remote sensing image set; a disturbance area identification and detection unit, used to construct a continuous change detection model to identify change areas in each remote sensing image in the second time-series remote sensing image set, to obtain disturbed areas and stable areas within the mangrove area, wherein the disturbed areas within the mangrove area are pixel change areas in the remote sensing images in the second time-series remote sensing image set; and a remote sensing image classification unit, used to classify mangrove images based on the mangrove area... Within the disturbed area, a random forest model is used to classify the remote sensing images in the second time-series remote sensing image set, resulting in pre-disaster mangrove remote sensing images, post-disaster mangrove remote sensing images, and recovery-period mangrove remote sensing images. The recovery-period mangrove remote sensing images are those of mangrove areas that have entered a natural recovery state after a natural disaster. A damage and recovery years analysis unit is used to analyze the damage to the mangrove area based on the pre-disaster and post-disaster mangrove remote sensing images. Combining this with a preset damage warning standard, it determines whether the damage level of the mangrove area exceeds a preset damage warning threshold. If so, it analyzes the recovery years of the mangrove area based on the recovery-period mangrove remote sensing images and the post-disaster mangrove remote sensing images, obtaining the recovery years parameter for the mangrove area. A recovery intervention unit is used to perform recovery intervention on the mangrove area based on the recovery years parameter and a preset recovery intervention standard.

[0008] Compared to existing technologies, the method described in this application firstly acquires a first time-series remote sensing image set of the target area, and then uses a sub-pixel analysis method to extract mangrove areas, resulting in a second time-series remote sensing image set. Secondly, a continuous change detection model is constructed to identify change areas in each remote sensing image in the second time-series remote sensing image set, obtaining disturbed and stable areas within the mangrove area. Thirdly, based on the disturbed areas within the mangrove area, a random forest model is used to classify each remote sensing image in the second time-series remote sensing image set, obtaining pre-disaster mangrove remote sensing images, post-disaster mangrove remote sensing images, and recovery period mangrove remote sensing images, and performing disaster damage analysis and recovery years analysis. Finally, based on the analysis results, restoration interventions are implemented in the mangrove area. Therefore, the method described in this application not only improves the accuracy of identifying mangrove areas in remote sensing images, but also accurately identifies the specific areas of mangroves that have suffered natural disasters by acquiring disturbed areas, thereby avoiding interference from irrelevant features such as water bodies or bare soil. Furthermore, by conducting disaster damage analysis and recovery year analysis, those skilled in the art can efficiently intervene in the recovery of disaster-stricken mangrove areas, thus effectively improving the efficiency of recovery detection, management, and protection of mangrove areas after natural disasters.

[0009] To better understand and implement this application, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description

[0010] Figure 1 A flowchart of the mangrove post-disaster recovery detection method based on remote sensing data provided in this application;

[0011] Figure 2 A flowchart illustrating the steps for obtaining a second time-series remote sensing image set in the mangrove post-disaster recovery detection method based on remote sensing data provided in this application;

[0012] Figure 3 A flowchart illustrating the steps for obtaining hybrid pixel decomposition results in the mangrove post-disaster recovery detection method based on remote sensing data provided in this application;

[0013] Figure 4 The structural principle diagram of the mangrove post-disaster recovery monitoring system based on remote sensing data provided in this application. Detailed Implementation

[0014] This application provides a method and system for detecting post-disaster recovery of mangroves based on remote sensing data. To make the purpose, technical solution, and effects of this application clearer and more explicit, the following detailed description is provided with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining this application and are not intended to limit this application.

[0015] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0016] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0017] The invention will be further explained below with reference to the accompanying drawings and the description of the embodiments.

[0018] Example 1

[0019] Please refer to Figure 1 , Figure 1 A flowchart illustrating a method for detecting post-disaster recovery of mangrove forests based on remote sensing data, provided in this application. The method includes the following steps:

[0020] S10. Obtain the first time-series remote sensing image set of the target area;

[0021] S20. Extract the mangrove areas from each remote sensing image in the first time series remote sensing image set using the sub-pixel analysis method to obtain the second time series remote sensing image set.

[0022] S30. Construct a continuous change detection model to identify change areas in each remote sensing image in the second time series remote sensing image set, and obtain the disturbed areas and stable areas within the mangrove area.

[0023] S40. Based on the disturbed areas within the mangrove forest area, the remote sensing images in the second time series remote sensing image set are classified using a random forest model to obtain pre-disaster mangrove remote sensing images, post-disaster mangrove remote sensing images, and mangrove remote sensing images during the recovery period.

