Wetland landscape space-time dynamic characteristic analysis method and system based on multi-source remote sensing

Abstract

The invention discloses a wetland landscape space-time dynamic characteristic analysis method and system based on multi-source remote sensing, relates to the technical field of wetland landscape dynamic analysis, and systematically captures a dynamic evolution process of a landscape under the combined action of nature and man-made interference by constructing a three-level framework of landscape structure parameters, interference factors and time sequence change response. The method specifically comprises the following steps: forming a data flow closed loop connected front and back among the modules, and dynamically tracking the response and feedback of the landscape structure along with time; spatial scale mapping and landscape patch feature evolution are comprehensively considered, and application requirements under different regional scales and time dimensions are met; the method does not stay at the current state description, but outputs a representation value capable of reflecting an evolution trend and a process mechanism, thereby expanding the analysis dimension. Multi-source remote sensing data are uniformly mapped to a structure unit with a space unit Ui as a core through a unit scale feature mapping and averaging processing unit, and dimension reduction aggregation of pixel-level data is achieved.

Description

Technical Field

[0001] This invention relates to the field of wetland landscape dynamic analysis technology, specifically to a method and system for analyzing the spatiotemporal dynamic characteristics of wetland landscapes based on multi-source remote sensing. Background Technology

[0002] Wetland landscape structure evolution involves monitoring changes in the spatial structure, patch succession, functional differentiation, and ecological integrity of wetlands using technological means. In recent years, with the improvement of the spatiotemporal resolution of remote sensing images and the refinement of ecological monitoring technologies, a new research trend has emerged: dynamic characterization of wetland landscapes based on the fusion of multi-source remote sensing and ecological data. Its core objective is to achieve comprehensive perception and analysis of the structural dynamics, process responses, and temporal changes of wetland systems under complex disturbance conditions.

[0003] Current wetland landscape monitoring methods mostly rely on response analysis driven by a single disturbance variable, such as analyzing only the impact of land use change on wetland patches or examining only the effects of a specific type of anthropogenic disturbance on wetland structure. These methods generally suffer from the following shortcomings: The data types are too limited, lacking a comprehensive consideration of the combined effects of natural and anthropogenic disturbances; the spatiotemporal processes are fragmented, lacking the ability to incorporate changes across multiple time scales and spatial levels into a single model framework for systematic analysis; and the static structure dominates, overemphasizing the current structural state while neglecting the evolution of the landscape and its responsiveness, making it difficult to reflect the true dynamic changes of the wetland system.

[0004] The aforementioned situation primarily stems from the coexistence of multiple disturbances and nonlinear superposition effects faced by wetland systems in real-world environments, including the intertwining of human-induced construction and natural disturbances. For example, infrastructure construction activities may be accompanied by natural factors such as abnormal precipitation or fluctuations in groundwater levels, making the changes in wetland systems highly complex. In the absence of a unified analytical framework, the following abnormal consequences often occur: Risk misjudgment: Dramatic fluctuations in localized plaques may be mistaken for systemic evolution, leading to flawed governance decisions; Neglecting recovery potential: Some areas may have structural recovery capabilities, but are misjudged as irreversible degradation zones from a static analysis perspective. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for analyzing the spatiotemporal dynamic characteristics of wetland landscapes based on multi-source remote sensing, thus solving the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing, comprising a multi-source data synchronous acquisition module, a spatial scale consistency processing module, a landscape structure parameter extraction module, a multiple interference factor analysis and quantification module, a coupled state calculation module, and a wetland landscape spatiotemporal dynamic comprehensive module. The multi-source data synchronous acquisition module collects wetland data through acquisition devices and fits it into the original remote sensing dataset YW; The spatial scale uniformity processing module preprocesses the original remote sensing dataset YW and maps the data to the unit scale to obtain the wetland spatial uniformity set SW; The landscape structure parameter extraction module analyzes each cell of the wetland spatial consistency set SW, extracts landscape structure parameters, and fits them into a landscape feature set FGW. The multi-disturbance factor analysis and quantification module is based on the unit scale of the wetland spatial uniform set SW, constructs the natural disturbance intensity Inat and the anthropogenic disturbance density Ihum, and fits them into the disturbance feature set RW; The coupling state calculation module performs perturbation calculation on the data in the perturbation feature set RW to obtain the perturbation coupling state value Cint; The wetland landscape spatiotemporal dynamic integrated module is based on the disturbance coupling state value Cint, combined with the landscape feature set FGW for unified characterization, constructs the landscape spatiotemporal dynamic representation value DLS, and provides feedback.

[0007] Preferably, the multi-source data synchronous acquisition module includes a multi-source sensing data acquisition unit and a multi-source data time fitting unit; The multi-source sensing data acquisition unit collects wetland data through various acquisition devices, including optical remote sensing image data O(x,y,t), radar image data R(x,y,t), digital elevation model data E(x,y,t), water depth ecological monitoring data WH(t), and land use record data L(x,y,t). Among them, the optical remote sensing image data O(x, y, t) is acquired by multispectral remote sensing satellites; the acquisition method is: a preset imaging cycle, covering once every 16 days, with a band resolution of 10–30m; Radar image data R(x, y, t) is acquired through synthetic aperture radar; acquisition method: all-weather, all-time acquisition, penetrating clouds and vegetation; The digital elevation model data E(x, y, t) is acquired by lidar; acquisition method: accuracy higher than 5 meters, periodic updates or on-site aerial surveys; Water depth ecological monitoring data WH(t) is acquired through water depth sensors and pressure level gauges; acquisition method: installed at typical water body monitoring points, with a sampling frequency set such as 10 min; Land use record data L(x, y, t) was acquired through UAV remote sensing and historical remote sensing classification maps; acquisition method: multi-source stitching, update cycle 1-3 months; The multi-source data temporal fitting unit performs linear interpolation on the water depth ecological monitoring data WH(t) to fill in missing values; spatially aligns the radar image data R(x,y,t) and optical remote sensing image data O(x,y,t) and projects them onto a unified reference grid; and removes noise from the digital elevation model data E(x,y,t) and land use record data L(x,y,t) and fits them into the original remote sensing dataset YW.

[0008] Preferably, the spatial scale uniformity processing module includes a spatial geometric correction and resolution resampling unit and a unit scale feature mapping and averaging processing unit; The Spatial Geometric Correction and Resolution Resampling Unit preprocesses the original remote sensing dataset YW, including geometric registration, resolution normalization resampling, and uniform rasterization. The geometric registration process involves performing a geometric transformation on the data in the original remote sensing dataset YW to transform the original image coordinates into the target image coordinates. The resolution normalization resampling process is as follows: linearly resample the data in the original remote sensing dataset YW to a unified spatial resolution step size of r×r. The unified rasterization process is as follows: U i ={(x,y)∣x∈[x i x i+r ), y∈[y i y i+r )}; In the formula, U i Let (x, y) represent the i-th spatial unit, and (x, y) represent the geographic spatial coordinates. i y i ) represents the spatial unit U i The starting coordinates are given by r, where r represents the spatial step size.

[0009] The spatial step size r is obtained by taking the square root of the ratio of the total area of ​​the study region to the desired number of subdivision units. In landscape ecological analysis, to ensure that spatial units can "identify" key patches, boundaries, and transition zones, the minimum identification scale of patches should be referenced; Preferably, the unit-scale feature mapping and averaging processing unit will process the spatial unit U i The original remote sensing dataset YW was aggregated and transformed to obtain the wetland spatial consistency set SW. The spatial uniform set SW of wetlands is obtained using the following formula: ; In the formula, SWo(t) represents the o-th data item in the wetland spatial consistency set SW at time t, and YWo(t) represents the o-th data item in the original remote sensing dataset YW at time t.

