Method for evaluating damage and recovery of coast vegetation after extreme storm and related equipment

Abstract

The invention discloses a coastal vegetation damage and recovery evaluation method after an extreme storm and related equipment, and relates to the technical field of data processing, and the method comprises the steps: constructing a synchronous grid stack; four states of storm, moisture, scattering and damage are mapped into a node set, and a pixel-level community lag causal graph is constructed according to each node; solving according to the pixel-level community lag causal graph to obtain a water film scattering expected value, and determining a net damage sequence by using the water film scattering expected value; dividing to obtain ecological partitions of different environments, and calculating a scattering value in each ecological partition so as to obtain a natural fluctuation envelope; and packaging the net damage sequence, the natural fluctuation envelope, the edge weight of pixel-level moisture and scattering in the pixel-level community lag causal diagram and the confidence factor into match input and inputting the match input into a parameter player, further outputting various parameters of vegetation damage and recovery, and then determining the peak damage rate, the half-recovery duration and the complete recovery duration of the vegetation. According to the invention, the evaluation accuracy of vegetation damage and recovery can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method for assessing coastal vegetation damage and recovery after extreme storms and related equipment. Background Art

[0002] Existing technical approaches for monitoring coastal vegetation damage after typhoon landfall often rely on the parallel use of optical and microwave remote sensing. A common approach is to first calculate vegetation indices such as NDVI and EVI using multi-temporal Sentinel-2 or Landsat. Pre- and post-disaster imagery is then differentiated to delineate suspected damaged areas. C-band Sentinel-1 dual-polarization scattering intensity is then used to compensate for cloud cover and imaging time differences. Optical thresholds and radar amplitude ratios are then superimposed to generate a preliminary damage distribution map. Both image types require sub-pixel geometric correction, and terrain correction and unified projection are used to achieve a comparable spatial reference.

[0003] Existing technologies generally lack a unified spatiotemporal reference and traceable causal chain, and are unable to simultaneously characterize the multi-level lag relationship between storm intensity, surface moisture, radar scattering, and vegetation damage at the pixel level. They also lack an explicit stripping mechanism for water film scattering interference, making it difficult to obtain reliable quantitative damage-recovery indicators in high-humidity backgrounds. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to propose a method and related equipment for assessing coastal vegetation damage and recovery after extreme storms, so as to improve the accuracy of vegetation damage and recovery assessment.

[0005] To achieve the above objectives, one aspect of an embodiment of the present application provides a method for assessing coastal vegetation damage and recovery after an extreme storm, the method comprising the following steps: The storm intensity, moisture, VV polarimetric scattering, VH polarimetric scattering, and damage index are integrated in the common pixel domain and time domain to construct a synchronized grid stack. Mapping the four states of storm, moisture, scattering and damage in the synchronized grid stack into a node set, and constructing a pixel-level community lag causal graph based on each node in the node set; Obtaining a water film scattering expectation value according to the pixel-level community lag causal graph, and determining a net damage sequence using the water film scattering expectation value; Dividing the synchronized grid stack into ecological zones of different environments, calculating the scattering value in each ecological zone to obtain a natural fluctuation envelope; The net damage sequence, the natural fluctuation envelope, the edge weights of pixel-level moisture and scattering in the pixel-level community lag causal graph, and the confidence factor are encapsulated as game inputs and input into a parameter player, and the parameter player is used to output various parameters of vegetation damage and recovery; The peak damage rate, half-recovery time and full recovery time of the vegetation are output according to the multiple parameters of vegetation damage and recovery.

[0006] In some embodiments, integrating storm intensity, moisture, VV polarization scattering, VH polarization scattering, and damage index in a common pixel domain and a time domain to construct a synchronized grid stack comprises the following steps: Obtaining recorded data for the first set number of consecutive days before and after the storm's landfall date; wherein the recorded data includes dual-polarization synthetic aperture radar radar images, soil moisture grids, typhoon intensity grids, and damage records from field plots; Adding a unified daily time scale to the recorded data; performing orbit refinement, digital elevation model correction, radiometric calibration, and slant range projection on the radar image in sequence, and then converting and reprojecting the image into a unified reference system; Resampling the typhoon intensity grid and the soil moisture grid using terrain weights and wind field weights based on radar pixel resolution; wherein the weight factors are automatically assigned under the dual constraints of elevation difference and radial distance of the typhoon wind field; The time series is completed using the observation date of the field sample as the anchor point, and the same pixel is kept in the same geographical coordinate on all dates through rigid body correction; The quality label is set in combination with the daily tide level and the incident angle threshold, and then the storm intensity, moisture, VV polarization scattering, VH polarization scattering and damage index are integrated in the common pixel domain and time domain to obtain the synchronized grid stack.

[0007] In some embodiments, mapping the four states of storm, moisture, scattering, and damage in the synchronized grid stack to a node set comprises the following steps: A community response priori database is generated using typhoon cases over the years, and a community three-level lag window is obtained based on the community response priori database; According to the principles of pixel consistency, community consistency, and lag matching, the four states of storm, moisture, scattering, and damage in the synchronized grid stack are mapped to the node set; The step of constructing a pixel-level community hysteresis causal graph based on each of the nodes in the node set comprises the following steps: The forward edge weights corresponding to storm and moisture, moisture and scattering, and scattering and damage are initialized according to the lagged correlation coefficient, normalized in parallel, and then truncated according to the prior; Adding negative edges of the scattering and damage when the probability of tide level residue or foam coating meets the set conditions; Traversing the sliding window along the time axis, the weight of each edge is iteratively updated by gradient descent until the pixel residual variance is lower than the threshold, thereby obtaining the pixel-level community lag causal graph.

[0008] In some embodiments, obtaining the expected value of water film scattering according to the pixel-level community hysteresis causal graph comprises the following steps: Synchronously extracting the VV polarized scattering, the VH polarized scattering, the moisture, and the edge weights of moisture and scattering obtained according to the pixel-level community lag causal graph at each pixel; The expected value of water film scattering is obtained by combining the edge weights of moisture and scattering with moisture and scattering gain functions; The method of determining a net damage sequence by using the water film scattering expectation value comprises the following steps: Weighting the dual-polarization scattered energy and deducting the water film scattering expectation value to obtain the structural scattering residual; After applying a two-day window first-order Savitzky–Golay filter to the structural scattering residuals, the mean of the residuals for the second set number of days before the storm landfall is extracted as a baseline and normalized to obtain the net damage series.

[0009] In some embodiments, the step of obtaining ecological zones of different environments according to the synchronized grid stack division includes the following steps: Reading elevation, aspect, conductivity, and community type rasters by pixel alignment according to the synchronized grid stack to form an environmental vector; The environmental vectors are divided into the ecological zones of different environments by using progressive similarity splitting clustering of weighted Mahalanobis distance; Calculating the scattering value in each ecological zone to obtain the natural fluctuation envelope includes the following steps: Extract dual-polarization radar images of the same month as the target storm from the storm-free images in the most recent set number of years to form a storm-free control set; Calculating the 5th percentile and the 95th percentile of the scattering value on a calendar day basis in each ecological zone based on the no-storm control set to obtain the natural fluctuation envelope; Abnormal curves that have been manually verified in the natural fluctuation envelope are eliminated.

[0010] In some embodiments, the step of using the parameters to output various parameters of vegetation damage and restoration includes the following steps: Using the parameters, the player generates a preliminary simulation curve based on the damage peak, incubation period, and recovery rate; Calculating a weighted residual score based on the confidence level and envelope crossing situation of the initial version simulation curve using residual players; Using the parameters, the player modifies the parameters according to the vector-level rules until the change value of the weighted residual score is lower than the error threshold or reaches the preset maximum round, and outputs the recovery amplitude, latent period, recovery rate and residual index as the multiple parameters of the vegetation damage and recovery.

[0011] In some embodiments, the step of outputting the peak damage rate, half-recovery time, and full recovery time of vegetation based on the multiple parameters of vegetation damage and recovery comprises the following steps: Correcting the timelines of various parameters of vegetation damage and recovery based on field water level records; Mapping the various parameters of vegetation damage and recovery into a piecewise exponential curve, and normalizing the curve with the peak daily envelope amplitude to obtain the peak damage rate; Locate the date when the net damage drops to half of the peak value from the peak date to obtain the half-recovery time; When the net damage is below the upper limit of the envelope for the third consecutive set number of days, the full recovery time is recorded; The peak damage rate, the half recovery time, and the full recovery time are output.