[0024] S50. Based on the pre-disaster mangrove remote sensing images and the post-disaster mangrove remote sensing images, conduct a disaster damage analysis on the mangrove area, and determine whether the degree of damage to the mangrove area exceeds the preset damage warning threshold by combining the preset disaster damage warning standard; if so, analyze the recovery years of the mangrove area based on the recovery period mangrove remote sensing images and the post-disaster mangrove remote sensing images to obtain the recovery years parameter of the mangrove area.

[0025] S60. Based on the restoration years parameter of the mangrove area and the preset restoration intervention standard, restore the mangrove area by performing restoration intervention.

[0026] Compared to existing technologies, the technical solution of this application integrates time-series remote sensing image analysis with machine learning models to achieve intelligent management of the entire process of disaster damage assessment and recovery intervention in mangrove areas: First, based on sub-pixel analysis methods, mangrove areas are accurately separated from mixed pixels, solving the problem of vegetation classification ambiguity caused by spatial resolution limitations in traditional remote sensing images, and effectively improving the accuracy of mangrove boundary identification; Second, by constructing a continuous change detection model to perform pixel-level dynamic monitoring of time-series images, and combining it with a random forest classification model, mangrove areas are divided into pre-disaster, post-disaster, and recovery phases, enabling those skilled in the art to implement this application... The proposed technical solution can accurately identify areas of instantaneous disturbance and long-term degradation trends caused by natural disasters such as typhoons and floods, and avoid detection errors in mangrove disaster areas caused by tidal fluctuations. It overcomes the limitations and accuracy constraints of existing technologies in capturing dynamic changes and classifying disaster impacts. Furthermore, by quantifying the degree of damage and the number of years to restore the forest, a correlation mechanism is established between disaster damage warning thresholds and restoration intervention standards, achieving a scientific closed loop from disaster monitoring to restoration decision-making. For example, based on the predicted number of years to restore the forest, ecological restoration strategies such as replanting density and hydrological regulation are automatically matched, significantly improving the restoration efficiency and sustainability of mangrove ecosystems. Therefore, this technical solution, through the coupling of different models across multiple stages and data collaboration, achieves for the first time in the field of mangrove protection intelligent management of "damage identification - degree assessment - restoration prediction - decision generation," providing reliable technical support for the precise protection of mangrove ecosystems.

[0027] For step S10, obtain the first time series remote sensing image set of the target area.

[0028] The target area is the area observed, recorded, and photographed by a remote sensing satellite. In one embodiment, the remote sensing satellite may be the Sentinel-2 remote sensing satellite or other remote sensing satellites. The first time-series remote sensing image set includes several time-series remote sensing images of the target area taken by the remote sensing satellite at different time points.

[0029] In one embodiment, step S10 includes the following steps:

[0030] S101. Acquire several remote sensing images of the target area taken at different time points.

[0031] The time node refers to the time when the remote sensing satellite performs the imaging. In one embodiment, the time node can be an exact time, such as 10:00 AM on July 1, 2024. In other embodiments, it can also be a short period of time, such as when shooting long-exposure images, which often requires an exposure time.

[0032] In one embodiment, the Sentinel-2 remote sensing image satellite covers the Earth's surface at a time interval of 5 days, meaning that new image data can be acquired approximately every 5 days in the same area. By periodically revisiting the target area using this remote sensing image satellite, several remote sensing images of the target area captured at different time points can be obtained.

[0033] S102. Preprocess the aforementioned remote sensing images respectively.

[0034] The preprocessing includes geometric correction, radiometric correction, image registration, and atmospheric correction.

[0035] S103. Based on the recording time of the remote sensing images of the target area recorded at different time nodes, sort the preprocessed remote sensing images of each target area recorded at different time nodes by time to obtain the first time series remote sensing image set of the target mangrove area.

[0036] For step S20, the mangrove areas in each remote sensing image in the first time series remote sensing image set are extracted by sub-pixel analysis method to obtain the second time series remote sensing image set.

[0037] The sub-pixel analysis method is a remote sensing image processing method that decomposes mixed pixels by utilizing information such as spectral features and / or spatial features, as well as prior knowledge or auxiliary data, to obtain more refined information on the distribution of ground features.