[0010] Preferably, the landscape structure parameter extraction module includes a wetland patch boundary identification calculation unit and a landscape structure parameter calculation unit; The wetland patch boundary identification calculation unit uses land use record data L(x, y, t) in spatial unit U. i The set of pixels marked as wetland type is selected from the internal data to form a patch region, and the spatial coverage area of ​​all pixels in the patch is counted to obtain the patch area A(i,t). By using boundary tracking methods, such as edge scanning based on pixel neighborhood and eight-neighborhood tracking operators, the outer contour of the patch is extracted, and the contour length P(i,t) is calculated. The landscape structure parameter calculation unit is based on the patch area A(i,t) and the outline length P(i,t) to calculate the patch perimeter area ratio Rpa, landscape shape tortuosity Csh and the mean patch adjacency distance Dnb, and collects them into a landscape feature set FGW. The patch perimeter area ratio Rpa is obtained by the ratio of patch area A(i,t) to contour length P(i,t); The method for obtaining the landscape shape tortuosity Csh is as follows: First, obtain the current spatial unit U. i The boundary contour of the wetland patch is obtained, and the contour length P(i,t) is obtained. Then, under the same area condition, the theoretical circular boundary length of the ideal circular patch is obtained. Finally, the landscape shape tortuosity Csh is obtained by dividing the contour length P(i,t) by the theoretical circular boundary length. The closer the actual patch shape is to a circle, the closer this ratio is to 1. The more irregular the patch shape and the more complex the boundary, the larger this ratio will be. The mean spacing between adjacent patches, Dnb, is obtained as follows: First, determine the current spatial unit U. i The location in space; then, from the wet map layer at the current time t, find the spatial unit U. i Other adjacent units, also labeled as wetland types, are denoted as adjacent patches. These adjacent patches are typically geographically connected to the current unit in a four- or eight-neighbor direction, or are within a short distance. For each adjacent patch, the shortest spatial distance to the current unit is measured; this distance is typically based on the straight-line distance between patch centroids or the shortest edge-to-edge distance. Each adjacent patch receives an independent distance value. Next, the total number of wetland patches adjacent to the current unit is counted, i.e., the number of adjacent patches. Finally, the distance values ​​of all adjacent patches are summed and divided by the number of adjacent patches to obtain the mean patch adjacency distance Dnb.

[0011] Preferably, the multiple interference factor analysis and quantification module includes a natural disturbance intensity calculation unit and an anthropogenic activity interference density calculation unit; The natural disturbance intensity calculation unit is based on the wetland spatial uniform set SW. It identifies and calculates the magnitude of natural disturbance changes caused by hydrological, humidity and topographic fluctuations in each spatial unit, including water depth time series change ΔWH and radar wave response ΔR, and fits it into the natural disturbance intensity quantity Inat. The method for obtaining the water depth time series variation ΔWH is as follows: First, find the spatial unit U at the current time t. i The data includes water depth ecological monitoring data; then, water depth ecological monitoring data at a fixed time interval before time t is found, and the difference between the two is calculated, and the absolute value is taken to obtain the spatial unit U. i Water depth time series variation ΔWH; The radar wave response ΔR is obtained as follows: First, find the spatial unit U at the current time t. i The radar image data within the time frame is used; then, radar image data at a fixed time interval before time t is found, and the difference between the two is calculated, and the absolute value is taken to obtain the spatial unit U. i The radar wave response ΔR; The formula for obtaining the natural disturbance intensity Inat is: Inat(i,t)=a1×ΔWH(i,t)+a2×ΔR(i,t); In the formula, Inat(i,t) represents the natural disturbance intensity of the i-th spatial unit at time t, ΔWH(i,t) represents the water depth time series change of the i-th spatial unit at time t, ΔR(i,t) represents the radar wave response of the i-th spatial unit at time t, a1 represents the water depth disturbance adjustment coefficient, and a2 represents the radar disturbance adjustment coefficient.

[0012] Specifically, the water depth disturbance adjustment coefficient a1 and the radar disturbance adjustment coefficient a2 are obtained as follows: Clearly define the spatial scope of the analysis and set a target time window, such as typical wetland evolution stages or the period before and after extreme events within the past five years. By calling historical or real-time remote sensing hydrological data sources, water depth data at the same time scale are extracted from spatial units within the target area, a time series is constructed, and the water depth change is calculated. Select a PolSAR radar data source, extract radar echo intensity sequences from spatial cells that are consistent with the water depth data, process them into time series format, and calculate the radar variation amplitude between adjacent time series. For each spatial cell, the fluctuation amplitude, standard deviation, and maximum change of ΔWH and ΔR are statistically analyzed to measure the severity and range of the two types of disturbance factors. The initial values ​​of the water depth disturbance adjustment coefficient a1 and the radar disturbance adjustment coefficient a2 are determined based on characteristic values ​​such as the variation range or standard deviation.

[0013] Preferably, the human activity disturbance density calculation unit identifies the degree of disturbance caused by human land use activities, such as construction expansion, farmland encroachment and road intersections, in each spatial unit, and quantifies it using spatial occupancy density. Detect whether the main land use category has changed, and construct a difference mask ΔL; The difference mask ΔL is obtained as follows: First, find the current spatial unit U. i The land use type corresponding to the current point in time, such as cultivated land, forest land, water body, and building land; then, find the current spatial unit U. i The land use type at the previous observation point in time, such as cultivated land, forest land, water body and building land; if the land use types at the two points in time are different, it indicates that land use changes have occurred in the region during the time period; When a change in land use type is detected, a value of 1 is assigned to the marker, indicating that there is land use disturbance in the spatial unit; When the land use type remains unchanged, the marker is assigned a value of 0, indicating that the spatial unit remains stable during this period; To extract the optical image change difference map ΔPros, the method is as follows: First, find the spatial unit U at the current time t. i The process involves first identifying optical remote sensing imagery data within a given timeframe (t); then finding optical remote sensing imagery data at a fixed time interval prior to time t, calculating the difference between the two, and taking the absolute value to obtain the spatial unit U. i Optical image variation difference diagram ΔPros; The difference mask ΔL and the optical image change difference map ΔPros are combined to calculate the human activity interference density Ihum. The human activity interference density Ihum indicates that the greater the optical change of the unit under the premise of land use category change, the higher the degree of human interference. If the land use remains unchanged, such as always being cultivated land, even if the optical change is large, it is not considered as new human interference. The density of human-caused interference, Ihum, is obtained by multiplying the difference mask ΔL and the optical image change difference map ΔPros. By fitting the natural disturbance intensity Inat and the human-induced disturbance density Ihum, the disturbance feature set RW is obtained.

[0014] Preferably, the coupling state calculation module includes an interference coupling strength calculation unit and an interference level discrimination unit; The interference coupling strength calculation unit calculates the interference coupling state value Cint based on the disturbance feature set RW; the acquisition method is as follows: The interference coupling state value Cint is obtained by taking the square root of the sum of the square of the natural disturbance intensity Inat and the square of the human-induced disturbance density Ihum. The interference level discrimination unit compares the acquired interference coupling state value Cint with the preset interference threshold Tnt to determine the disturbance state; the determination method is as follows: When the interference coupling state value Cint < 0.5 × interference threshold Tnt, it indicates a stable state; When 0.5 × interference threshold Tnt ≤ interference coupling state value Cint ≤ interference threshold Tnt, it indicates a moderate disturbance state. When the interference threshold Tnt is less than the interference coupling state value Cint, it indicates a strong interference state.