[0012] To achieve the above objectives, another aspect of the present application provides a device for assessing coastal vegetation damage and recovery after an extreme storm, the device comprising: A data acquisition unit is used to integrate storm intensity, moisture, VV polarization scattering, VH polarization scattering and damage index in the common pixel domain and time domain to construct a synchronized grid stack; A graph construction unit is configured to map the four states of storm, moisture, scattering, and damage in the synchronized grid stack into a node set, and construct a pixel-level community lag causal graph based on each node in the node set; a damage determination unit, configured to obtain an expected value of water film scattering according to the pixel-level community hysteresis causal graph, and determine a net damage sequence using the expected value of water film scattering; A natural fluctuation envelope determination unit is configured to obtain ecological zones of different environments according to the synchronous grid stack division, and calculate a scattering value in each ecological zone to obtain a natural fluctuation envelope; A parameter fitting unit is used to encapsulate the net damage sequence, the natural fluctuation envelope, the edge weights of pixel-level moisture and scattering in the pixel-level community lag causal graph, and the confidence factor as game inputs and input them into a parameter player, and use the parameter player to output multiple parameters of vegetation damage and recovery; The damage and recovery evaluation unit is used to output the peak damage rate, half recovery time and full recovery time of vegetation according to the multiple parameters of vegetation damage and recovery.

[0013] To achieve the above-mentioned purpose, another aspect of an embodiment of the present application provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned method when executing the computer program.

[0014] To achieve the above-mentioned purpose, another aspect of an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above-mentioned method when executed by a processor.

[0015] The embodiments of the present application include at least the following beneficial effects: The present application provides a method and related equipment for assessing coastal vegetation damage and recovery after extreme storms. The present application scheme integrates storm intensity, moisture, VV polarization scattering, VH polarization scattering and damage index in a common pixel domain and a time domain to construct a synchronized grid stack; maps the four states of storm, moisture, scattering and damage in the synchronized grid stack into a node set, and constructs a pixel-level community lag causal graph based on each node in the node set; obtains the expected value of water film scattering based on the pixel-level community lag causal graph, and uses the expected value of water film scattering to determine the net damage sequence; obtains ecological zones of different environments based on the synchronized grid stack, calculates the scattering value in each ecological zone to obtain the natural fluctuation envelope; encapsulates the net damage sequence, natural fluctuation envelope, edge weights of pixel-level moisture and scattering in the pixel-level community lag causal graph, and confidence factors as game inputs and inputs them into a parameter player, and uses the parameter player to output multiple parameters of vegetation damage and recovery; outputs the peak damage rate, half-recovery time and full recovery time of vegetation based on the multiple parameters of vegetation damage and recovery. This application uses a storm-moisture-scattering synchronized grid stack to maintain the same step size and resolution for typhoon intensity, soil moisture, and dual-polarization scattering in the pixel domain, providing an equivalent space-time coordinate system for subsequent analysis, avoiding convolution errors caused by inconsistent resolution of multi-source field quantities, and thus improving the accuracy of vegetation damage and recovery assessments. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0017] Figure 1 Comparison diagrams of some embodiments provided in the present application and the prior art; Figure 2 A flow chart of a method for assessing coastal vegetation damage and restoration after an extreme storm provided in an embodiment of the present application; Figure 3 An example flow chart of a method for assessing coastal vegetation damage and restoration after an extreme storm provided in an embodiment of the present application; Figure 4 A processing flow chart of a cause-effect diagram baseline provided in an embodiment of the present application; Figure 5 A process flow chart of water film-structure demixing provided in an embodiment of the present application; Figure 6 A flowchart for constructing an ecological resilience baseline provided in an embodiment of the present application; Figure 7 A schematic diagram of the structure of a device for assessing coastal vegetation damage and recovery after an extreme storm provided in an embodiment of the present application; Figure 8 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are merely examples of devices and methods consistent with some aspects of the embodiments of the present application as detailed in the appended claims.

[0019] It will be understood that the terms "first", "second", etc. used in this application may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0020] The terms "at least one", "plurality", "each", "any", etc. used in this application include "at least one", "two" or more, "plurality" or "each", "any" or "any one", "each" or "any one" as used herein.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0022] Before describing the embodiments of the present application in detail, some of the related technologies involved in the embodiments of the present application are first described as follows: 1. Parameter Player: In the game framework, the parameter player is responsible for generating and updating the simulated damage-recovery curve. Its core task is to perform vectorized adjustments to the three parameters (initial damage amplitude A, incubation period τ, and recovery rate k) based on the current fitting residuals, so that the next round of simulation is closer to the observed series.

[0023] 2. Residual player: The residual player plays the role of "finding faults". By calculating the difference between the current simulation curve and the actual net damage sequence, the sliding window accumulation method is used to locate the maximum continuous residual segment, and a weighted residual score is formed based on this. It provides feedback on the direction and magnitude of correction to the parameter player.

[0024] 3. Confidence Factor: The confidence factor (Cp) is derived from the pixel-level water scattering edge weight mapping and is a quantitative measure of the reliability of the observation at that pixel. During the residual weighting phase, Cp is used to adjust the weights between the residual score and the parameter update, ensuring that high-confidence pixels converge faster and have a greater impact in the game.

[0025] 4. Sliding Window Accumulation Method: Also known as the sliding window accumulation method, this involves moving a fixed-length window over the residual sequence, accumulating the absolute values ​​of the residuals within the window, and finding the interval with the maximum sum. This method can highlight periods of concentrated consecutive errors, helping residual analysts pinpoint the curve segments most in need of correction.

[0026] 5. Weighted Residual Score: The residual player calculates a weighted residual score (Sr) based on the maximum residual segment located, combined with a confidence factor and an out-of-bounds penalty. Sr reflects both the degree to which the fit deviates from the observation and the priority of that deviation within the overall game.

[0027] 6. Game Iteration and Termination Conditions: A "game" refers to the entire process of alternating optimization between the parameter player and the residual player on the same pixel. After each round, if both the residual score and residual variance fall below a preset threshold, or if the maximum number of iterations is reached, the pixel fit is considered converged and the iteration terminates.

[0028] To reduce false detections caused by tidal range and muddy water reflections, some existing technologies incorporate the MODIS water color index in the preprocessing stage to remove highly turbid water surfaces, while also relying on tide forecast models or measured water level records to mask flooded time slices. On the radar side, transient specular scattering pixels are simply disabled. All indicators are then resampled to a regular 30m or 10m grid, and digital elevation models are used to adjust for slope and shadow effects, ensuring consistent terrain reference across multiple data sources.

[0029] Time series analysis often employs exponential smoothing or the BreaksForAdditiveSeasonandTrend model, capturing sudden drops in the annual vegetation index sequence and treating typhoons as short-term negative jumps. The recovery slope is then used to fit the ecological resilience curve and calculate the half-recovery time. Because this approach assumes a single index can describe both damage and recovery, post-regression correction is often required, combining the field sample breakage rate with the results to improve numerical accuracy.

[0030] On the other hand, some studies, drawing on the physical scattering mechanism, use the coherence coefficient of dual-polarization radar to separate volume scattering from specular scattering and infer the direction of the fall. Other work, based on the coherence variation of interferometric SAR, interprets the decrease in coherence as structural damage. These methods require a sufficiently short imaging interval to maintain interferometric quality and are susceptible to rapid moisture fluctuations. Therefore, they are often combined with SMAP or ASCAT soil moisture grids for regression denoising.

[0031] In recent years, machine learning approaches have been explored. Researchers compile wind speed fields, tide levels, soil moisture, radar scattering, and optical indices to construct pixel-level feature vectors, which are then fed into random forests or convolutional neural networks to output damage probabilities or levels. Training relies on historical typhoon samples, while inference utilizes daily satellite quick-scan imagery to achieve near-real-time assessments. This approach offers greater flexibility in integrating high-dimensional features, but its interpretability is limited, requiring continuous replenishment of annotated data to maintain generalization capabilities.

[0032] Problems with existing technologies: Current pre- and post-disaster differential methods often rely on a simple superposition of optical vegetation indices and radar amplitudes. Capturing temporal features often relies on empirical methods based on single-point dips and exponential rebounds. They lack independent characterization of the rapidly changing soil moisture content and free water film on leaves in the days following landfall, making it difficult to distinguish between the distinct physical signals of "high-humidity enhanced scattering" and "structural breakage attenuation." The same pixel often exhibits increased scattering during the water film effect phase but correspondingly decreased scattering during the breakage phase. Using a uniform threshold for processing in existing models can easily lead to both false and missed detections.

[0033] Although the idea of ​​separating body scattering by interferometric coherence or dual-polarization coherence coefficient is closer to the scattering mechanism, it is highly sensitive to imaging interval, moisture disturbance after rain, and interferometric baseline stability. Once a long baseline or a drastic change in the geoelectric constant after heavy rain occurs, the coherence drops significantly, making the result lack spatiotemporal continuity.