[0038] In one embodiment, by applying a sub-pixel analysis method to each remote sensing image in the first time-series remote sensing image set, it is determined whether each pixel in the remote sensing image meets a preset threshold rule, thereby determining whether the pixel belongs to a mangrove area; then, all extracted mangrove area images are reorganized and sorted according to the time order of the corresponding images in the first time-series remote sensing image set to form a second time-series remote sensing image set.

[0039] In this embodiment, the pixel is a pixel in the corresponding remote sensing image.

[0040] Please refer to Figure 2 In one embodiment, step S20 includes the following steps:

[0041] S201. Perform hybrid pixel decomposition on each remote sensing image in the first time series remote sensing image set to obtain the abundance information of each land feature endmember in each pixel of the remote sensing image in the first time series remote sensing image set, and obtain the hybrid pixel decomposition result.

[0042] Wherein, the land feature end-member is a pure land feature in the target area; the land feature end-member includes at least mangroves and irrelevant land features (such as vegetation, water bodies and buildings); the abundance information is the proportion of the land feature end-member in a pixel.

[0043] S202. Based on the hybrid pixel decomposition results, the pixel exchange method is used to predict the spatial distribution information of mangroves in each pixel of the remote sensing image in the first time series remote sensing image set, and a sub-pixel land cover distribution map of mangroves is obtained.

[0044] The pixel swapping method is an algorithm for extracting ground feature information at the sub-pixel level.

[0045] In one embodiment, a pixel swapping method is employed to assume that the distribution of land cover categories in the remote sensing images of the first time-series remote sensing image set is spatially correlated within and between pixels (i.e., pixels that are closer to each other are more likely to belong to the same type of land cover compared to pixels that are farther apart). By swapping the positions of sub-pixels, the attraction of the same land cover type is maximized to improve the spatial resolution of the remote sensing images.

[0046] S203. Based on the sub-pixel land cover distribution map of the mangrove forest and combined with the preset remote sensing image features of the mangrove forest, a supervised classification method and / or an unsupervised classification method are used to extract the mangrove forest area from each remote sensing image in the first time series remote sensing image set, obtain the corresponding number of remote sensing images of the mangrove forest area, and obtain the second time series remote sensing image set.

[0047] The preset remote sensing image features of mangroves include at least the spectral features, texture features, and structural features of mangroves.

[0048] Please also refer to Figure 3 In one embodiment, step S201 includes the following steps:

[0049] S2011. Set the pixels in the remote sensing images of the first time-series remote sensing image set as a linear combination of the spectral features of the ground object endmembers:

[0050] r = Eα + ε;

[0051] Where r is the spectral reflectance vector of the pixel; E is the end-member spectral matrix, and each column of the end-member spectral matrix represents the spectral characteristics of an end-member; α is the abundance vector representing the proportion of the corresponding end-members in the pixel; and ε is a preset error term.

[0052] S2012. Using the fully constrained least squares method, the abundance inversion of the linear combination is performed and the reconstruction error is minimized to obtain the first problem to be solved:

[0053]

[0054] Where, α i Let m be the abundance vector of the proportion of the i-th land feature endmember in the pixel, and m be the number of land feature endmembers.

[0055] S013. Construct the abundance Lagrangian function for the first problem to be solved:

[0056]

[0057] Where λ and μ are preset Lagrange multipliers.

[0058] S2014. Solve the abundance Lagrangian function to obtain the abundance information of each land feature end-member in the remote sensing image of the first time series remote sensing image set in each pixel, and obtain the mixed pixel decomposition result.

[0059] For step S30, constructing a continuous change detection model, the change region is identified for each remote sensing image in the second time series remote sensing image set, and the disturbed region and stable region within the mangrove area are obtained.

[0060] The continuous change detection model relies on remote sensing image data and identifies and analyzes changes in land cover by comparing image information at different time points; the disturbed area within the mangrove area refers to the pixel change area in the remote sensing images in the second time series remote sensing image set.

[0061] In one embodiment, step S30 includes the following steps:

[0062] S301. Input the tidal height data at the time of acquisition of each remote sensing image in the second time series remote sensing image set and the corresponding remote sensing image in the first time series remote sensing image set into the preset DECODER model, fit the harmonic time series, and obtain the tidal factor composite detection model:

[0063]

[0064] in, Let be the predicted value of the i-th analytical variable of a pixel in the remote sensing image at date x in the Julian calendar, where T is the number of days in a year, T = 365.25, k is the order of the harmonic components, and a 0,i Let a be the baseline value of the i-th analysis variable. k,i Let b be the first harmonic coefficient of the i-th analytical variable. k,i Let c be the second harmonic coefficient of the i-th analytical variable. 1,i Let c be the interannual variation coefficient of the i-th analytical variable. 2,i Let w be the tidal influence coefficient of the i-th analysis variable, and w be the interpolated tidal level corresponding to 10:00 AM on the Julian date x.