[0015] Preferably, the wetland landscape spatiotemporal dynamic integrated module includes a disturbance coordinated change analysis unit and a spatiotemporal evolution feedback analysis unit; The disturbance-coordinated change analysis unit is based on the disturbance coupling state value Cint, which couples the structural changes of each spatial unit with the disturbance intensity to calculate and obtain the landscape spatiotemporal dynamic characterization value DLS. The spatial-temporal dynamic representation value of landscape (DLS) is obtained using the following formula: ; In the formula, DLS(i,t) represents the spatiotemporal dynamic characterization value of the i-th spatial unit at time t, Cint(i,t) represents the disturbance coupling state value of the i-th spatial unit at time t, Rpa(i,t) represents the patch perimeter area ratio of the i-th spatial unit at time t, Rpa(i,t-Δt) represents the patch perimeter area ratio of the i-th spatial unit at time t-Δt, Csh(i,t) represents the landscape shape tortuosity of the i-th spatial unit at time t, Dnb(i,t) represents the average patch adjacency spacing of the i-th spatial unit at time t, and b represents the adjacency relationship adjustment coefficient. In particular, Cint(i,t) represents the driving term of the disturbance intensity, specifically the strength of the disturbance effect on the landscape at time t; in the multiplicative structure, it serves as the activation coefficient of the overall dynamic response; if Cint is 0, it indicates that the disturbance of this unit is extremely low, the overall DLS is zero, and pseudo-dynamics is excluded. |Rpa(i,t)-Rpa(i,t-Δt)| represents the structural change magnitude term, which describes the intensity of landscape structural change and is a dynamic indicator within a time window; multiplying it with the disturbance term means that the disturbance only has a real impact on the dynamics when the structure actually changes. [Csh(i,t)+b×Dnb(i,t)] represents the spatial structure adjustment term. The more complex the shape, the looser the adjacency, and the more likely it is to be unstable. The adjacency relationship adjustment coefficient b controls the intensity of the adjacency effect, making the model adaptable in the two types of landscapes dominated by shape adjacency. The three types of parameters with different physical meanings and different spatial statistical directions are multiplied and coupled to construct a dynamic feature quantity that can reflect the multi-factor coupling relationship between spatial structure, morphological response and disturbance state. The steps to obtain the adjacency relationship adjustment coefficient b are as follows: The study area was divided into several representative sample areas; Within each sample area: calculate the average adjacency spacing across multiple time phases; simultaneously record landscape structural stability indices, assess the impact trend of Dnb changes on structural indices, and determine the intensity of its moderating effect; Fit a function to each region; the horizontal axis represents the change in Dnb(i,t), and the vertical axis represents the change in structural indices; if the fitting trend is obvious, it indicates that the structure of the region is sensitive to the adjacent spacing, and a higher b value should be set; if the correlation is weak, a lower b value should be set. The sensitivity response levels fitted to all regions are normalized to the interval [0.1, 1]; this normalized value is the regional baseline value of the adjacency adjustment coefficient b.

[0016] The spatiotemporal evolution feedback analysis unit performs horizontal and vertical feedback evaluation on the landscape spatiotemporal dynamic characterization value DLS, identifies dynamic change trends, and obtains the time series change trend value ΔDLS. The time series trend value ΔDLS is obtained by the difference between the landscape spatiotemporal dynamic characterization value at time t and the landscape spatiotemporal dynamic characterization value at time t-Δt. When the time series trend value ΔDLS is continuously rising, it indicates that the interference effect is continuously increasing, and the structure is being disturbed and deconstructed; the potential region status is: structural degradation region, interference expansion region; When the time series trend value ΔDLS is continuously zero and tends to be stable, it indicates that the disturbance is in a balanced state; the potential region state is: a relatively stable region or an adaptation region. When the time series trend value ΔDLS is continuously negative, it indicates that the disturbance is weakening or the landscape is recovering to its original state; the potential area status is: recovery zone or recovery potential area; When the time series trend value ΔDLS is in a high-frequency fluctuation state, it indicates that it is in a critical state or an unstable transition state; the potential region states are: boundary region, evolutionary transition region, and disturbance edge region.

[0017] A method for analyzing the spatiotemporal dynamic characteristics of wetland landscapes based on multi-source remote sensing includes the following steps: Step 1: The multi-source data synchronous acquisition module collects wetland data through the acquisition device and fits it into the original remote sensing dataset YW; Step 2: The spatial scale uniformity processing module preprocesses the original remote sensing dataset YW and maps the data to the unit scale to obtain the wetland spatial uniformity set SW; Step 3: The landscape structure parameter extraction module analyzes each cell of the wetland spatial consistency set SW, extracts the landscape structure parameters, and fits them into the landscape feature set FGW. Step 4: The multi-disturbance factor analysis and quantification module is based on the unit scale of the wetland spatial consistency set SW, constructs the natural disturbance intensity Inat and the anthropogenic disturbance density Ihum, and fits them into the disturbance feature set RW. Step 5: The coupling state calculation module performs perturbation calculation on the data in the perturbation feature set RW to obtain the perturbation coupling state value Cint; Step Six: The wetland landscape spatiotemporal dynamic integrated module uses the disturbance coupling state value Cint as a basis, combines it with the landscape feature set FGW for unified characterization, constructs the landscape spatiotemporal dynamic representation value DLS, and provides feedback.

[0018] This invention provides a method and system for analyzing the spatiotemporal dynamic features of wetland landscapes based on multi-source remote sensing, which has the following beneficial effects: (1) During system operation, by constructing a three-level framework of landscape structure parameters, disturbance factors and temporal change response, the system captures the dynamic evolution process of the landscape under the combined effects of natural and human disturbances. It comprehensively considers spatial scale mapping and landscape patch feature evolution to adapt to application needs under different regional scales and time dimensions; it no longer stops at the current state description, but outputs characterization values ​​that can reflect the evolution trend and process mechanism, thus expanding the analysis dimensions.

[0019] (2) By constructing a land use type difference mask, it accurately identifies whether changes in the main category of human activities have occurred in each spatial unit; it introduces an optical image change difference map to reflect structural changes such as surface reflectance in the same area; it integrates the two into a product form of an anthropogenic disturbance intensity index, which not only detects whether changes have occurred, but also quantifies the degree of change. It can accurately mark patches affected by human disturbances on a spatial scale; it enables explicit identification of landscape disturbances caused by intervention activities such as engineering construction, farmland reclamation, and road construction; this masking mechanism reduces the dependence on artificially labeled data and improves the adaptability and objectivity of disturbance identification.

[0020] (3) The intensity of natural disturbances and the density of human-induced disturbances are incorporated into a unified calculation framework, and the fused disturbance state is obtained through a combined coupling function. Compared with the traditional approach of analyzing each category separately, this method achieves a fusion assessment of disturbance sources, which can take into account the cumulative effects of sudden and persistent disturbances. Three gradient levels are set: stable state, moderate disturbance state, and strong disturbance state, which improves the accuracy of state identification. It can accurately identify abnormal changes in the boundary ambiguity zone under the background of multiple disturbances, and provide a logical boundary basis for subsequent control strategy formulation and response mechanism triggering.

[0021] By combining parameters such as the perimeter area ratio, shape tortuosity, and adjacent spacing of landscape patches, the response morphology of landscape structure under different disturbance intensities is comprehensively characterized; a landscape spatiotemporal dynamic characterization value (DLS) is constructed to achieve dynamic identification of structural changes at the unit scale; historical comparison is introduced to enhance the causal judgment logic between "disturbance-change".

