Wetland identification method and device based on groundwater depth and multi-source remote sensing data
Patent Information
- Application Number
- CN202610938463.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-26
- Publication Date
- 2026-10-13
Smart Images
Figure CN122903948A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of data processing technology, and more particularly to the field of remote sensing image classification technology, specifically to a wetland identification method and apparatus based on groundwater depth and multi-source remote sensing data. Background Technology
[0002] According to wetland classification, wetlands refer to "natural or artificial, permanent or intermittent marshes, peatlands, and water areas with still or flowing, fresh, brackish, or saline water bodies, including sea areas where the water depth does not exceed 6 meters at low tide." Meanwhile, according to land survey classification, wetlands are defined as "mangrove areas, natural or artificial, permanent or intermittent marshes, peatlands, salt pans, tidal flats, etc." Based on these definitions, wetlands, in addition to the water bodies typically identified by remote sensing, also include non-aquatic surface areas where the groundwater level is close to the surface and the soil water remains saturated for extended periods.
[0003] Current methods for wetland identification based on remote sensing data are mostly based on spectral reflectance characteristics. For example, they often collect multispectral remote sensing images of the target area (concentrated in the visible and near-infrared bands), and use features such as grayscale distribution and texture characteristics to manually delineate water areas and non-water areas based on image processing methods or automatically divide them based on machine learning methods.
[0004] However, while current methods are highly accurate in identifying wetlands with water bodies, they are often less accurate in extracting other types of wetlands. For example, they are quite accurate in classifying water bodies and non-water bodies, but in seasonal wetlands, dry season wetlands, or areas with dense vegetation cover (such as shrub marshes), pure image methods are prone to missing or misjudging, easily misclassifying dry land as wetlands, or classifying dried-up wetlands as non-wetlands.
[0005] Furthermore, the accuracy bottleneck of this method stems from its neglect of the fundamental mechanisms of wetland formation. For instance, the hydrological boundaries of marshland, peatlands, and some former riverbed wetlands are not determined by surface water accumulation, but rather by the relative relationship between groundwater level and topography. These types of wetlands often appear as "non-water bodies" in remote sensing imagery, meaning they are covered by dense vegetation or seasonally dry, causing current methods to miss them and resulting in low wetland identification accuracy. Summary of the Invention
[0006] This disclosure provides a method, apparatus, equipment, and storage medium for wetland identification based on groundwater depth and multi-source remote sensing data.
[0007] According to a first aspect of this disclosure, a wetland identification method based on groundwater depth and multi-source remote sensing data is provided. The method includes: Acquire multi-source remote sensing image data and groundwater level data; The multi-source remote sensing image data and groundwater level data are preprocessed; Based on physical mechanisms, feature extraction was performed on preprocessed multi-source remote sensing image data and groundwater level data to obtain physically aligned surface feature factors and groundwater feature factors. The surface feature factors and groundwater feature factors are fused to obtain a fused feature vector; The fused feature vector is input into a preset classifier, which outputs a wetland classification result map. The preset physical rules are embedded as hard constraints in the decision-making process of the preset classifier. The preset physical rules are set based on the physical consistency relationship between groundwater depth and remote sensing features to force correction of the classification results.
[0008] In addition to the aspects and any possible implementations described above, a further implementation is provided, wherein the multi-source remote sensing image data includes optical image data, radar image data, and topographic image data; and the groundwater level data includes groundwater depth. The preprocessing of the multi-source remote sensing image data and groundwater level data includes: Initial preprocessing is performed on the multi-source remote sensing image data and groundwater level data respectively; Based on the topographic humidity index in the preprocessed multi-source remote sensing image data, spatial interpolation is performed on the groundwater depth to generate a groundwater depth raster layer with the same spatial resolution as the multi-source remote sensing image data.
[0009] In addition to the aspects and any possible implementations described above, a further implementation is provided, wherein the physical mechanism-based feature extraction of preprocessed multi-source remote sensing image data and groundwater level data to obtain physically aligned surface feature factors and groundwater feature factors includes: Surface feature factors were constructed based on the preprocessed multi-source remote sensing image data; Based on the groundwater depth raster layer, groundwater feature factors are constructed according to the preprocessed groundwater level data. The surface characteristic factors include optical characteristics, radar characteristics, and temporal dynamic characteristics; the optical characteristics include the normalized vegetation index, the improved normalized water index, and the modified soil-regulated vegetation index; the radar characteristics include the dual-polarization backscattering coefficient and polarization decomposition parameters; the temporal dynamic characteristics include the maximum, minimum, mean, standard deviation, and phase of each characteristic index over time; and the groundwater characteristic factors include groundwater level depth and flooding frequency.