[0065] Based on the inventive concept of this application, those skilled in the art can combine common knowledge in the field to set several appropriate index parameters as the analysis variables to implement the mangrove post-disaster recovery detection method based on remote sensing data described in this application. For example, these parameters can be the Normalized Difference Vegetation Index (NDVI), the Modified Normalized Water Index (MNDW), or the three tassel cap (TC) components. Alternatively, they can be new index parameters formed by performing weighted allocation or combination transformation on several of the index parameters.

[0066] S302. Input each remote sensing image in the second time series remote sensing image set into the tidal factor composite detection model, perform change area identification, and obtain the disturbed area and stable area within the mangrove area.

[0067] Meanwhile, in one embodiment, the step S302 of "inputting each remote sensing image in the second time-series remote sensing image set into the tidal factor composite detection model and performing change area identification" includes:

[0068] S3021. Obtain the actual values ​​of the analysis variables of the pixels in the remote sensing images of the second time series remote sensing image set;

[0069] S3022. Input each remote sensing image in the second time series remote sensing image set into the tidal factor composite detection model to obtain the predicted value of the analysis variable of the corresponding pixel.

[0070] S3023. Subtract the predicted value from the actual value to obtain the residual value of the analysis variable. If the residual value of the analysis variable is greater than the preset change threshold, it is determined that the pixel has changed.

[0071] In addition, this application provides some steps that can be applied to the mangrove post-disaster recovery detection method based on remote sensing data, for verifying the data accuracy of the tidal factor composite detection model:

[0072] S303. The performance of the tidal factor composite detection model is evaluated using the five-fold cross-validation method. If the accuracy of the tidal factor composite detection model is greater than the preset evaluation standard, the tidal factor composite detection model is determined to have passed the data accuracy verification.

[0073] In one embodiment, the original dataset is randomly divided into five mutually exclusive subsets. Each time, one subset is selected as the test set, and the remaining four subsets are used as the training set for training and validation. This process is repeated five times, with a different subset selected as the test set each time, ensuring that each subset has a chance to be used as the test set. Finally, the average of the five validation results is used as the final performance evaluation of the model.

[0074] For step S40, based on the disturbed areas within the mangrove forest area, the remote sensing images in the second time series remote sensing image set are classified using a random forest model to obtain pre-disaster mangrove remote sensing images, post-disaster mangrove remote sensing images, and mangrove remote sensing images during the recovery period.

[0075] The pre-disaster mangrove remote sensing images are remote sensing images of mangrove areas before the natural disaster; the post-disaster mangrove remote sensing images are remote sensing images of mangrove areas after the natural disaster; and the recovery period mangrove remote sensing images are remote sensing images of mangrove areas that have entered a natural recovery state after the natural disaster.

[0076] In one embodiment, those skilled in the art can use color information (such as values ​​of red, green, and blue bands) in remote sensing images as color features, and / or use methods such as gray-level co-occurrence matrix to extract texture features from remote sensing images, and / or extract shape features (such as area, perimeter, shape index, etc.) of mangrove areas to perform feature extraction in order to continue implementing step S40 of the method described in this application; those skilled in the art can obtain the time point when the corresponding mangrove area encountered a natural disaster and the end of the natural disaster by executing step S40.

[0077] In one embodiment, those skilled in the art can verify the classification results of the random forest model through methods such as visual interpretation or field surveys to ensure the accuracy of remote sensing image classification.

[0078] For step S50, based on the pre-disaster mangrove remote sensing image and the post-disaster mangrove remote sensing image, a disaster damage analysis is performed on the mangrove area. Combined with the preset disaster damage early warning standard, it is determined whether the degree of damage to the mangrove area exceeds the preset damage early warning threshold. If so, based on the recovery period mangrove remote sensing image and the post-disaster mangrove remote sensing image, a recovery year analysis is performed on the mangrove area to obtain the recovery year parameter of the mangrove area.