[0022] (4) Establish a multi-source remote sensing data synchronous acquisition mechanism to support the synchronous access of multi-modal data, including optical remote sensing and radar remote sensing; construct a unified format of raw remote sensing dataset through preprocessing and fitting. This eliminates the asynchronous problem caused by differences in sampling time, spatial resolution and observation angle of different remote sensing platforms; and lays the data foundation for subsequent scale unification, index comparison and time series analysis.

[0023] Heterogeneous remote sensing data are mapped to a unified unit scale to form a spatially consistent dataset; each spatial unit is structurally represented, exhibiting good regional consistency. This improves the spatial alignment accuracy between data from different sources and ensures that subsequent landscape structure parameter extraction results are consistent and comparable in spatial dimensions. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the block diagram of the wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing of the present invention. Figure 2 This is a schematic diagram illustrating the steps of the wetland landscape spatiotemporal dynamic feature analysis method based on multi-source remote sensing of the present invention; Figure 3 This is a flowchart of the interference coupling state value acquisition process of the present invention; Figure 4 This is a trend chart of the spatiotemporal dynamic characterization values ​​of the landscape on which this invention is based. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0026] Example 1 This invention provides a system for analyzing the spatiotemporal dynamic features of wetland landscapes based on multi-source remote sensing. Please refer to [link / reference]. Figures 1 to 4 The module includes a multi-source data synchronous acquisition module, a spatial scale consistency processing module, a landscape structure parameter extraction module, a multi-interference factor analysis and quantification module, a coupled state calculation module, and a wetland landscape spatiotemporal dynamic integration module. The multi-source data synchronous acquisition module collects wetland data through acquisition devices and fits it into the original remote sensing dataset YW; The spatial scale uniformity processing module preprocesses the original remote sensing dataset YW and maps the data to the unit scale to obtain the wetland spatial uniformity set SW; The landscape structure parameter extraction module analyzes each cell of the wetland spatial consistency set SW, extracts landscape structure parameters, and fits them into a landscape feature set FGW. The multi-disturbance factor analysis and quantification module is based on the unit scale of the wetland spatial uniform set SW, constructs the natural disturbance intensity Inat and the anthropogenic disturbance density Ihum, and fits them into the disturbance feature set RW; The coupling state calculation module performs perturbation calculation on the data in the perturbation feature set RW to obtain the perturbation coupling state value Cint; The wetland landscape spatiotemporal dynamic integrated module is based on the disturbance coupling state value Cint, combined with the landscape feature set FGW for unified characterization, constructs the landscape spatiotemporal dynamic representation value DLS, and provides feedback.

[0027] Specifically, the multi-source data synchronous acquisition module and the spatial scale consistency module communicate through local API calls; the spatial scale consistency module and the landscape structure parameter extraction module communicate through memory mapping or JSON-RPC; the spatial scale consistency module and the multiple interference factor analysis module communicate through memory sharing; and the coupled state calculation module and the landscape spatiotemporal dynamic integration module communicate through direct Socket connection. In this embodiment, the system abandons the traditional analytical approach that treats wetland landscapes as static structures or results driven by a single factor. Instead, it constructs a three-tiered framework of landscape structural parameters, disturbance factors, and temporal change responses to capture the dynamic evolution of the landscape under the combined influence of natural and anthropogenic disturbances. Specifically, it comprehensively considers spatial scale mapping and landscape patch feature evolution to adapt to application needs at different regional scales and time dimensions; it no longer focuses on describing the current state but outputs characterization values ​​that reflect evolution trends and process mechanisms, thus expanding the analytical dimensions.

[0028] The system quantifies and superimposes the intensity of natural disturbances and the density of anthropogenic disturbances within a unified data scale and computational framework. By constructing the disturbance coupling state value Cint, it solves the problem of previous independent and fragmented processing of multiple factors. Its beneficial effects include: unified response modeling of the interactions and superposition effects among disturbance factors, reflecting the complex disturbance mechanisms behind landscape changes; identification of key areas with strong coupled disturbances and abrupt structural responses, providing directional guidance for subsequent ecological early warning and intervention; and construction of systematic causal logical relationships, improving the interpretability and predictive accuracy of the analytical model.

[0029] The DLS dynamic characterization values ​​output by the wetland landscape spatiotemporal dynamic integrated module, while preserving the details of structural parameters, reflect the coupling state between disturbance intensity and landscape response, making the landscape state between different regions or different time segments comparable and traceable: at the spatial level, it can locate the clustering or diffusion areas of structural changes caused by disturbance; at the temporal level, it can identify key change nodes such as abrupt change points, transition periods, and recovery stages; and it establishes a quantitative path for the dynamic chain response between disturbance, structural change, and feedback effects, which is conducive to the systematic construction of management decision-making.

[0030] Example 2 This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 and Figure 3 Specifically: the multi-source data synchronous acquisition module includes a multi-source sensing data acquisition unit and a multi-source data time fitting unit; The multi-source sensing data acquisition unit collects wetland data through various acquisition devices, including optical remote sensing image data O(x,y,t), radar image data R(x,y,t), digital elevation model data E(x,y,t), water depth ecological monitoring data WH(t), and land use record data L(x,y,t). Among them, the optical remote sensing image data O(x, y, t) was acquired through multispectral remote sensing satellites; Radar image data R(x, y, t) is acquired through synthetic aperture radar. The digital elevation model data E(x, y, t) was acquired using lidar. Water depth ecological monitoring data WH(t) is acquired through water depth sensors and pressure level gauges; Land use record data L(x, y, t) was acquired through UAV remote sensing and historical remote sensing classification maps; The multi-source data temporal fitting unit performs linear interpolation on the water depth ecological monitoring data WH(t) to fill in missing values; spatially aligns the radar image data R(x,y,t) and optical remote sensing image data O(x,y,t) and projects them onto a unified reference grid; and removes noise from the digital elevation model data E(x,y,t) and land use record data L(x,y,t) and fits them into the original remote sensing dataset YW.

[0031] The spatial scale uniformity processing module includes a spatial geometric correction and resolution resampling unit and a unit scale feature mapping and averaging processing unit; The Spatial Geometric Correction and Resolution Resampling Unit preprocesses the original remote sensing dataset YW, including geometric registration, resolution normalization resampling, and uniform rasterization. The geometric registration process involves performing a geometric transformation on the data in the original remote sensing dataset YW to transform the original image coordinates into the target image coordinates. The resolution normalization resampling process is as follows: linearly resample the data in the original remote sensing dataset YW to a unified spatial resolution step size of r×r. The unified rasterization process is as follows: U i ={(x,y)∣x∈[x i x i+r ), y∈[y i y i+r )}; In the formula, U i Let (x, y) represent the i-th spatial unit, and (x, y) represent the geographic spatial coordinates. i y i ) represents the spatial unit U i The starting coordinates are given by r, which represents the spatial step size.

[0032] In this embodiment, traditional wetland landscape analysis often suffers from problems such as temporal asynchrony, spatial misalignment, and inconsistent resolution of multi-source data, making them difficult to integrate and leading to biased analysis results. This system unifies the acquisition process by incorporating multiple types of data, including optical remote sensing, radar data, digital elevation models, water depth monitoring, and land use data, and fills in missing data in the temporal dimension through linear interpolation, while performing rasterization mapping and coordinate projection unification in the spatial dimension, thus solving the problems of data heterogeneity and inconsistency at the source.