[0034] Data-driven approaches, such as machine learning, demonstrate flexibility in integrating multi-source features, but they rely heavily on the completeness of historical typhoon annotations. Model performance degrades dramatically when training data is insufficient or ecological communities change. Furthermore, black-box decision-making struggles to incorporate prior knowledge about typhoon intensity, tide levels, and microtopography, resulting in a lack of interpretable physical coherence in the output.

[0035] Existing technologies generally lack a unified spatiotemporal reference and traceable causal chain, and are unable to simultaneously characterize the multi-level lag relationship between storm intensity, surface moisture, radar scattering, and vegetation damage at the pixel level. They also lack an explicit stripping mechanism for water film scattering interference, making it difficult to obtain reliable quantitative damage-recovery indicators in high-humidity backgrounds.

[0036] The embodiments of the present application aim to address the technical problem of difficulty in accurately extracting the true structural damage of coastal vegetation and its subsequent recovery dynamics in the environment of high humidity and strong tidal disturbance after the landfall of an extreme storm. The specific goal is to establish an interpretable, quantifiable storm-moisture-scattering causal chain with community-specific hysteresis at the pixel scale, and on this basis, to strip away the water film scattering gain to obtain a temporally continuous and physically consistent vegetation damage-recovery indicator.

[0037] For example, some embodiments of the present application are compared with the prior art. Figure 1 shown.

[0038] The embodiment of the present application provides a method for assessing damage and recovery of coastal vegetation after an extreme storm and related equipment, and relates to the field of data processing technology. The embodiment of the present application provides a method for assessing damage and recovery of coastal vegetation after an extreme storm and related equipment, which can be applied to a terminal, can also be applied to a server, and can also be software running in a terminal or a server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and a car terminal, etc., but is not limited to this; the server side can be configured as an independent physical server, or as a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network; the software can be an application that implements a method for assessing damage and recovery of coastal vegetation after an extreme storm, etc., but is not limited to the above forms.

[0039] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments in which tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0040] Reference Figure 2 The present application embodiment provides a method for assessing coastal vegetation damage and restoration after an extreme storm. The method may include but is not limited to steps S200 to S250, as follows: S200: Integrate storm intensity, moisture, VV polarimetric scattering, VH polarimetric scattering, and damage index in the common pixel and time domains to construct a synchronized grid stack. S210: Mapping the four states of storm, moisture, scattering, and damage in the synchronized grid stack into a node set, and constructing a pixel-level community lag causal graph based on each node in the node set; S220: Obtaining a water film scattering expectation value according to the pixel-level community lag causal graph, and determining a net damage sequence using the water film scattering expectation value; S230: obtaining ecological zones of different environments according to the synchronized grid stack division, calculating a scattering value in each of the ecological zones to obtain a natural fluctuation envelope; S240: Encapsulating the net damage sequence, the natural fluctuation envelope, the edge weights of pixel-level moisture and scattering in the pixel-level community lag causal graph, and the confidence factor as game inputs and inputting them into a parameter player, and using the parameter player to output multiple parameters of vegetation damage and recovery; S250: Outputting the peak damage rate, half-recovery time, and full recovery time of the vegetation according to the multiple parameters of vegetation damage and recovery.

[0041] Optionally, integrating storm intensity, moisture, VV polarization scattering, VH polarization scattering, and damage index in the common pixel domain and the time domain to construct a synchronized grid stack comprises the following steps: Obtaining recorded data for the first set number of consecutive days before and after the storm's landfall date; wherein the recorded data includes dual-polarization synthetic aperture radar radar images, soil moisture grids, typhoon intensity grids, and damage records from field plots; Adding a unified daily time scale to the recorded data; performing orbit refinement, digital elevation model correction, radiometric calibration, and slant range projection on the radar image in sequence, and then converting and reprojecting the image into a unified reference system; Resampling the typhoon intensity grid and the soil moisture grid using terrain weights and wind field weights based on radar pixel resolution; wherein the weight factors are automatically assigned under the dual constraints of elevation difference and radial distance of the typhoon wind field; The time series is completed using the observation date of the field sample as the anchor point, and the same pixel is kept in the same geographical coordinate on all dates through rigid body correction; The quality label is set in combination with the daily tide level and the incident angle threshold, and then the storm intensity, moisture, VV polarization scattering, VH polarization scattering and damage index are integrated in the common pixel domain and time domain to obtain the synchronized grid stack.

[0042] Optionally, mapping the four states of storm, moisture, scattering, and damage in the synchronized grid stack into a node set comprises the following steps: A community response priori database is generated using typhoon cases over the years, and a community three-level lag window is obtained based on the community response priori database; According to the principles of pixel consistency, community consistency, and lag matching, the four states of storm, moisture, scattering, and damage in the synchronized grid stack are mapped to the node set; The step of constructing a pixel-level community hysteresis causal graph based on each of the nodes in the node set comprises the following steps: The forward edge weights corresponding to storm and moisture, moisture and scattering, and scattering and damage are initialized according to the lagged correlation coefficient, normalized in parallel, and then truncated according to the prior; Adding negative edges of the scattering and damage when the probability of tide level residue or foam coating meets the set conditions; Traversing the sliding window along the time axis, the weight of each edge is iteratively updated by gradient descent until the pixel residual variance is lower than the threshold, thereby obtaining the pixel-level community lag causal graph.

[0043] Optionally, obtaining the expected value of water film scattering according to the pixel-level community hysteresis causal graph comprises the following steps: Synchronously extracting the VV polarized scattering, the VH polarized scattering, the moisture, and the edge weights of moisture and scattering obtained according to the pixel-level community lag causal graph at each pixel; The expected value of water film scattering is obtained by combining the edge weights of moisture and scattering with moisture and scattering gain functions; The method of determining a net damage sequence by using the water film scattering expectation value comprises the following steps: Weighting the dual-polarization scattered energy and deducting the water film scattering expectation value to obtain the structural scattering residual; After applying a two-day window first-order Savitzky–Golay filter to the structural scattering residuals, the mean of the residuals for the second set number of days before the storm landfall is extracted as a baseline and normalized to obtain the net damage series.

[0044] Optionally, obtaining ecological zones of different environments according to the synchronized grid stack division comprises the following steps: Reading elevation, aspect, conductivity, and community type rasters by pixel alignment according to the synchronized grid stack to form an environmental vector; The environmental vectors are divided into the ecological zones of different environments by using progressive similarity splitting clustering of weighted Mahalanobis distance; Calculating the scattering value in each ecological zone to obtain the natural fluctuation envelope includes the following steps: Extract dual-polarization radar images of the same month as the target storm from the storm-free images in the most recent set number of years to form a storm-free control set; Calculating the 5th percentile and the 95th percentile of the scattering value on a calendar day basis in each ecological zone based on the no-storm control set to obtain the natural fluctuation envelope; Abnormal curves that have been manually verified in the natural fluctuation envelope are eliminated.

[0045] Optionally, the step of using the parameter player to output various parameters of vegetation damage and restoration includes the following steps: Using the parameters, the player generates a preliminary simulation curve based on the damage peak, incubation period, and recovery rate; Calculating a weighted residual score based on the confidence level and envelope crossing situation of the initial version simulation curve using residual players; Using the parameters, the player modifies the parameters according to the vector-level rules until the change value of the weighted residual score is lower than the error threshold or reaches the preset maximum round, and outputs the recovery amplitude, latent period, recovery rate and residual index as the multiple parameters of the vegetation damage and recovery.

[0046] Optionally, outputting the peak damage rate, half-recovery time, and full recovery time of vegetation based on the multiple parameters of vegetation damage and recovery comprises the following steps: Correcting the timelines of various parameters of vegetation damage and recovery based on field water level records; Mapping the various parameters of vegetation damage and recovery into a piecewise exponential curve, and normalizing the curve with the peak daily envelope amplitude to obtain the peak damage rate; Locate the date when the net damage drops to half of the peak value from the peak date to obtain the half-recovery time; When the net damage is below the upper limit of the envelope for the third consecutive set number of days, the full recovery time is recorded; The peak damage rate, the half recovery time, and the full recovery time are output.

[0047] Next, some optional embodiments of the present application will be introduced and explained in detail with reference to specific application examples.

[0048] Reference Figure 3 This embodiment provides a method for evaluating coastal vegetation damage and recovery after extreme storms based on scenario-coupled causal unmixing and enhanced game optimization, including the following steps S1 to S6: S1. Construct a synchronized storm-moisture-scattering grid stack, specifically: obtain dual-polarization synthetic aperture radar imagery, passive microwave soil moisture raster, typhoon center intensity log, and field sample damage records for thirty consecutive days before and after the storm landfall date, and attach a unified daily time scale to all data; perform orbit refinement, digital elevation model correction, radiometric calibration, and slant range projection conversion on the radar imagery, and then reproject it to a unified reference system; perform a joint topographic weight-wind field weight resampling of the typhoon intensity raster and soil moisture raster based on the radar pixel resolution, where the weight factor is automatically assigned under the dual constraints of elevation difference and radial distance of the typhoon wind field; complete the time series using the sample observation date as the anchor point, and use rigid body correction to ensure that the same pixel maintains consistent geographic coordinates on all dates; set quality labels based on daily tide level and incident angle thresholds, and finally integrate storm intensity, moisture, VV polarization scattering, VH polarization scattering, and damage index in the common pixel domain and time domain to obtain a synchronized grid stack.