[0079] In one embodiment, step S50 includes the following steps:

[0080] S501. Calculate the disaster damage in the mangrove area according to the disaster damage calculation formula:

[0081] ΔNDVI=(NDVI Rpost -NDVI Cpost )-(NDVI Rpre -NDVI Cpre );

[0082] Wherein, △NDVI is the normalized difference in vegetation index (NDVI) of the mangrove region. Rpost NDVI is the normalized differential vegetation index (NDVI) of the mangrove area at a post-disaster time point during a disaster year. Cpost NDVI is the normalized differential vegetation index (NDVI) for the mangrove area at the same post-disaster time point during a normal year. Rpre The Normalized Difference Vegetation Index (NDVI) for the mangrove region at a pre-disaster time point during a disaster year. Cpre The normalized vegetation index is the same pre-disaster time point in a normal year for the mangrove area.

[0083] S502. When △NDVI is greater than 0, it is determined that the mangrove area is undamaged; when △NDVI is less than 0, it is determined that the degree of damage to the mangrove area exceeds the preset damage warning threshold.

[0084] S503. When △NDVI is less than 0, perform a restoration years analysis on the target mangrove area according to the restoration years calculation formula:

[0085]

[0086] Where Y2R is the restoration years parameter of the target mangrove area, ρ mangrove To predict the annual total value of the mangrove region using a period of time prior to the disturbance, ρdieback To predict the annual total value of the mangrove region using a period of time following the disturbance, x july,1st This refers to the annual overall value for the mangrove area as of July 1st of the year in which the disturbance occurred. The rate of mangrove death after disturbance is denoted as T, and the annual recovery rate of the mangrove area is denoted as T.

[0087] Wherein, the annual total value is a comprehensive description of the analytical variables in step S301 over a one-year time period. In one embodiment, the annual total value is the annual average value of the analytical variables in step S301.

[0088] For step S60, restoration intervention is carried out on the mangrove area according to the restoration years parameter of the mangrove area and the preset restoration intervention standard.

[0089] In one embodiment, those skilled in the art can intervene in the restoration of the mangrove area through habitat restoration (including site preparation and habitat creation, creating interspersed high, medium and low mudflats through terrain modification to provide growth environments for different vegetation communities; preserving appropriate exposed mudflats to form an ecological pattern in which mangroves, mudflats and tidal channels are interspersed, etc.), seedling selection and afforestation (selecting suitable mangrove species according to the climate conditions, substrate type, mudflat elevation, salinity and hydrodynamic conditions of the restoration site; ensuring appropriate initial planting spacing for newly planted mangroves to ensure sufficient number of plants per hectare; selecting seedlings with well-developed root systems and robust growth for planting to improve the survival rate of afforestation, etc.), and stock enhancement (using artificial methods to directly release or transfer eggs, larvae or adults of aquatic organisms into natural waters such as the ocean and mudflats to restore or increase the population and improve and optimize the community structure of the water area).

[0090] In one embodiment, those skilled in the art can use the normalized vegetation index (NDI) of the mangrove area as an intervention reference to determine whether artificial intervention needs to continue. If the NDI of the mangrove area recovers to 80% of the pre-disaster level, artificial intervention can be stopped.

[0091] Example 2

[0092] Please refer to Figure 4 This application also provides a mangrove post-disaster recovery detection system based on remote sensing data to implement the steps of the mangrove post-disaster recovery detection method based on remote sensing data described in the above embodiments. The mangrove post-disaster recovery detection system based on remote sensing data includes: a first time-series remote sensing image acquisition unit 1001, a second time-series remote sensing image acquisition unit 1002, a disturbance area identification and detection unit 1003, a remote sensing image classification unit 1004, a disaster damage and recovery years analysis unit 1005, and a recovery intervention unit 1006.

[0093] The first time-series remote sensing image acquisition unit 1001 is used to acquire a first time-series remote sensing image set of the target area, wherein the first time-series remote sensing image set includes several time-series remote sensing images of the target area taken at different time nodes.

[0094] The second time-series remote sensing image acquisition unit 1002 is used to extract the mangrove area in each remote sensing image in the first time-series remote sensing image set by using a sub-pixel analysis method, so as to obtain the second time-series remote sensing image set.

[0095] The disturbance region identification and detection unit 1003 is used to construct a continuous change detection model, identify the change region of each remote sensing image in the second time series remote sensing image set, and obtain the disturbance region and stable region in the mangrove area, wherein the disturbance region in the mangrove area is the pixel change region in the remote sensing image in the second time series remote sensing image set.

[0096] The remote sensing image classification unit 1004 is used to classify each remote sensing image in the second time series remote sensing image set based on the disturbed area in the mangrove area using a random forest model, to obtain pre-disaster mangrove remote sensing images, post-disaster mangrove remote sensing images, and recovery period mangrove remote sensing images. The recovery period mangrove remote sensing images are remote sensing images of mangrove areas that have entered a natural recovery state after encountering a natural disaster.