[0033] In this embodiment, multi-source remote sensing data is remapped to standard spatial units using a unified rasterization method. Instead of directly processing the original pixel-level data, the analysis is based on these units, balancing data representation accuracy and processing efficiency. This scale-consistent design enables equal-scale comparisons between different regions horizontally and supports trend overlay analysis of multi-period data vertically.

[0034] Through multi-source sensing and scale unification, the system ultimately formed a well-structured, spatiotemporally coordinated, and noise-removed remote sensing raw dataset, YW. This dataset not only possesses multi-dimensional ecological information such as image texture, topographic relief, vegetation status, water depth changes, and land use dynamics, but also ensures data traceability and regional consistency. Compared to traditional methods that often involve single data sources or single-period image comparisons, this system provides a foundation for systematically tracking the evolution of disturbance effects.

[0035] Example 3 This embodiment is an explanation based on Embodiment 2. Please refer to it. Figure 1 Specifically: Unit-scale feature mapping and averaging processing unit will process spatial unit U i The original remote sensing dataset YW was aggregated and transformed to obtain the wetland spatial consistency set SW. The spatial uniform set SW of wetlands is obtained using the following formula: ; In the formula, SWo(t) represents the o-th data item in the wetland spatial consistency set SW at time t, and YWo(t) represents the o-th data item in the original remote sensing dataset YW at time t.

[0036] The landscape structure parameter extraction module includes a wetland patch boundary identification calculation unit and a landscape structure parameter calculation unit; The wetland patch boundary identification calculation unit uses land use record data L(x, y, t) in spatial unit U. i The set of pixels marked as wetland type is selected from the internal data to form a patch region, and the spatial coverage area of ​​all pixels in the patch is counted to obtain the patch area A(i,t). The outer contour of the patch is extracted by boundary tracking, and the contour length P(i,t) is calculated. The landscape structure parameter calculation unit is based on the patch area A(i,t) and the outline length P(i,t) to calculate the patch perimeter area ratio Rpa, landscape shape tortuosity Csh and the mean patch adjacency distance Dnb, and collects them into a landscape feature set FGW. The patch perimeter area ratio Rpa is obtained by the ratio of patch area A(i,t) to contour length P(i,t); The method for obtaining the landscape shape tortuosity Csh is as follows: First, obtain the current spatial unit U. i The boundary contour of the wetland patch is obtained, and the contour length P(i,t) is obtained. Then, under the same area condition, the theoretical circular boundary length of the ideal circular patch is obtained. Finally, the landscape shape tortuosity Csh is obtained by dividing the contour length P(i,t) by the theoretical circular boundary length. The mean spacing between adjacent patches, Dnb, is obtained as follows: First, determine the current spatial unit U. i The location in space; then, from the wet map layer at the current time t, find the spatial unit U. i Other adjacent units that are also labeled as wetland types are denoted as adjacent patches. For each adjacent patch, the shortest spatial distance between it and the current unit is measured. Then, the total number of wetland patches adjacent to the current unit is counted, i.e., the number of adjacent patches. Finally, the distance values ​​of all adjacent patches are added together and then divided by the number of adjacent patches to obtain the average patch adjacency distance Dnb.

[0037] In this embodiment, multi-source remote sensing data are uniformly mapped to spatial units U through unit-scale feature mapping and averaging processing units. i Based on the core structural unit, dimensionality reduction and aggregation of pixel-level data are achieved, forming a spatially consistent structural representation set. Compared with traditional methods that analyze the original image pixel by pixel, this aggregation processing greatly improves processing efficiency and analytical comparability while preserving spatial feature structure, and is particularly suitable for extracting temporal structural trends from multi-period remote sensing data.

[0038] This embodiment extracts the boundary and contour features of wetland patches by aggregating and filtering wetland type pixels, and further calculates morphological parameters such as perimeter area ratio and tortuosity. This approach goes beyond mere area statistics, combining contour morphology and spatial adjacency to fully reconstruct the geometric features and spatial distribution of patch structures, thereby reflecting the sensitivity and stability of their ecological process responses.

[0039] By aggregating parameters such as area, perimeter, shape tortuosity, and adjacency spacing into a landscape feature set (FGW), this system establishes a characterization system based on structural changes, breaking through the limitations of previous single-dimensional analyses that relied solely on area or type statistics. This feature set supports systematic structural comparisons across multiple time points and regions, enabling the identification of landscape anomalies caused by overlapping disturbances, such as boundary fragmentation, patch shrinkage, or connection disruptions, and tracing their occurrence through time series analysis of the parameters.

[0040] Example 4 This embodiment is an explanation based on Embodiment 3. Please refer to it. Figure 3Specifically: the multi-interference factor analysis and quantification module includes a natural disturbance intensity calculation unit and an anthropogenic activity interference density calculation unit; The natural disturbance intensity calculation unit is based on the wetland spatial uniform set SW. It identifies and calculates the magnitude of natural disturbance changes caused by hydrological, humidity and topographic fluctuations in each spatial unit, including water depth time series change ΔWH and radar wave response ΔR, and fits it into the natural disturbance intensity quantity Inat. The method for obtaining the water depth time series variation ΔWH is as follows: First, find the spatial unit U at the current time t. i The data includes water depth ecological monitoring data; then, water depth ecological monitoring data at a fixed time interval before time t is found, and the difference between the two is calculated, and the absolute value is taken to obtain the spatial unit U. i Water depth time series variation ΔWH; The radar wave response ΔR is obtained as follows: First, find the spatial unit U at the current time t. i The radar image data within the time frame is used; then, radar image data at a fixed time interval before time t is found, and the difference between the two is calculated, and the absolute value is taken to obtain the spatial unit U. i The radar wave response ΔR; The formula for obtaining the natural disturbance intensity Inat is: Inat(i,t)=a1×ΔWH(i,t)+a2×ΔR(i,t); In the formula, Inat(i,t) represents the natural disturbance intensity of the i-th spatial unit at time t, ΔWH(i,t) represents the water depth time series change of the i-th spatial unit at time t, ΔR(i,t) represents the radar wave response of the i-th spatial unit at time t, a1 represents the water depth disturbance adjustment coefficient, and a2 represents the radar disturbance adjustment coefficient.

[0041] The human activity disturbance density calculation unit identifies the degree of disturbance caused by human land use activities in each spatial unit and quantifies it using spatial occupancy density. Detect whether the main land use category has changed, and construct a difference mask ΔL; The difference mask ΔL is obtained as follows: First, find the current spatial unit U. i Find the land use type corresponding to the current time point; then, find the current spatial unit U. i The land use type at the previous observation point in time; if the land use types at the two points in time are different, it indicates that land use has changed in the region during the time period. When a change in land use type is detected, a value of 1 is assigned to the marker, indicating that there is land use disturbance in the spatial unit; When the land use type remains unchanged, the marker is assigned a value of 0, indicating that the spatial unit remains stable during this period; To extract the optical image change difference map ΔPros, the method is as follows: First, find the spatial unit U at the current time t. i The process involves first identifying optical remote sensing imagery data within a given timeframe (t); then finding optical remote sensing imagery data at a fixed time interval prior to time t, calculating the difference between the two, and taking the absolute value to obtain the spatial unit U. i Optical image variation difference diagram ΔPros; The obtained difference mask ΔL and the optical image change difference map ΔPros are combined to calculate the human activity interference density Ihum; The density of human-caused interference, Ihum, is obtained by multiplying the difference mask ΔL and the optical image change difference map ΔPros. By fitting the natural disturbance intensity Inat and the human-induced disturbance density Ihum, the disturbance feature set RW is obtained.