[0049] S2. Construct a pixel-level community lag causal diagram, specifically: use typhoon cases from previous years to generate a community response prior library to obtain a community-specific three-level lag window; map the four states of storm, moisture, scattering, and damage into a node set according to the principles of pixel consistency, community consistency, and lag matching; initialize the three positive edge weights of storm-moisture, moisture-scattering, and scattering-damage according to the lag correlation coefficient, normalize them in parallel, and then truncate them according to the prior; add a scattering-damage suppression edge and assign a negative weight when the probability of tidal residual or foam coating meets the set conditions; then traverse the sliding window along the time axis, and iteratively update the edge weights through gradient descent until the pixel residual variance is lower than the threshold. For example, Figure 4 This is a processing flow chart of a cause-effect diagram baseline provided in this embodiment.

[0050] S3, water film-structure scattering unmixing, specifically: synchronously pull VV scattering, VH scattering, surface moisture and the moisture-scattering edge weight obtained in step S2 at each pixel, use the edge weight combined with the moisture-scattering gain function to obtain the expected value of water film scattering, weight the dual-polarization scattered energy and deduct the expected value of water film scattering to obtain the structural scattering residual, apply a two-day window first-order Savitzky–Golay filter to the structural scattering residual, extract the mean residual of the seven days before the storm landfall as the baseline and normalize it to obtain the net damage sequence. For example, Figure 5 A processing flow chart of water film-structure demixing provided in this embodiment.

[0051] S4. Construct an ecological resilience baseline, specifically by: reading the elevation, aspect, conductivity, and community type grids at pixel level to form an environmental vector, using progressive similarity splitting clustering with weighted Mahalanobis distance to divide ecological zones; extracting dual-polarization radar images from the same month as the target storm from the storm-free images of the past ten years to form a storm-free control set, calculating the 5th and 95th percentiles of the scattering value on a calendar day for each ecological zone to obtain the natural fluctuation envelope, and removing abnormal curves that have been manually verified. For example, Figure 6 A flowchart for constructing an ecological resilience baseline is provided for this embodiment.

[0052] S5. Fit the vegetation damage-recovery curve based on the enhanced game, specifically: encapsulate the net damage sequence of step S3, the natural fluctuation envelope of step S4, the pixel-level moisture-scattering edge weight and the confidence factor as the game input, and the parameter player generates the initial simulation curve according to the damage peak, latent period and recovery rate. The residual player calculates the weighted residual score according to the confidence and envelope out-of-bounds situation. The parameter player corrects the parameters according to the vector-level rules until the residual score change is lower than the error threshold or reaches the preset maximum round, and outputs the recovery amplitude, latent period, recovery rate and residual index.

[0053] S6. Calculate the damage-recovery index, specifically as follows: based on the time axis calibrated by the on-site water level records, map the parameters obtained in step S5 into a segmented exponential curve, and normalize it with the peak daily envelope amplitude to obtain the peak damage rate; locate the date when the net damage drops to half of the peak value from the peak date to obtain the half-recovery time; when the net damage is below the envelope upper limit for five consecutive days, record the full recovery time; and output the peak damage rate, half-recovery time, and full recovery time rasters.

[0054] More specifically, this embodiment can be implemented through the following implementations: Step S1 includes the following steps: S11. Obtain dual-polarization synthetic aperture radar images for 30 consecutive days before and after the storm's landfall date. , passive microwave soil moisture grid , Typhoon Center Intensity Log and field sample damage records . Constructing daily time series based on Gregorian calendar dates ,Unified time scale labels are attached to the four types of data to achieve preliminary alignment of multi-source data in the time domain.

[0055] S12, yes Each image in the image is refined, terrain distortion is corrected using a digital elevation model, and radiometric calibration and slant range projection conversion are completed, converting digital quantization values ​​into backscatter coefficients. .

[0056] All images were reprojected to a unified coastal reference system to ensure comparability and spatial homogeneity between images.

[0057] S13, based on SAR pixel resolution Construct the target grid. and The terrain weight and wind field weight are jointly resampled. The weight factor is automatically allocated under the dual constraints of elevation difference and radial distance of typhoon wind field, so that the fine-scale moisture Typhoon intensity It can express the lateral seepage gradient and wind speed spatial attenuation characteristics caused by micro-topography.

[0058] S14. Collection of field sample observation dates As an anchor point, 、 and Corresponds to the most recent observation day; sets a placeholder when a day lacks observations Preserve time series integrity.

[0059] S15. Extract two types of strong scattering artificial targets, namely stable coastline and harbor retaining wall, to form a control point set , calculate the cross-date shift by phase correlation and sub-pixel interpolation and rotation angle , performing rigid body correction on each image, ultimately making the same pixel have consistent geographic coordinates on all dates.

[0060] S16, combined with daily tide levels and the angle of incidence Threshold ,shield or Pixels with a signal-to-noise ratio lower than 3dB are given a quality label. , and the rest of the pixel labels are set to Eliminate the interference of seawater-covered areas and extreme squint areas on backscatter reliability.

[0061] S17. In the common pixel domain and time domain Integrate storm intensity, moisture, dual-polarization scattering, and quadrat damage index to construct a storm-moisture-scattering synchronized grid stack: ; in: : Pixel ,date The five-dimensional feature column vector of ; : typhoon intensity index of the corresponding pixel after spatial interpolation; : soil moisture content after terrain weight-wind field weight sampling; : VV polarization backscattering coefficient; : VH polarization backscattering coefficient; : vegetation damage index corresponding to the field sample; : A synchronized grid stack with a global pixel-time dual index, used as a unified input for subsequent causal graph modeling; : pixel index; : time index; : effective pixel set of coastal zone; : Unified daily scale time series.

[0062] Step S2 includes the following steps: S21. Response baseline established. In many years of typhoon landfall, salt marshes, dune shrubs, and seagrass beds exhibited distinct water expansion rates, radar scattering gains, and physical breakage thresholds in response to high-humidity storm environments due to differences in morphology, root systems, and leaf wax. To capture this community-element-time three-dimensional response pattern, the validation quadrat data from previous years were integrated with the corresponding moisture, scattering, and damage sequences. The peak time, peak amplitude, and half-life duration were statistically analyzed by community to generate a priori database. .

[0063] Centralized recording of three-level hysteresis window for parameters in the library 、 、 , used to characterize the temporal gradient of storm-induced surface moisture jumps, water film-structure scattering peak shifts, and structural damage manifestations. The baseline inherits the actual physical time series and provides community-dependent temporal constraints for subsequent node mapping.

[0064] S22. To solve the problem of aligning multiple source elements in space and time, this step converts the four states of storm, moisture, scattering, and damage into node sets according to the three principles of same pixel, community consistency, and hysteresis matching. A one-to-one correspondence is established through a community-specific hysteresis window: ; in: : Pixel At the moment Typhoon intensity; : Typhoons experience community-specific hysteresis after landfall of surface moisture; 、 : Dual-polarization backscattering sequence, including dual contributions from water film and structure; : On-site determination of vegetation damage index; : At the reference time and the five-dimensional node vector formed by all levels of lag; : Pixel The complete set of nodes along the entire timeline.

[0065] Through the above node mapping, observations originating from different physical processes are synchronized to the same pixel index and community scale, ensuring that subsequent edge weight calculations follow the causal context of the same source, the same community, and the same lag.

[0066] S23. Directed path setting. Consider that a storm first changes the surface humidity. High humidity then increases radar backscattering through the water film effect. Ultimately, when leaves break or fall, this manifests as a physical chain of structural scattering attenuation. In this step, three forward edges are written into the adjacency matrix and the edge weights are initialized using a calculation method with a lagged correlation coefficient: ; ; in: : Pixel The three-level directed graph adjacency matrix of ; :storm The initial weight of the moisture edge; :Moisture Initial weights of scattering edges; :scattering The initial weight of the damaged edge; , ; , ; , ; : Community No. Level hysteresis window; ; : Complete timeline.

[0067] Based on the above calculation method, the edge weight not only considers the numerical correlation, but also explicitly incorporates the community hysteresis characteristics, so that the early scattering peak caused by water film and the damage hysteresis caused by structural destruction can be quantified in the initial weighting stage.