[0097] The disaster damage and recovery years analysis unit 1005 is used to perform disaster damage analysis on the mangrove area based on the pre-disaster mangrove remote sensing image and the post-disaster mangrove remote sensing image, and to determine whether the degree of damage to the mangrove area exceeds the preset damage warning threshold based on the preset disaster damage warning standard; if so, it performs recovery years analysis on the mangrove area based on the recovery period mangrove remote sensing image and the post-disaster mangrove remote sensing image to obtain the recovery years parameter of the mangrove area.

[0098] The restoration intervention unit 1006 is used to perform restoration intervention on the mangrove area according to the restoration years parameter of the mangrove area and the preset restoration intervention standard.

[0099] It should be noted that the above embodiment of the mangrove post-disaster recovery detection system based on remote sensing data is only illustrated by the division of the above functional modules when implementing a mangrove post-disaster recovery detection method based on remote sensing data. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0100] Furthermore, the mangrove post-disaster recovery detection system based on remote sensing data provided in the above embodiments and the mangrove post-disaster recovery detection method based on remote sensing data in Embodiment 1 belong to the same concept. The implementation process is detailed in the method embodiment, namely Embodiment 1, and will not be repeated here.

[0101] The embodiments described above are merely examples of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and this application also intends to include these modifications and variations.

Claims

1. A method for detecting post-disaster recovery of mangroves based on remote sensing data, comprising the following steps: Acquire a first time-series remote sensing image set of the target area, wherein the first time-series remote sensing image set includes several time-series remote sensing images of the target area taken at different time nodes; The mangrove areas in each remote sensing image in the first time series remote sensing image set are extracted by sub-pixel analysis method to obtain the second time series remote sensing image set. A continuous change detection model is constructed to identify change regions in each remote sensing image in the second time-series remote sensing image set, obtaining disturbed and stable regions within the mangrove area. The disturbed regions within the mangrove area are pixel-change regions in the remote sensing images of the second time-series remote sensing image set. This includes: inputting tidal height data at the time of acquisition from each remote sensing image in the second time-series remote sensing image set and the corresponding remote sensing image in the first time-series remote sensing image set into a preset DECODER model, fitting a harmonic time series, and obtaining a tidal factor composite detection model. in, For Julian dates x The first pixel in the remote sensing image at that time i The predicted values ​​of the analytical variables, where T is the number of days in a year, T=365.

25. k Let be the order of the harmonic components. a 0,i Let i be the baseline value of the i-th analysis variable. a k,i For the first i The first harmonic coefficient of the analysis variable b k,i For the first i The second harmonic coefficient of the analysis variable c 1,i For the first i The interannual variation coefficients of the analytical variables, c 2,i For the first i The tidal influence coefficients of the analytical variables, w For Julian dates x The tidal level interpolation corresponding to 10:00 AM is performed; each remote sensing image in the second time series remote sensing image set is input into the tidal factor composite detection model to perform change area identification and obtain the disturbed area and stable area within the mangrove area; Based on the disturbed areas within the mangrove region, the remote sensing images in the second time series remote sensing image set are classified using a random forest model to obtain pre-disaster mangrove remote sensing images, post-disaster mangrove remote sensing images, and recovery period mangrove remote sensing images. The recovery period mangrove remote sensing images are remote sensing images of mangrove regions that have entered a natural recovery state after encountering a natural disaster. Based on the pre-disaster and post-disaster remote sensing images of mangroves, a disaster damage analysis is performed on the mangrove area. Combined with a preset disaster damage early warning standard, it is determined whether the degree of damage to the mangrove area exceeds a preset damage early warning threshold. If so, based on the recovery period mangrove remote sensing images and the post-disaster mangrove remote sensing images, a recovery year analysis is performed on the mangrove area to obtain the recovery year parameter of the mangrove area. Restoration interventions are carried out on the mangrove area based on the restoration years parameter and preset restoration intervention standards.

2. The method for detecting post-disaster recovery of mangroves based on remote sensing data according to claim 1, the step of acquiring the first time-series remote sensing image set of the target area includes: Acquire several remote sensing images of the target area taken at different time points; The aforementioned remote sensing images are preprocessed, including geometric correction, radiometric correction, image registration, and atmospheric correction. Based on the recording time of the remote sensing images of the target area taken at different time points, the preprocessed remote sensing images of each target area taken at different time points are sorted by time to obtain the first time series remote sensing image set of the target mangrove area.