[0042] In this embodiment, by introducing natural disturbance observations reflecting two dimensions—water depth time series changes and radar fluctuations—a response index system sensitive to changes in ecological state is constructed. By combining ecological monitoring data with remote sensing radar information, dynamic capture of natural hydrological and topographic fluctuations is achieved. Each disturbance dimension has physical interpretability and a temporal evolution basis, avoiding misjudgment problems caused by human interference.

[0043] It can precisely depict the ecological response process caused by changes in natural conditions such as rainfall, seasonal water variation, and topographic water accumulation; it provides a basis for distinguishing between natural changes and human intervention for subsequent disturbance attribution; it supports the construction of reference models for natural background disturbances and provides underlying data support for wetland evolution trend analysis.

[0044] By constructing a land use type difference mask, the system accurately identifies whether changes in the main category of human activities have occurred in each spatial unit. An optical image change difference map is introduced to reflect structural changes such as surface reflectance within the same area. These two aspects are then merged into a product-form human disturbance intensity index, which not only detects whether changes have occurred but also quantifies their degree. This allows for precise marking of patches affected by human disturbance at a spatial scale, enabling explicit identification of landscape disturbances caused by interventions such as engineering construction, farmland reclamation, and road construction. This masking mechanism reduces reliance on manually labeled data, improving the adaptability and objectivity of disturbance identification.

[0045] Unlike traditional interference identification methods that treat all changes as uniform phenomena, this approach constructs two subsystems: natural disturbances and human-induced disturbances, which are then converged into a unified interference feature set. The feature set structure is decomposable and combinable, and can be used for both independent analysis and fusion attribution. The fitted interference feature set is not only used to identify interferences, but also to construct disturbance trend models, spatial distribution heat maps, and other multi-level analysis tasks.

[0046] Example 5 This embodiment is an explanation based on Embodiment 4. Please refer to it. Figure 3 and Figure 4 Specifically: the coupling state calculation module includes an interference coupling strength calculation unit and an interference level discrimination unit; The interference coupling strength calculation unit calculates the interference coupling state value Cint based on the disturbance feature set RW; the acquisition method is as follows: The interference coupling state value Cint is obtained by taking the square root of the sum of the square of the natural disturbance intensity Inat and the square of the human-induced disturbance density Ihum. The interference level discrimination unit compares the acquired interference coupling state value Cint with the preset interference threshold Tnt to determine the disturbance state; the determination method is as follows: When the interference coupling state value Cint < 0.5 × interference threshold Tnt, it indicates a stable state; When 0.5 × interference threshold Tnt ≤ interference coupling state value Cint ≤ interference threshold Tnt, it indicates a moderate disturbance state. When the interference threshold Tnt is less than the interference coupling state value Cint, it indicates a strong interference state.

[0047] The wetland landscape spatiotemporal dynamic integrated module includes a disturbance coordinated change analysis unit and a spatiotemporal evolution feedback analysis unit; The disturbance-coordinated change analysis unit is based on the disturbance coupling state value Cint, which couples the structural changes of each spatial unit with the disturbance intensity to calculate and obtain the landscape spatiotemporal dynamic characterization value DLS. The spatial-temporal dynamic representation value of landscape (DLS) is obtained using the following formula: ; In the formula, DLS(i,t) represents the spatiotemporal dynamic characterization value of the i-th spatial unit at time t, Cint(i,t) represents the disturbance coupling state value of the i-th spatial unit at time t, Rpa(i,t) represents the patch perimeter area ratio of the i-th spatial unit at time t, Rpa(i,t-Δt) represents the patch perimeter area ratio of the i-th spatial unit at time t-Δt, Csh(i,t) represents the landscape shape tortuosity of the i-th spatial unit at time t, Dnb(i,t) represents the average patch adjacency spacing of the i-th spatial unit at time t, and b represents the adjacency relationship adjustment coefficient. The spatiotemporal evolution feedback analysis unit performs horizontal and vertical feedback evaluation on the landscape spatiotemporal dynamic characterization value DLS, identifies dynamic change trends, and obtains the time series change trend value ΔDLS. The time series trend value ΔDLS is obtained by the difference between the landscape spatiotemporal dynamic characterization value at time t and the landscape spatiotemporal dynamic characterization value at time t-Δt. When the time series trend value ΔDLS is continuously rising in a positive value, it indicates that the interference effect is continuously increasing and the structure is being disturbed and deconstructed. When the time series trend value ΔDLS is continuously zero and tends to be stable, it indicates that the disturbance is in a balanced state. When the time series trend value ΔDLS is continuously negative, it indicates that the disturbance is weakening or the landscape is returning to its original state. When the time series trend value ΔDLS is in a high-frequency fluctuation state, it indicates that it is in a critical state or an unstable transition state.

[0048] In this embodiment, the intensity of natural disturbances and the density of human-induced disturbances are incorporated into a unified calculation framework, and the fused disturbance state is obtained through a combined coupling function. Compared with the traditional approach of analyzing each category separately, this method achieves a fusion assessment of disturbance sources, taking into account the cumulative effects of both sudden and persistent disturbances. Three gradient levels—stable state, moderate disturbance state, and strong disturbance state—are set to improve the accuracy of state identification. This method accurately identifies abnormal changes in ambiguous boundary zones under multiple disturbance backgrounds, providing a logical demarcation basis for subsequent control strategy formulation and response mechanism triggering.

[0049] By combining parameters such as the perimeter area ratio, shape tortuosity, and adjacency spacing of landscape patches, this study comprehensively characterizes the response morphology of landscape structures under different disturbance intensities. It constructs a landscape spatiotemporal dynamic characterization value (DLS) to dynamically identify structural changes at the unit scale. Historical comparisons are introduced to enhance the causal judgment logic between disturbance and change. This clarifies whether changes in disturbance intensity have damaged, deconstructed, or reconstructed the landscape structure; it supports a one-to-one correspondence between changes in disturbance levels and changes in the morphology of ecological patches, constructing a traceable data chain; and it reflects the sensitivity differences among different landscape patches within a spatial unit, supporting differentiated management.

[0050] The system designs the time-series trend value ΔDLS as a time-series trend indicator, possessing the ability for longitudinal comparison and trend accumulation; it defines four types of feedback states, constructing a state transition model based on static state + dynamic trend; and it can be adapted to multi-period remote sensing data or monitoring data for real-time updates. It can continuously track whether disturbance processes intensify, stabilize, or are recovering; it can provide early warning judgments for rapidly evolving or severely disturbed areas; and it can serve as a basis for dividing wetland structure recovery and degradation periods, supporting dynamic protection strategies.

[0051] Example 6 For a method of analyzing the spatiotemporal dynamic characteristics of wetland landscapes based on multi-source remote sensing, please refer to [reference needed]. Figure 2 Specifically, it includes the following steps: Step 1: The multi-source data synchronous acquisition module collects wetland data through the acquisition device and fits it into the original remote sensing dataset YW; Step 2: The spatial scale uniformity processing module preprocesses the original remote sensing dataset YW and maps the data to the unit scale to obtain the wetland spatial uniformity set SW; Step 3: The landscape structure parameter extraction module analyzes each cell of the wetland spatial consistency set SW, extracts the landscape structure parameters, and fits them into the landscape feature set FGW. Step 4: The multi-disturbance factor analysis and quantification module is based on the unit scale of the wetland spatial consistency set SW, constructs the natural disturbance intensity Inat and the anthropogenic disturbance density Ihum, and fits them into the disturbance feature set RW. Step 5: The coupling state calculation module performs perturbation calculation on the data in the perturbation feature set RW to obtain the perturbation coupling state value Cint; Step Six: The wetland landscape spatiotemporal dynamic integrated module uses the disturbance coupling state value Cint as a basis, combines it with the landscape feature set FGW for unified characterization, constructs the landscape spatiotemporal dynamic representation value DLS, and provides feedback.