[0068] S24, yes The three forward edge weights are normalized to ensure that the sum of the outgoing edge weights is always 1. At the same time, the weight interval of the corresponding community in the prior library is read As the upper and lower bounds, the normalized results are truncated to the inside of the interval to generate the initial causal weight matrix of the pixel .

[0069] S25, during the high tide stage of the storm, some low-lying pixels are submerged in seawater for a long time and accompanied by foam filling, resulting in high specular scattering and random bubble scattering. In order to isolate this type of non-vegetation scattering, this step detects the tide level residual Probability with foam cladding ,like or In scattering Add inhibition edges to the damage direction and assign negative weights The suppression edge reduces the risk of misjudgment by weakening the transmission of structural scattering signals to damaged nodes.

[0070] S26. In high humidity scenarios, the thickness of the water film, leaf curling, and light changes all evolve dynamically over time, and the initial weights are difficult to accurately reflect the true causal strength in the long term. This step traverses the sliding window along the time axis. , calculate the observed damage and the current The residual between the damages is deduced and the weights of each edge are updated by gradient descent. The iteration terminates when , and the final causal weight tensor is output .

[0071] The results will be used in the subsequent water film-structure unmixing step to provide pixel-level, community-specific physical priors for stripping interference and reconstructing pure structural damage sequences.

[0072] Step S3 includes the following steps: S31, pixel extraction, for the causal graph domain Each pixel within , according to the timeline Simultaneously pull four types of observations: VV polarization scattering , VH polarization scattering , surface moisture , and moisture calculated by S2 Scattering causal weight The pulling process completes pixel-level alignment based on the same row and column coordinates, and labels the community With quality label They are bound together to ensure that subsequent unmixing not only takes into account the differences in water absorption rates between communities, but also inherits the confidence mark of S1–S2.

[0073] S32. Storms cause the relative humidity of the near-ground layer and the thickness of the free water film on the leaf surface to soar simultaneously, which in turn produces a linear-saturation composite gain on the backscatter. Modulate the water film response intensity and The moisture value within the daily high humidity window is averaged to obtain the expected water film scattering condition: ; in: : Pixel ,time The expected value of water film scattering; :Moisture The scattering causal edge weight reflects the water film gain sensitivity of the pixel; : The experimentally calibrated moisture-scattering gain function shows a characteristic of increasing first and then saturating; : length of the effective window of high humidity after the storm (days); : Pixel exist Surface moisture observations at all times; : Community The multi-year average of water film scattering is used to compensate for low humidity background.

[0074] Physical damage such as S33, broken plant stems, and loose roots will reduce the volume scattering and birefringence intensity. To quantify this attenuation, the dual-polarization scattering is weighted by energy conservation and the water film expectation calculated by S32 is deducted to obtain the structural scattering residual: ; in: : Pixel ,time The structural scattering residual of : VV and VH polarization energy weighting coefficient, satisfying ; : Dual-polarization backscatter measured value; : Water film scattering expectation estimated by S32.

[0075] S34, time series continuity correction, pulse noise caused by instantaneous tidal jump or foam decay, The series exhibits high-frequency spikes. To preserve the true low-frequency characteristics of the destruction-buffer-repair ecological process while suppressing meaningless high-frequency perturbations, this step applies a first-order Savitzky–Golay filter with a two-day window to the residual series. This filter maintains the locations of inflection points while smoothing the amplitudes, making the damage peaks and recovery trends more physically interpretable.

[0076] S35. The difference in the original scattering intensity of different pixels may reach 10–15 dB. If it is not normalized, it will make it difficult to unify the learning rate during subsequent game fitting.

[0077] This step extracts the mean of the structural residuals for the seven days before the storm landfall. as a steady-state baseline and calculate the normalized net damage: ; Normalized results The scale is a range where 0 indicates no damage or complete recovery and 1 corresponds to peak damage caused by the storm, creating a uniform scale for comparison across pixels and communities.

[0078] S36, finally the normalized net damage By row pixel index , by column time index Assemble into a two-dimensional matrix . The matrix elements also carry quality labels With community tags , to support the next step of enhancing the game algorithm to converge first in high-confidence pixels and post-correct low-confidence pixels, so as to achieve a robust fitting of the vegetation damage-recovery dynamics after extreme storms.

[0079] Step S4 includes the following steps: S41. The wind resilience of coastal vegetation is highly coupled with the environmental background: Elevation Directly determines the duration of tidal immersion; slope Affects solar radiation and evapotranspiration; bottom conductivity Reflects salt penetration rate; community type This step calls these four rasters and aligns them with the pixel indexes one by one to form the environment vector At the same time, the four-dimensional sample variance is counted, and the covariance matrix used for the subsequent weighted Mahalanobis distance is It is pre-calculated at this stage to ensure that different dimensions are comparable in distance measurement.

[0080] S42. Ecological zoning is delineated. Considering that salt marshes and seagrass beds have similar topography but completely different community functions, and that the southeast slope area of ​​the dune shrubland exhibits independent resilience due to stronger evaporation on the windward side, this step uses the progressive similarity splitting algorithm to identify the ecological zones. Perform hierarchical clustering.

[0081] The algorithm first takes the entire domain as the root node and calculates the weighted Mahalanobis distance of any pixel pair: ; in: : Pixel The weighted Mahalanobis distance of is used to measure the comprehensive niche difference; : The environmental vector of the two pixels; : The diagonal covariance matrix composed of the variance of each dimension of the environment vector; : The overall variance of the four factors.

[0082] Then iteratively select the largest The pixel pairs are divided along the connecting line until the maximum distance inside any sub-area is lower than the threshold The process ensures that the zoning retains significant ecological differences while avoiding excessive fragmentation, and ultimately outputs the three major ecological zones of salt marsh, dune shrub, and seagrass bed and several micro-landscape sub-zone masks. .

[0083] S43. The resilience baseline must be evaluated under the same illumination and temperature background, otherwise natural seasonal fluctuations will be misjudged as storm effects. This step queries the typhoon routes and sea wave warning records of the past ten years, eliminating all periods affected by typhoons; in the remaining years, Sentinel-1 dual-polarization SAR images of the same Gregorian calendar month as the target storm event are intercepted to construct a no-storm control set. During the process, the trajectory is kept consistent with the imaging geometry to avoid the introduction of additional scattering noise due to differences in viewing angles.

[0084] S44. In each partition For storm-free sequences, the calendar days Aggregate the scattered pixels of all years, calculate the 5th and 95th percentiles, and form the natural fluctuation envelope of this area: ; in: : Partition On calendar day The lower and upper limits of natural scattering; : Percentile operator; : VV polarization scattering in historical storm-free years; : The time index is mapped to calendar days. The 5%–95% interval covers normal fluctuations such as tidal rise and fall and vegetation growth, while also maintaining tolerance for extreme values, establishing an objective threshold for damage anomaly detection.

[0085] S45. Reclamation construction or emergency desilting in some years will cause a jump or a sharp drop in the scattering. If it is directly written into the baseline, it will cover up the true natural state. This step compares each historical curve with the envelope to locate the The method then retrieves isolated peaks and valleys from the data set. Land use change archives and on-site inspection records are then used. If human disturbance is confirmed, the corresponding curve is removed from the dataset. The remaining samples are then smoothed using LOESS to remove high-frequency noise, making the envelope more consistent with long-term natural fluctuations.

[0086] S46. Write the upper and lower bounds of each partition's envelope back to its pixel domain to generate two toughness baseline grids. . When the same pixel is fitted in the subsequent game, if the net damage curve Beyond , then the penalty term will be triggered in the parameter player's profit function, forcing the fitting result to return to the reasonable range of the partition, avoiding global distortion caused by a small number of abnormal pixels. At the same time, the partition statistics table is output , which makes it easier for regulatory authorities to quickly compare the resilience differences between storm years and natural years, and guide subsequent repair strategies.

[0087] Step S5 includes the following steps: S51, the vegetation response caused by extreme storms often spans several weeks to several months. The accuracy of time alignment directly determines the convergence efficiency of game fitting. In this step, the net damage sequence calculated by S3 is first converted to The partition toughness envelope generated by S4 By unified timeline Complete alignment and remove the quality labels marked in S1 Invalid observation. Then read the pixel-level moisture Scattering edge weight , using the linear mapping function Generate confidence factors, Set by the global median, ensuring that the confidence level is distributed in The maximum number of rounds in the game is preset. and convergence error threshold , completing the encapsulation of pixel-level game input.

[0088] S52: At the beginning of the storm's landing, plants are rapidly damaged by the impact of fractures, and then enter a latent period before gradually starting the physiological repair mechanism. For this two-stage process, the parameter player first increases the baseline along the Find the damage peak on the top, and set the corresponding time as , based on which the initial damage amplitude is given Incubation period The time difference between the damage peak and the first inflection point of the curve is taken, and the initial recovery rate is obtained by the envelope slope Driven by this ternary parameter, the expression for generating the initial simulation curve is: ; in: : First simulation of net damage; : initial damage amplitude; : incubation period; : Initial estimate of recovery rate; : storm landfall reference time; : global nonlinear acceleration factor, take 1.2; : Attenuation modulation coefficient, take .