[0010] As described above and in any possible implementation, a further implementation is provided, wherein the preset physical rules include: When the groundwater depth is less than or equal to the preset minimum threshold and the normalized vegetation index is greater than the first threshold, the pixel is determined to be a wetland. When the groundwater depth is greater than the preset maximum threshold and the improved normalized water index is greater than the second threshold, the pixel is determined based on the radar backscattering coefficient. When the groundwater depth is less than the preset maximum threshold, the time difference between the peak time of the improved normalized water index and the valley time of the groundwater depth is less than or equal to the preset number of days, and the average value of the valley and peak values of the groundwater depth is greater than the third threshold, the humidity confidence of this pixel is set to high.
[0011] In addition to the aspects and any possible implementations described above, a further implementation is provided in which the preset classifier is a random forest classifier, used to run the preset physical rules in the post-processing stage of the random forest classifier.
[0012] In addition to the aspects and any possible implementations described above, a further implementation is provided, wherein the method further includes: The accuracy of the wetland classification results map was verified based on independent field survey points. Morphological opening and closing operations are used to optimize the boundary of the wetland classification result map based on the groundwater depth contour lines; the groundwater depth contour lines are extracted based on the groundwater depth raster layer.
[0013] In addition to the aspects described above and any possible implementations, a further implementation is provided in which the wetland classification result map includes a classification confidence layer for distinguishing between wetlands, seasonal wetlands and non-wetlands.
[0014] According to a second aspect of this disclosure, a wetland identification device based on groundwater depth and multi-source remote sensing data is provided. The device includes: The acquisition module is used to acquire multi-source remote sensing image data and groundwater level data; The preprocessing module is used to preprocess the multi-source remote sensing image data and groundwater level data; The extraction module is used to extract features from preprocessed multi-source remote sensing image data and groundwater level data based on physical mechanisms, and obtain physically aligned surface feature factors and groundwater feature factors. The fusion module is used to perform feature fusion on the surface feature factors and groundwater feature factors to obtain a fused feature vector. The generation module is used to input the fused feature vector into a preset classifier and output a wetland classification result map; the preset physical rules are embedded as hard constraints in the decision-making process of the preset classifier; the preset physical rules are set based on the physical consistency relationship between groundwater depth and remote sensing features to force correction of the classification results.
[0015] According to a third aspect of this disclosure, an electronic device is provided. The electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method described above.
[0016] According to a fourth aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the method described above.
[0017] This application provides a wetland identification method based on groundwater depth and multi-source remote sensing data. The method acquires and preprocesses multi-source remote sensing image data and groundwater level data. Based on a physical mechanism, it extracts features from the preprocessed multi-source remote sensing image data and groundwater level data to obtain physically aligned surface feature factors and groundwater feature factors. It then fuses these surface and groundwater feature factors to obtain a fused feature vector. This fused feature vector is input into a preset classifier, which outputs a wetland classification result map. The preset classifier's decision-making process incorporates preset physical rules as hard constraints. These rules are set based on the physical consistency relationship between groundwater depth and remote sensing features to forcibly correct the classification results. Based on this, it can integrate multi-source remote sensing data, employ machine learning algorithms, and introduce groundwater level constraints. This introduces the causal control effect of groundwater level into the identification model, effectively solving the problem of diverse and difficult-to-identify non-water body wetland features in remote sensing wetland identification, thereby improving wetland identification accuracy.
[0018] It should be understood that the description in the Summary of the Invention section is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description
[0019] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of this disclosure. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 A flowchart of a wetland identification method based on groundwater depth and multi-source remote sensing data according to an embodiment of the present disclosure is shown; Figure 2 A flowchart is shown for another wetland identification method based on groundwater depth and multi-source remote sensing data according to an embodiment of the present disclosure; Figure 3 A block diagram of a wetland identification device based on groundwater depth and multi-source remote sensing data according to an embodiment of the present disclosure is shown; Figure 4 A block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.
[0021] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.