3. The method for detecting post-disaster recovery of mangroves based on remote sensing data according to claim 1, the step of extracting mangrove areas from each remote sensing image in the first time-series remote sensing image set through sub-pixel analysis to obtain the second time-series remote sensing image set includes: Hybrid pixel decomposition is performed on each remote sensing image in the first time-series remote sensing image set to obtain the abundance information of each land feature endmember in each pixel of the remote sensing images in the first time-series remote sensing image set, thus obtaining the hybrid pixel decomposition result. The land feature endmember is a pure land feature in the target area; the land feature endmember includes at least mangroves and unrelated land features; the abundance information is the proportion of the land feature endmember in a single pixel. Based on the hybrid pixel decomposition results, the pixel exchange method is used to predict the spatial distribution information of mangroves in each pixel of the remote sensing images in the first time series remote sensing image set, and a sub-pixel land cover distribution map of mangroves is obtained. Based on the sub-pixel land cover distribution map of the mangrove forest, and combined with the preset remote sensing image features of the mangrove forest, a supervised classification method and / or an unsupervised classification method are used to extract the mangrove forest areas from each remote sensing image in the first time series remote sensing image set, obtain the corresponding number of remote sensing images of the mangrove forest areas, and obtain the second time series remote sensing image set. The preset remote sensing image features of the mangrove forest include at least the spectral features, texture features, and structural features of the mangrove forest.

4. The method for detecting post-disaster recovery of mangroves based on remote sensing data according to claim 3, comprising the steps of performing hybrid pixel decomposition on each remote sensing image in the first time-series remote sensing image set to obtain the abundance information of each land cover endmember in each pixel of the remote sensing image in the first time-series remote sensing image set, and obtaining the hybrid pixel decomposition result, including: The pixels in the remote sensing images of the first time-series remote sensing image set are set as a linear combination of the spectral features of the ground object endmembers: ； Where r is the spectral reflectance vector of the pixel; E is the end-member spectral matrix of the ground features, and each column of the end-member spectral matrix represents the spectral characteristics of an end-member of the ground features. α An abundance vector representing the proportion of various land cover endmembers in a pixel; ε This is a preset error term; Using the fully constrained least squares method, abundance inversion is performed on the linear combination and the reconstruction error is minimized to obtain the first problem to be solved: ； in, α i For the corresponding number i The abundance vector of the proportion of each land feature endmember in a pixel. m The number of end-members of the ground features; Construct the abundance Lagrangian function for the first problem to be solved: ； Where, λ and μ These are the pre-defined Lagrange multipliers; The abundance Lagrangian function is solved to obtain the abundance information of each land feature end-member in the target area in each pixel of the remote sensing image in the first time series remote sensing image set, and the mixed pixel decomposition result is obtained.

5. The method for detecting post-disaster recovery of mangroves based on remote sensing data according to claim 4, characterized in that, It also includes a step for verifying the data accuracy of the tidal factor composite detection model: The performance of the tidal factor composite detection model is evaluated using a five-fold cross-validation method. If the accuracy of the tidal factor composite detection model is greater than the preset evaluation standard, the tidal factor composite detection model is determined to have passed the data accuracy verification. The steps of inputting each remote sensing image from the second time-series remote sensing image set into the tidal factor composite detection model to perform change area identification include: Obtain the actual values ​​of the analytical variables of the pixels in the remote sensing images of the second time series remote sensing image set; Each remote sensing image in the second time series remote sensing image set is input into the tidal factor composite detection model to obtain the predicted value of the analysis variable of the corresponding pixel; Subtracting the predicted value from the actual value yields the residual value of the analysis variable. If the residual value of the analysis variable is greater than a preset change threshold, it is determined that the pixel has changed.

6. The method for detecting post-disaster recovery of mangroves based on remote sensing data according to claim 4, wherein the step of performing disaster damage analysis on the mangrove area based on the pre-disaster mangrove remote sensing image and the post-disaster mangrove remote sensing image includes: Based on the disaster damage calculation formula, the disaster damage to the mangrove area was calculated: ； Wherein, △NDVI represents the normalized difference in vegetation index (NDVI) of the mangrove region. NDVI Rpost The normalized vegetation index (NDI) of the mangrove area at a post-disaster time point during a disaster year. NDVI Cpost The normalized vegetation index (NDI) represents the mangrove area at the same post-disaster time point during a normal year. NDVI Rpre The normalized vegetation index (NDI) of the mangrove area at a pre-disaster time point during the disaster year. NDVI Cpre The normalized vegetation index is the same pre-disaster time point in a normal year for the mangrove area.