[0089] The first version of the simulation curve expression uses the Sigmoid function to capture the latent-startup transition in the front part, and the exponential term in the back part simulates the repair rate that decreases over time, making the curve have the consistency of ecological process.

[0090] S53, residual players find fault. Under high humidity interference, the net damage sequence often contains two types of errors: short spikes and slow seepage. If the residual is minimized globally at one time, the curve shape will easily be distorted by the spikes. The residual player first calculates the difference between the two curves. , and then find the maximum continuous residual segment through the sliding window accumulation method For this error signal, the residual player constructs a double weight of confidence and baseline violation to form a weighted residual score: ; in: : first round residual score; : confidence factor; : The penalty coefficients for exceeding the upper and lower bounds are both 1.5; : indicator function.

[0091] Based on this design, ecologically unacceptable deviations can be exposed first, while the weights of noise pixels are reduced to prevent them from misleading parameter players.

[0092] S54, parameter players according to The magnitude and direction of the deviation are used to implement vector-level correction on the ternary parameters: if the deviation is positive and occurs in the later stage of the repair, then the To speed up recovery; if the deviation is negative and concentrated in the latent segment, reduce and fine-tune The correction range strictly follows ,in is the learning rate, is the direction vector, which is determined by the adaptive shape of the deviation segment. In this process, the confidence pass Already pre-amplified, high-confidence pixels will get a larger step size and thus converge earlier.

[0093] S55, the game enters After the round, the residual players re-evaluate ,like and , the game continues; otherwise it ends and records .

[0094] Based on this, similar to the high-confidence salt marsh pixels, the water film scattering has been completely stripped off, and the residual mainly comes from the real break recovery, which usually converges in 5 to 7 rounds; while the seagrass bed pixels are greatly affected by the tidal start, and most of them need to be close to .

[0095] S56. After all pixels have completed the game, the final parameters and residual indicators are written back to the four grids: Recovery range , incubation period , recovery rate , residual penalty A zoning statistics table and error distribution histogram are also output, allowing ecological authorities to assess storm recovery project priorities and key vulnerable areas. Through the aforementioned enhanced game, this step successfully achieves pixel-level robust fitting of vegetation structure destruction and recovery dynamics after removing water film scattering interference.

[0096] Step S6 includes the following steps: S61, parameter grid loading, after the typhoon passes, the residual tide level and sea fog backflow will cause the risk of misjudgment of the time axis, so first read After three layers of grids, the storm landfall reference date was set using on-site water level records. Backward correction is performed to the actual high tide peak date to ensure that the curve starting point is synchronized with the maximum structural impact. All grids are then projected onto a unified daily time axis. Observable identification bits are set for pixels that are easily obscured by salt marsh water, and missing days are automatically skipped in subsequent calculations. This step provides a data foundation with complete time series and guaranteed quality for curve reconstruction.

[0097] S62. During the incubation period, the vegetation is still in a state of tissue water saturation and growth stagnation. Even if the leaves fall over, they will not recover immediately. The segment directly maintains the peak As the tide recedes and light returns, the repair rate is gradually dominated by respiration and can be The piecewise function can be specifically expressed as: ; Therefore, it is consistent with the physiological delay and avoids the uncertainty of multi-parameter fitting. The values ​​will gradually approach , reflecting the maximum natural restoration limit allowed by the ecological niche.

[0098] S63. The instantaneous wind pressure of extreme storms is determined by both the amount of drifting debris and the amount of wind pressure. In order to convert this absolute amplitude into a relative index comparable across regions, this step is based on the peak day. Envelope amplitude of the day Normalize and get the peak damage rate .like , indicating that the structural damage of this pixel has exceeded the upper limit of historical natural fluctuations and is a severely damaged area that requires urgent intervention.

[0099] S64, half recovery time It reflects the time required for vegetation to transition from a critical survival zone to a self-healing zone, and is a quantitative indicator of early restoration efficiency after a storm impact. Scan daily from the peak date, once detected (allowing 2% tolerance), i.e. recording date .short They often correspond to areas with sunny slopes or good bottom permeability, verifying the regulatory effects of environmental factors in S1–S4 on the recovery rate.

[0100] S65, even if the exponential decay trend is good, high humidity or subsequent small storms will cause secondary disturbances. Therefore, this step requires First fall After five consecutive days of stability, the full recovery time is recorded. If a second transgression occurs within five days, the area is marked as unrecovered and designated as a priority area for subsequent monitoring. This design aligns with the frequent intertidal inundation of coastal areas and the recurring salt stress experienced by vegetation.

[0101] S66, peak damage rate, half recovery time , full recovery time The three indicator grids can be directly superimposed on the original causal weight layer for dynamic linkage display by the intelligent monitoring platform. Calculate the mean, standard deviation, and 25%–75% quantiles, and generate a partition layer and statistical table.

[0102] The data results are directly fed back to the management department to quickly identify the dual vulnerable areas of severe damage and slow recovery, and to implement targeted restoration measures such as artificial support, grass planting and sand fixation, or tidal gully dredging to achieve differentiated ecological governance after extreme storms.

[0103] In summary, the technical solution proposed in this embodiment first constructs a synchronized "storm-moisture-scattering" grid stack at the data organization level, and then achieves strict alignment of multi-source field quantities in the pixel domain and daily scale through joint resampling of terrain weights and wind field weights. The resampling weights are determined by the dual factors of local elevation difference and radial distance of the typhoon wind field. This not only preserves the lateral seepage gradient controlled by microtopography, but also synchronously maps the spatial attenuation of wind speed, allowing the temporal evolution of typhoon intensity and soil moisture fields to be coupled to radar scattering pixels with consistent resolution. Furthermore, the solution uses two types of high-signal-to-noise artificial targets—stable shorelines and harbor retaining walls—to perform cross-date rigid body correction, ensuring that the same pixel has stable geographic coordinates across the entire timeline. This processing not only provides a unified and frame-free data cube for subsequent algorithms but also establishes a traceable comparison benchmark for the drastically changing high-humidity background in the short period after storm landfall. This is the most significant structural difference from traditional pre-disaster and post-disaster scene difference or long-baseline interferometry strategies.

[0104] At the causal modeling level, this embodiment targets the physical resilience characteristics of different vegetation communities, pre-statistics the typhoon sample curves of previous years, extracts three sets of parameters: peak amplitude, peak time, and half-life time, forms a community response priori library, and provides a three-level lag window. Subsequently, in the pixel-level node set, the "storm intensity node", "soil moisture node", "VV / VH dual-polarization scattering node", and "damage index node" are explicitly distinguished, and three positive main edges of storm-moisture, moisture-scattering, and scattering-damage, as well as a negative edge of tidal bubble suppression, are established based on the community-specific lag relationship. The initial edge weight is calculated by the lagged correlation coefficient and normalized on the column vector, and then truncated in combination with the edge weight interval set by the priori library to ensure that the weight reflects both real-time correlation and meets the constraints of the community's physical time series. To overcome edge weight drift caused by water film thickness, leaf curling, and light variations in high humidity environments, the proposed approach applies a sliding window gradient descent strategy over the entire timeline. Using the residuals of observed and inferred damage as the objective function, the edge weights are adaptively updated until the residual variance converges, resulting in a causal weight tensor with temporal continuity and physical interpretability. This causal graph construction and dynamic calibration mechanism, centered around a community lag window, differs from existing approaches that directly establish feature-label mappings using empirical thresholds or black-box learning. It explicitly captures the multi-stage transmission and driving rate differences of the water film-fracture-recovery chain.

[0105] During the damage extraction and recovery fitting phase, this embodiment uses pixel-level moisture-scattering edge weights as modulation coefficients. Combined with an experimentally calibrated first-increase-then-saturation gain function, the moisture sequence within the high-humidity window is convolved to obtain the water film scattering expectation. This expectation is then subtracted using dual-polarization energy conservation weighting to obtain the pure structural scattering residual. Subsequently, a short-window Savitzky–Golay filter is used to suppress tidal impulse noise, and the mean value of the seven days before the storm landfall is introduced to normalize the resulting net damage sequence. To fit the damage-recovery curve, a two-agent augmented game of "parameter player-residual player" is designed: the parameter player describes the entire process of incubation, initiation, and repair using a sigmoid-exponential piecewise function, while the residual player weights the coherence error based on confidence and toughness envelope violations, driving the adaptive convergence of the parameter vector during iterations. Upon termination of the game, three rasters representing the recovery amplitude, incubation period, and recovery rate are output, and the peak damage rate, half-recovery time, and full recovery time are further calculated. This step-by-step unmixing and game fitting strategy avoids the damage overestimation or recovery distortion problems of traditional single exponential or static machine learning models under high humidity disturbances by explicitly stripping off the water film gain and using residual feedback to dynamically correct the model parameters, and constitutes the core algorithm that this application intends to protect.