[0052] In this embodiment, a multi-source remote sensing data synchronous acquisition mechanism is established to support the synchronous access of multi-modal data, including optical remote sensing and radar remote sensing. A unified format of raw remote sensing dataset is constructed through preprocessing and fitting. This eliminates the asynchronous problems caused by differences in sampling time, spatial resolution, and observation angle among different remote sensing platforms, laying a data foundation for subsequent scale unification, index comparison, and time series analysis.

[0053] Heterogeneous remote sensing data are mapped to a unified unit scale to form a spatially consistent dataset; each spatial unit is structurally represented, exhibiting good regional consistency. This improves the spatial alignment accuracy between data from different sources and ensures that subsequent landscape structure parameter extraction results are consistent and comparable in spatial dimensions.

[0054] The design incorporates extraction logic specific to wetland landscape structures, including patch-scale characteristics, boundary variation features, and shape complexity. It constructs a landscape feature set to reflect the heterogeneity and responsiveness to change within spatial units. This supports a microscopic characterization of the dynamic changes in wetland patches and provides a quantitative indicator basis for determining key ecological characteristics such as ecological connectivity and structural integrity.

[0055] It clearly distinguishes between natural disturbances and human interventions, and constructs separate disturbance intensity and density indices; it supports disturbance distribution analysis based on spatial unit scale. It accurately reflects the type and impact degree of disturbance sources; it helps identify disturbance hotspots and high-frequency areas of intervention behavior, providing data support for subsequent disturbance management. It integrates natural and human disturbances to construct a unified disturbance coupling state index; it supports quantitative analysis of the overall disturbance impact.

[0056] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. A system for analyzing the spatiotemporal dynamic characteristics of wetland landscapes based on multi-source remote sensing, characterized in that: The system includes a multi-source data synchronous acquisition module, a spatial scale consistency processing module, a landscape structure parameter extraction module, a multiple interference factor analysis and quantification module, a coupled state calculation module, and a wetland landscape spatiotemporal dynamic integration module. The multi-source data synchronous acquisition module collects wetland data through acquisition devices and fits it into the original remote sensing dataset YW; The spatial scale uniformity processing module preprocesses the original remote sensing dataset YW and maps the data to the unit scale to obtain the wetland spatial uniformity set SW; The landscape structure parameter extraction module analyzes each cell of the wetland spatial consistency set SW, extracts landscape structure parameters, and fits them into a landscape feature set FGW. The multi-disturbance factor analysis and quantification module is based on the unit scale of the wetland spatial uniform set SW, constructs the natural disturbance intensity Inat and the anthropogenic disturbance density Ihum, and fits them into the disturbance feature set RW; The coupling state calculation module performs perturbation calculation on the data in the perturbation feature set RW to obtain the perturbation coupling state value Cint; The wetland landscape spatiotemporal dynamic integrated module is based on the disturbance coupling state value Cint, combined with the landscape feature set FGW for unified characterization, constructs the landscape spatiotemporal dynamic representation value DLS, and provides feedback.

2. The wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing according to claim 1, characterized in that: The multi-source data synchronous acquisition module includes a multi-source sensing data acquisition unit and a multi-source data time fitting unit; The multi-source sensing data acquisition unit collects wetland data through various acquisition devices, including optical remote sensing image data O(x,y,t), radar image data R(x,y,t), digital elevation model data E(x,y,t), water depth ecological monitoring data WH(t), and land use record data L(x,y,t). Among them, the optical remote sensing image data O(x, y, t) was acquired through multispectral remote sensing satellites; Radar image data R(x, y, t) is acquired through synthetic aperture radar. The digital elevation model data E(x, y, t) was acquired using lidar. Water depth ecological monitoring data WH(t) is acquired through water depth sensors and pressure level gauges; Land use record data L(x, y, t) was acquired through UAV remote sensing and historical remote sensing classification maps; The multi-source data temporal fitting unit performs linear interpolation on the water depth ecological monitoring data WH(t) to fill in missing values; spatially aligns the radar image data R(x,y,t) and optical remote sensing image data O(x,y,t) and projects them onto a unified reference grid; and removes noise from the digital elevation model data E(x,y,t) and land use record data L(x,y,t) and fits them into the original remote sensing dataset YW.

3. The wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing according to claim 2, characterized in that: The spatial scale uniformity processing module includes a spatial geometric correction and resolution resampling unit and a unit scale feature mapping and averaging processing unit; The Spatial Geometric Correction and Resolution Resampling Unit preprocesses the original remote sensing dataset YW, including geometric registration, resolution normalization resampling, and uniform rasterization. The geometric registration process involves performing a geometric transformation on the data in the original remote sensing dataset YW to transform the original image coordinates into the target image coordinates. The resolution normalization resampling process is as follows: linearly resample the data in the original remote sensing dataset YW to a unified spatial resolution step size of r×r. The unified rasterization process is as follows: OR i ={(x, y) | x∈[x i ,x i+r ),y∈[y i ,and i+r )}; In the formula, U i Let (x, y) represent the i-th spatial unit, and (x, y) represent the geographic spatial coordinates. i y i ) represents the spatial unit U i The starting coordinates are given by r, which represents the spatial step size.

4. The wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing according to claim 3, characterized in that: Unit-scale feature mapping and averaging processing unit will process spatial unit U i The original remote sensing dataset YW was aggregated and transformed to obtain the wetland spatial consistency set SW. The spatial uniform set SW of wetlands is obtained using the following formula: ； In the formula, SWo(t) represents the o-th data item in the wetland spatial consistency set SW at time t, and YWo(t) represents the o-th data item in the original remote sensing dataset YW at time t.

5. The wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing according to claim 4, characterized in that: The landscape structure parameter extraction module includes a wetland patch boundary identification calculation unit and a landscape structure parameter calculation unit; The wetland patch boundary identification calculation unit uses land use record data L(x, y, t) in spatial unit U. i The set of pixels marked as wetland type is selected from the internal data to form a patch region, and the spatial coverage area of ​​all pixels in the patch is counted to obtain the patch area A(i,t). The outer contour of the patch is extracted by boundary tracking, and the contour length P(i,t) is calculated. The landscape structure parameter calculation unit is based on the patch area A(i,t) and the outline length P(i,t) to calculate the patch perimeter area ratio Rpa, landscape shape tortuosity Csh and the mean patch adjacency distance Dnb, and collects them into a landscape feature set FGW. The patch perimeter area ratio Rpa is obtained by the ratio of patch area A(i,t) to contour length P(i,t); The method for obtaining the landscape shape tortuosity Csh is as follows: First, obtain the current spatial unit U. i The boundary contour of the wetland patch is obtained, and the contour length P(i,t) is obtained. Then, under the same area condition, the theoretical circular boundary length of the ideal circular patch is obtained. Finally, the landscape shape tortuosity Csh is obtained by dividing the contour length P(i,t) by the theoretical circular boundary length. The mean spacing between adjacent patches, Dnb, is obtained as follows: First, determine the current spatial unit U. i The location in space; then, from the wet map layer at the current time t, find the spatial unit U. i Other adjacent units that are also labeled as wetland types are denoted as adjacent patches. For each adjacent patch, the shortest spatial distance between it and the current unit is measured. Then, the total number of wetland patches adjacent to the current unit is counted, i.e., the number of adjacent patches. Finally, the distance values ​​of all adjacent patches are added together and then divided by the number of adjacent patches to obtain the average patch adjacency distance Dnb.