7. The method for detecting post-disaster recovery of mangroves based on remote sensing data according to claim 6, in conjunction with a preset disaster damage warning standard, the step of determining whether the degree of damage to the mangrove area exceeds a preset damage warning threshold includes: When △NDVI is greater than 0, it is determined that the mangrove area is undamaged; When △NDVI is less than 0, it is determined that the damage level of the mangrove area exceeds the preset damage warning threshold.

8. The method for detecting post-disaster recovery of mangroves based on remote sensing data according to claim 7, wherein the step of performing a recovery year analysis on the mangrove area based on the remote sensing image of the mangroves during the recovery period and the remote sensing image of the mangroves after the disaster, to obtain the recovery year parameter of the mangrove area, includes: When △NDVI is less than 0, the restoration years of the mangrove area are analyzed according to the restoration years calculation formula: in, Y 2 R The parameter representing the number of years the mangrove area has been restored is given. ρ mangrove To predict the annual total value of the mangrove area using a period of time prior to the disturbance, ρ dieback To predict the annual total value of the mangrove area using a period of time following the disturbance, x july,1st This refers to the annual overall value for the mangrove area as of July 1st of the year in which the disturbance occurred. The rate of mangrove death after disturbance is denoted as T, and the annual recovery rate of the mangrove area is denoted as T.

9. A mangrove post-disaster recovery monitoring system based on remote sensing data, characterized in that, The mangrove post-disaster recovery monitoring system includes: The first time-series remote sensing image acquisition unit is used to acquire a first time-series remote sensing image set of the target area, wherein the first time-series remote sensing image set includes several time-series remote sensing images of the target area taken at different time nodes. The second time-series remote sensing image acquisition unit is used to extract the mangrove area from each remote sensing image in the first time-series remote sensing image set through sub-pixel analysis method, so as to obtain the second time-series remote sensing image set. The disturbance region identification and detection unit is used to construct a continuous change detection model, identify change regions in each remote sensing image in the second time-series remote sensing image set, and obtain the disturbance regions and stable regions within the mangrove area. The disturbance regions within the mangrove area are the pixel change regions in the remote sensing images of the second time-series remote sensing image set. This includes: inputting the tidal height data at the time of acquisition of each remote sensing image in the second time-series remote sensing image set and the corresponding remote sensing image in the first time-series remote sensing image set into a preset DECODER model, fitting a harmonic time series, and obtaining a tidal factor composite detection model. in, For Julian dates x The first pixel in the remote sensing image at that time i The predicted values ​​of the analytical variables, where T is the number of days in a year, T=365.

25. k Let be the order of the harmonic components. a 0,i Let i be the baseline value of the i-th analysis variable. a k,i For the first i The first harmonic coefficient of the analysis variable b k,i For the first i The second harmonic coefficient of the analysis variable c 1,i For the first i The interannual variation coefficients of the analytical variables, c 2,i For the first i The tidal influence coefficients of the analytical variables, w For Julian dates x The tidal level interpolation corresponding to 10:00 AM is performed; each remote sensing image in the second time series remote sensing image set is input into the tidal factor composite detection model to perform change area identification and obtain the disturbed area and stable area within the mangrove area; The remote sensing image classification unit is used to classify each remote sensing image in the second time series remote sensing image set based on the disturbed area within the mangrove area using a random forest model, to obtain pre-disaster mangrove remote sensing images, post-disaster mangrove remote sensing images, and recovery period mangrove remote sensing images. The recovery period mangrove remote sensing images are remote sensing images of mangrove areas that have entered a natural recovery state after encountering a natural disaster. The damage and recovery years analysis unit is used to analyze the damage to the mangrove area based on the pre-disaster and post-disaster remote sensing images of the mangroves, and to determine whether the damage to the mangrove area exceeds the preset damage warning threshold based on the preset damage warning standard. If so, the unit analyzes the recovery years of the mangrove area based on the recovery period remote sensing images of the mangroves and the post-disaster remote sensing images of the mangroves to obtain the recovery years parameter of the mangrove area. The restoration intervention unit is used to perform restoration interventions on the mangrove area based on the restoration years parameter of the mangrove area and preset restoration intervention standards.

Images