[0106] Compared to existing approaches based on pre- and post-disaster differentials or fixed-window exponential models, this embodiment utilizes a synchronized storm-moisture-scattering grid stack and joint resampling of terrain and wind weights, ensuring that typhoon intensity, soil moisture, and dual-polarization scattering maintain the same step size and resolution within the pixel domain. This structure provides an equivalent spatiotemporal coordinate system for subsequent analysis, avoiding convolution errors caused by inconsistent resolution of multi-source field quantities. While existing methods typically rely on image- or partition-level thresholds, this approach establishes a pixel-level causal graph with community-specific lag windows, explicitly quantifying the multi-level progressive relationship between storm, moisture, scattering, and damage. Edge weights are continuously updated through sliding-window gradient descent, achieving dynamic adaptive deduction that matches physical processes.

[0107] In the damage extraction process, this embodiment uses moisture-scattering edge weights to modulate the high-humidity window, first stripping off the water film gain and then calculating the structural scattering residual, thereby fundamentally separating the two opposing signals of "increased scattering due to rising water content" and "decreased scattering due to structural breakage." Traditional amplitude ratio or coherence threshold methods cannot provide this resolution in high-humidity environments. Furthermore, an enhanced game mechanism of parameter player-residual player is introduced to link and update ecological process parameters such as the incubation period and recovery rate with the confidence and resilience envelope, so that the damage-recovery curve strikes a balance between morphological continuity and statistical consistency, rather than relying on a large number of labeled samples to train a black-box model. Through the above-mentioned differentiated design, this embodiment can maintain stable detection of structural damage and fine quantification of the recovery process under conditions of drastic tidal changes and strong short-term water film interference, reflecting its targeted adaptation to extreme coastal storm scenarios and its superiority in principle.

[0108] A specific embodiment is as follows: Following a super typhoon, a provincial bay area wetland conservation center launched an 18-month project to monitor damage and recovery of salt marshes, dune shrublands, and adjacent seagrass beds. Historically, rapid assessments in this area have been conducted using pre- and post-disaster NDVI difference methods, but these methods suffer from high false positive rates and difficulty quantifying half-recovery times in the presence of high humidity and salt fog. This monitoring effort, jointly conducted by the Institute of Oceanography's remote sensing team and the Department of Natural Resources, fully incorporated the "storm-moisture-scattering" synchronized grid stack and pixel-level causal game model proposed in this example.

[0109] During the data preparation phase, this example collected 11 Sentinel-1A / BIW dual-polarization SAR images (10m) from August 28, 2022, to October 12, 2022. The SMAPL3 daily soil moisture product (9km) was also downloaded, and a 6-hour intensity track of the typhoon's center was obtained from JTWC. In the field, 32 plots were deployed in salt marshes and sand dunes, and DI (Damage Index) was recorded using inclinometers and break rate scales. All raster data were track refined, SRTM-1 DEM corrected, radiometrically calibrated, and slant range converted before being reprojected to UTM-50N. The JTWC intensity grid and the SMAP moisture grid were then resampled to 10m resolution using the ΔH–d_r dual-constraint algorithm. The resulting data were merged into a synchronized stack containing Ty, SM, VV, VH, and DI, covering the complete 30-day daily time series before and after landfall.

[0110] During the causal modeling phase, this example constructed a priori database using quadrat records from five typhoon events from 2016 to 2021. Statistics were obtained for salt marsh lag windows {1d, 3d, 6d}, dune shrubs {2d, 5d, 7d}, and seagrass beds {0d, 2d, 4d}. Following the pixel-community consistency rule, Ty, SM, VV / VH, and DI were mapped to nodes, and three positive primary edge weights were initialized. Negative inhibitory edges were automatically added to pixels with a 300mm tide residual and a foam probability of 0.65. The causal weight tensor converged when the global residual variance was iterated to 0.018.

[0111] In the water film-structure demixing process, this embodiment is based on The expected water film scattering was calculated, and a structured scattering residual sequence was generated under the dual-polarization energy conservation constraint. A subsequent 2D Savitzky-Golay filter effectively attenuated the spikes caused by the receding tide. The net damage ND(t), normalized by the mean value over the seven days before landfall, was obtained. Peak values ​​for salt marshes reached 0.83, sand dunes 0.57, and seagrass beds 0.48.

[0112] The ecological resilience baseline was divided into five micro-landscape sub-areas using the weighted Mahalanobis distance splitting algorithm, and the 5%-95% natural fluctuation envelope was calculated using storm-free Sentinel-1 images from the same months of 2012-2021. In the enhanced game, the parameter player used the Sigmoid-exponential piecewise function to give the initial curve. The residual player performs weighted picking based on confidence and out-of-bounds conditions, and each pixel converges after an average of 6.3 rounds. Output the recovery amplitude A, latency, and recovery rate grid, and further calculate the peak damage rate PD and half-recovery time. and full recovery time.

[0113] In the phased acceptance on 2022-10-18, this embodiment selects The results of the comparison of the traditional NDVI difference method, the dual-polarization amplitude ratio method and this embodiment in the densely sampled area are shown in Table 1:

[0114] Table 1 The results show that this embodiment has smaller numerical deviation in peak damage location and half-recovery time estimation, and the comprehensive detection of damaged pixels is more accurate. A 13 percentage point increase will be made. A priority list for repairs will be issued in January 2023, with PD>0.7 and The salt marsh mudflats were identified as key areas for emergency sand fixation and tidal gully dredging, and the subsequent on-site review and confirmation plan had an inference accuracy rate of 87%.

[0115] This example fully verifies the feasibility of applying synchronized grid stacking, community lagged causal diagrams, and water film-structure unmixing to quantitative monitoring of vegetation damage after extreme storms. It also demonstrates the stability of enhanced game curve fitting in the estimation of recovery process parameters, laying a technical foundation for subsequent promotion in longer time series and larger areas.

[0116] Reference Figure 7 The present application also provides a device for evaluating coastal vegetation damage and restoration after an extreme storm, which can implement the above-mentioned method for evaluating coastal vegetation damage and restoration after an extreme storm. The device includes: A data acquisition unit is used to integrate storm intensity, moisture, VV polarization scattering, VH polarization scattering and damage index in the common pixel domain and time domain to construct a synchronized grid stack; A graph construction unit is configured to map the four states of storm, moisture, scattering, and damage in the synchronized grid stack into a node set, and construct a pixel-level community lag causal graph based on each node in the node set; a damage determination unit, configured to obtain an expected value of water film scattering according to the pixel-level community hysteresis causal graph, and determine a net damage sequence using the expected value of water film scattering; A natural fluctuation envelope determination unit is configured to obtain ecological zones of different environments according to the synchronous grid stack division, and calculate a scattering value in each ecological zone to obtain a natural fluctuation envelope; A parameter fitting unit is used to encapsulate the net damage sequence, the natural fluctuation envelope, the edge weights of pixel-level moisture and scattering in the pixel-level community lag causal graph, and the confidence factor as game inputs and input them into a parameter player, and use the parameter player to output multiple parameters of vegetation damage and recovery; The damage and recovery evaluation unit is used to output the peak damage rate, half recovery time and full recovery time of vegetation according to the multiple parameters of vegetation damage and recovery.

[0117] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0118] The present application also provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method of the present application. The electronic device can be any smart terminal, such as a tablet computer or an in-vehicle computer.

[0119] It can be understood that the contents of the above method embodiments are all applicable to the embodiments of the present device, the functions specifically implemented by the embodiments of the present device are the same as those of the method of the present application, and the beneficial effects achieved are also the same as those achieved by the method of the present application.

[0120] See also Figure 8 , Figure 8 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes: The processor 801 can be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application. The memory 802 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 802 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 802 and is called by the processor 801 to execute the methods of the embodiments of this application. Input / output interface 803, used to implement information input and output; Communication interface 804, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.); Bus 805 , which transmits information between various components of the device (e.g., processor 801 , memory 802 , input / output interface 803 , and communication interface 804 ); The processor 801 , the memory 802 , the input / output interface 803 and the communication interface 804 are connected to each other in communication within the device via a bus 805 .

[0121] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the method of the present application is implemented.