6. The wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing according to claim 5, characterized in that: The multi-interference factor analysis and quantification module includes a natural disturbance intensity calculation unit and an anthropogenic disturbance density calculation unit; The natural disturbance intensity calculation unit is based on the wetland spatial uniform set SW. It identifies and calculates the magnitude of natural disturbance changes caused by hydrological, humidity and topographic fluctuations in each spatial unit, including water depth time series change ΔWH and radar wave response ΔR, and fits it into the natural disturbance intensity quantity Inat. The method for obtaining the water depth time series variation ΔWH is as follows: First, find the spatial unit U at the current time t. i The data includes water depth ecological monitoring data; then, water depth ecological monitoring data at a fixed time interval before time t is found, and the difference between the two is calculated, and the absolute value is taken to obtain the spatial unit U. i Water depth time series variation ΔWH; The radar wave response ΔR is obtained as follows: First, find the spatial unit U at the current time t. i The radar image data within the time frame is used; then, radar image data at a fixed time interval before time t is found, and the difference between the two is calculated, and the absolute value is taken to obtain the spatial unit U. i The radar wave response ΔR; The formula for obtaining the natural disturbance intensity Inat is: Inat(i,t)=a1×ΔWH(i,t)+a2×ΔR(i,t); In the formula, Inat(i,t) represents the natural disturbance intensity of the i-th spatial unit at time t, ΔWH(i,t) represents the water depth time series change of the i-th spatial unit at time t, ΔR(i,t) represents the radar wave response of the i-th spatial unit at time t, a1 represents the water depth disturbance adjustment coefficient, and a2 represents the radar disturbance adjustment coefficient.

7. The wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing according to claim 6, characterized in that: The human activity disturbance density calculation unit identifies the degree of disturbance caused by human land use activities in each spatial unit and quantifies it using spatial occupancy density. Detect whether the main land use category has changed, and construct a difference mask ΔL; The difference mask ΔL is obtained as follows: First, find the current spatial unit U. i Find the land use type corresponding to the current time point; then, find the current spatial unit U. i The land use type at the previous observation point in time; if the land use types at the two points in time are different, it indicates that land use has changed in the region during the time period. When a change in land use type is detected, a value of 1 is assigned to the marker, indicating that there is land use disturbance in the spatial unit; When the land use type remains unchanged, the marker is assigned a value of 0, indicating that the spatial unit remains stable during this period; To extract the optical image change difference map ΔPros, the method is as follows: First, find the spatial unit U at the current time t. i The process involves first identifying optical remote sensing imagery data within a given timeframe (t); then finding optical remote sensing imagery data at a fixed time interval prior to time t, calculating the difference between the two, and taking the absolute value to obtain the spatial unit U. i Optical image variation difference diagram ΔPros; The obtained difference mask ΔL and the optical image change difference map ΔPros are combined to calculate the human activity interference density Ihum; The density of human-caused interference, Ihum, is obtained by multiplying the difference mask ΔL and the optical image change difference map ΔPros. By fitting the natural disturbance intensity Inat and the human-induced disturbance density Ihum, the disturbance feature set RW is obtained.

8. The wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing according to claim 7, characterized in that: The coupling state calculation module includes an interference coupling strength calculation unit and an interference level discrimination unit; The interference coupling strength calculation unit calculates the interference coupling state value Cint based on the disturbance feature set RW; the acquisition method is as follows: The interference coupling state value Cint is obtained by taking the square root of the sum of the square of the natural disturbance intensity Inat and the square of the human-induced disturbance density Ihum. The interference level discrimination unit compares the acquired interference coupling state value Cint with the preset interference threshold Tnt to determine the disturbance state; the determination method is as follows: When the interference coupling state value Cint < 0.5 × interference threshold Tnt, it indicates a stable state; When 0.5 × interference threshold Tnt ≤ interference coupling state value Cint ≤ interference threshold Tnt, it indicates a moderate disturbance state. When the interference threshold Tnt is less than the interference coupling state value Cint, it indicates a strong interference state.

9. The wetland landscape spatiotemporal dynamic feature analysis system based on multi-source remote sensing according to claim 8, characterized in that: The wetland landscape spatiotemporal dynamic integrated module includes a disturbance coordinated change analysis unit and a spatiotemporal evolution feedback analysis unit; The disturbance-coordinated change analysis unit is based on the disturbance coupling state value Cint, which couples the structural changes of each spatial unit with the disturbance intensity to calculate and obtain the landscape spatiotemporal dynamic characterization value DLS. The spatial-temporal dynamic representation value of landscape (DLS) is obtained using the following formula: ； In the formula, DLS(i,t) represents the spatiotemporal dynamic characterization value of the i-th spatial unit at time t, Cint(i,t) represents the disturbance coupling state value of the i-th spatial unit at time t, Rpa(i,t) represents the patch perimeter area ratio of the i-th spatial unit at time t, Rpa(i,t-Δt) represents the patch perimeter area ratio of the i-th spatial unit at time t-Δt, Csh(i,t) represents the landscape shape tortuosity of the i-th spatial unit at time t, Dnb(i,t) represents the average patch adjacency distance of the i-th spatial unit at time t, and b represents the adjacency relationship adjustment coefficient. The spatiotemporal evolution feedback analysis unit performs horizontal and vertical feedback evaluation on the landscape spatiotemporal dynamic characterization value DLS, identifies dynamic change trends, and obtains the time series change trend value ΔDLS. The time series trend value ΔDLS is obtained by the difference between the landscape spatiotemporal dynamic characterization value at time t and the landscape spatiotemporal dynamic characterization value at time t-Δt. When the time series trend value ΔDLS is continuously rising in a positive value, it indicates that the interference effect is continuously increasing and the structure is being disturbed and deconstructed. When the time series trend value ΔDLS is continuously zero and tends to be stable, it indicates that the disturbance is in a balanced state. When the time series trend value ΔDLS is continuously negative, it indicates that the disturbance is weakening or the landscape is returning to its original state. When the time series trend value ΔDLS is in a high-frequency fluctuation state, it indicates that it is in a critical state or an unstable transition state.

10. A method for analyzing the spatiotemporal dynamic features of wetland landscapes based on multi-source remote sensing, applied to the spatiotemporal dynamic feature analysis system for wetland landscapes based on multi-source remote sensing as described in any one of claims 1 to 9, characterized in that: Includes the following steps: Step 1: The multi-source data synchronous acquisition module collects wetland data through the acquisition device and fits it into the original remote sensing dataset YW; Step 2: The spatial scale uniformity processing module preprocesses the original remote sensing dataset YW and maps the data to the unit scale to obtain the wetland spatial uniformity set SW; Step 3: The landscape structure parameter extraction module analyzes each cell of the wetland spatial consistency set SW, extracts the landscape structure parameters, and fits them into the landscape feature set FGW. Step 4: The multi-disturbance factor analysis and quantification module is based on the unit scale of the wetland spatial consistency set SW, constructs the natural disturbance intensity Inat and the anthropogenic disturbance density Ihum, and fits them into the disturbance feature set RW. Step 5: The coupling state calculation module performs perturbation calculation on the data in the perturbation feature set RW to obtain the perturbation coupling state value Cint; Step Six: The wetland landscape spatiotemporal dynamic integrated module uses the disturbance coupling state value Cint as a basis, combines it with the landscape feature set FGW for unified characterization, constructs the landscape spatiotemporal dynamic representation value DLS, and provides feedback.

Images