[0122] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiment, the functions specifically implemented by the present storage medium embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0123] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0124] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0125] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0126] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0127] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0128] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0129] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0130] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0131] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0132] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0133] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0134] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A method for assessing coastal vegetation damage and recovery after extreme storms, characterized in that: The method comprises the following steps: The storm intensity, moisture, VV polarimetric scattering, VH polarimetric scattering, and damage index are integrated in the common pixel domain and time domain to construct a synchronized grid stack. Mapping the four states of storm, moisture, scattering and damage in the synchronized grid stack into a node set, and constructing a pixel-level community lag causal graph based on each node in the node set; Obtaining a water film scattering expectation value according to the pixel-level community lag causal graph, and determining a net damage sequence using the water film scattering expectation value; Dividing the synchronized grid stack into ecological zones of different environments, calculating the scattering value in each ecological zone to obtain a natural fluctuation envelope; The net damage sequence, the natural fluctuation envelope, the edge weights of pixel-level moisture and scattering in the pixel-level community lag causal graph, and the confidence factor are encapsulated as game inputs and input into a parameter player, and the parameter player is used to output various parameters of vegetation damage and recovery; Outputting the peak damage rate, half-recovery time and full recovery time of vegetation according to the multiple parameters of vegetation damage and recovery; The step of integrating storm intensity, moisture, VV polarization scattering, VH polarization scattering, and damage index in the common pixel domain and the time domain to construct a synchronized grid stack includes the following steps: Obtaining recorded data for the first set number of consecutive days before and after the storm's landfall date; wherein the recorded data includes dual-polarization synthetic aperture radar radar images, soil moisture grids, typhoon intensity grids, and damage records from field plots; Adding a unified daily time scale to the recorded data; performing orbit refinement, digital elevation model correction, radiometric calibration, and slant range projection on the radar image in sequence, and then converting and reprojecting the image into a unified reference system; Resampling the typhoon intensity grid and the soil moisture grid using terrain weights and wind field weights based on radar pixel resolution; wherein the weight factors are automatically assigned under the dual constraints of elevation difference and radial distance of the typhoon wind field; The time series is completed using the observation date of the field sample as the anchor point, and the same pixel is kept in the same geographical coordinate on all dates through rigid body correction; The quality label is set in combination with the daily tide level and the incident angle threshold, and then the storm intensity, moisture, VV polarization scattering, VH polarization scattering and damage index are integrated in the common pixel domain and time domain to obtain the synchronized grid stack.

2. The method for assessing coastal vegetation damage and restoration after an extreme storm according to claim 1, characterized in that: Mapping the four states of storm, moisture, scattering, and damage in the synchronized grid stack into a node set includes the following steps: A community response priori database is generated using typhoon cases over the years, and a community three-level lag window is obtained based on the community response priori database; According to the principles of pixel consistency, community consistency, and lag matching, the four states of storm, moisture, scattering, and damage in the synchronized grid stack are mapped to the node set; The step of constructing a pixel-level community hysteresis causal graph based on each of the nodes in the node set comprises the following steps: The forward edge weights corresponding to storm and moisture, moisture and scattering, and scattering and damage are initialized according to the lagged correlation coefficient, normalized in parallel, and then truncated according to the prior; Adding negative edges of the scattering and damage when the probability of tide level residue or foam coating meets the set conditions; Traversing the sliding window along the time axis, the weight of each edge is iteratively updated by gradient descent until the pixel residual variance is lower than the threshold, thereby obtaining the pixel-level community lag causal graph.

3. The method for assessing coastal vegetation damage and restoration after an extreme storm according to claim 1, characterized in that: The method of obtaining the expected value of water film scattering according to the pixel-level community hysteresis causal graph comprises the following steps: Synchronously extracting the VV polarized scattering, the VH polarized scattering, the moisture, and the edge weights of moisture and scattering obtained according to the pixel-level community lag causal graph at each pixel; The expected value of water film scattering is obtained by combining the edge weights of moisture and scattering with moisture and scattering gain functions; The method of determining a net damage sequence by using the water film scattering expectation value comprises the following steps: Weighting the dual-polarization scattered energy and deducting the water film scattering expectation value to obtain the structural scattering residual; After applying a two-day window first-order Savitzky–Golay filter to the structural scattering residuals, the mean of the residuals for the second set number of days before the storm landfall is extracted as a baseline and normalized to obtain the net damage series.

4. The method for assessing coastal vegetation damage and restoration after an extreme storm according to claim 1, characterized in that: The method of obtaining ecological zones of different environments according to the synchronized grid stack division includes the following steps: Reading elevation, aspect, conductivity, and community type rasters by pixel alignment according to the synchronized grid stack to form an environmental vector; The environmental vectors are divided into the ecological zones of different environments by using progressive similarity splitting clustering of weighted Mahalanobis distance; Calculating the scattering value in each ecological zone to obtain the natural fluctuation envelope includes the following steps: Extract dual-polarization radar images of the same month as the target storm from the storm-free images in the most recent set number of years to form a storm-free control set; Calculating the 5th percentile and the 95th percentile of the scattering value on a calendar day basis in each ecological zone based on the no-storm control set to obtain the natural fluctuation envelope; Abnormal curves that have been manually verified in the natural fluctuation envelope are eliminated.

5. The method for assessing coastal vegetation damage and restoration after an extreme storm according to claim 1, characterized in that: The method of using the parameters to output various parameters of vegetation damage and restoration includes the following steps: Using the parameters, the player generates a preliminary simulation curve based on the damage peak, incubation period, and recovery rate; Calculating a weighted residual score based on the confidence level and envelope crossing situation of the initial version simulation curve using residual players; Using the parameters, the player modifies the parameters according to the vector-level rules until the change value of the weighted residual score is lower than the error threshold or reaches the preset maximum round, and outputs the recovery amplitude, latent period, recovery rate and residual index as the multiple parameters of the vegetation damage and recovery.

6. A method for assessing coastal vegetation damage and restoration after an extreme storm according to any one of claims 1 to 5, characterized in that: Outputting the peak damage rate, half-recovery time, and full recovery time of vegetation based on the multiple parameters of vegetation damage and recovery includes the following steps: Correcting the timelines of various parameters of vegetation damage and recovery based on field water level records; Mapping the various parameters of vegetation damage and recovery into a piecewise exponential curve, and normalizing the curve with the peak daily envelope amplitude to obtain the peak damage rate; Locate the date when the net damage drops to half of the peak value from the peak date to obtain the half-recovery time; When the net damage is below the upper limit of the envelope for the third consecutive set number of days, the full recovery time is recorded; The peak damage rate, the half recovery time, and the full recovery time are output.

7. A device for assessing coastal vegetation damage and recovery after extreme storms, characterized in that: The device comprises: A data acquisition unit is used to integrate storm intensity, moisture, VV polarization scattering, VH polarization scattering and damage index in the common pixel domain and time domain to construct a synchronized grid stack; A graph construction unit is configured to map the four states of storm, moisture, scattering, and damage in the synchronized grid stack into a node set, and construct a pixel-level community lag causal graph based on each node in the node set; a damage determination unit, configured to obtain an expected value of water film scattering according to the pixel-level community hysteresis causal graph, and determine a net damage sequence using the expected value of water film scattering; A natural fluctuation envelope determination unit is configured to obtain ecological zones of different environments according to the synchronous grid stack division, and calculate a scattering value in each ecological zone to obtain a natural fluctuation envelope; A parameter fitting unit is used to encapsulate the net damage sequence, the natural fluctuation envelope, the edge weights of pixel-level moisture and scattering in the pixel-level community lag causal graph, and the confidence factor as game inputs and input them into a parameter player, and use the parameter player to output multiple parameters of vegetation damage and recovery; A damage and recovery assessment unit, configured to output a peak damage rate, a half-recovery time, and a full recovery time of vegetation based on the plurality of parameters of vegetation damage and recovery; The step of integrating storm intensity, moisture, VV polarization scattering, VH polarization scattering, and damage index in the common pixel domain and the time domain to construct a synchronized grid stack includes the following steps: Obtaining recorded data for the first set number of consecutive days before and after the storm's landfall date; wherein the recorded data includes dual-polarization synthetic aperture radar radar images, soil moisture grids, typhoon intensity grids, and damage records from field plots; Adding a unified daily time scale to the recorded data; performing orbit refinement, digital elevation model correction, radiometric calibration, and slant range projection on the radar image in sequence, and then converting and reprojecting the image into a unified reference system; Resampling the typhoon intensity grid and the soil moisture grid using terrain weights and wind field weights based on radar pixel resolution; wherein the weight factors are automatically assigned under the dual constraints of elevation difference and radial distance of the typhoon wind field; The time series is completed using the observation date of the field sample as the anchor point, and the same pixel is kept in the same geographical coordinate on all dates through rigid body correction; The quality label is set in combination with the daily tide level and the incident angle threshold, and then the storm intensity, moisture, VV polarization scattering, VH polarization scattering and damage index are integrated in the common pixel domain and time domain to obtain the synchronized grid stack.

8. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images