[0022] In this disclosure, multi-source remote sensing data can be integrated, machine learning algorithms can be used, and groundwater level constraints can be introduced. That is, the cause and control of groundwater level are introduced into the identification model, which effectively solves the problem of the diverse and difficult identification of non-water wetland characteristics in remote sensing wetland identification, thereby improving the wetland identification accuracy.
[0023] Figure 1 A flowchart of a wetland identification method 100 based on groundwater depth and multi-source remote sensing data according to an embodiment of the present disclosure is shown.
[0024] In box 110, acquire multi-source remote sensing image data and groundwater level data.
[0025] In some embodiments, time-series remote sensing image data of the target area over a complete hydrological year (or at least including the wet and dry seasons) can be acquired as multi-source remote sensing image data.
[0026] In some embodiments, groundwater level data can be obtained by collecting daily (or ten-day) burial depth data of all groundwater monitoring wells of various levels within the target area.
[0027] In box 120, preprocessing is performed on multi-source remote sensing image data and groundwater level data.
[0028] In some embodiments, multi-source remote sensing image data may include optical image data, radar image data, and topographic image data.
[0029] In some embodiments, groundwater level data may include groundwater depth.
[0030] In some embodiments, the preprocessing of multi-source remote sensing image data and groundwater level data specifically includes: Initial preprocessing was performed on multi-source remote sensing image data and groundwater level data, respectively. Based on the topographic humidity index in the preprocessed multi-source remote sensing image data, spatial interpolation is performed on the groundwater depth to generate a groundwater depth raster layer with the same spatial resolution as the multi-source remote sensing image data.
[0031] In some embodiments, initial preprocessing is performed on multi-source remote sensing image data and groundwater level data, specifically including: Initial preprocessing of optical image data includes: Data from Sentinel-2 (Sentinel-2A and Sentinel-2B satellites) or Landsat-8 / 9 satellites can be selected for preprocessing such as radiometric calibration, atmospheric correction, and cloud masking. Atmospheric correction can be performed using the Second Simulation of the Satellite Signal in the Solar Spectrum (6S) or the Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH) model. Initial preprocessing of radar image data includes: Sentinel-1 dual-polarization (Vertical-Vertical Polarization (VV) + Vertical-Horizontal Polarization (VH)) data can be selected. Preprocessing includes orbit correction, radiometric calibration, multi-look processing, terrain correction, and refined Lee filtering. Among these, terrain correction can use the Shuttle RadarTopography Mission Digital Elevation Model (SRTM DEM). Initial preprocessing of the terrain image data includes: Acquire a 12.5m digital elevation model from the Advanced Land Observing Satellite (ALOS) or a 30m digital elevation model from the Shuttle Radar Topography Mission (SRTM), and calculate topographic factors: slope, aspect, and topographic wetness index (TWI). Initial pretreatment of groundwater depth includes: For each monitoring well, the following hydrological characteristic values are calculated: Multi-year average high water level depth: The average of several values of the highest water level (minimum depth) in the time series; Multi-year average low water level depth: The average of several values of the lowest water level (maximum depth) in the time series.
[0032] In some embodiments, to solve the spatial matching problem between point data and area images, terrain-constrained co-kriging interpolation can be used.
[0033] Specifically, the topographic humidity index (TWI) extracted above is used as a covariate. Since the topographic humidity index reflects the tendency of water accumulation under gravity, it is used to constrain the groundwater level depth through interpolation, generating a groundwater depth raster layer with the same spatial resolution as the remote sensing image.
[0034] In box 130, feature extraction is performed on the preprocessed multi-source remote sensing image data and groundwater level data based on physical mechanisms to obtain physically aligned surface feature factors and groundwater feature factors.
[0035] In some embodiments, surface feature factors include optical features, radar features, and temporal dynamic features.
[0036] In some embodiments, optical features include the Normalized Difference Vegetation Index (NDVI), the Modified Normalized Difference Water Index (MNDWI), and the Optimized Soil-Adjusted Vegetation Index (OSAVI). NDVI = (NIR - R) / (NIR + R), which is the most commonly used vegetation index; denser vegetation corresponds to a higher index value. MNDWI = (Green - SWIR) / (Green + SWIR), this index more clearly displays the fine features of water bodies through the shortwave infrared band, effectively distinguishing shadowed areas from actual water bodies in images. OSAVI = 1.16 × (NIR - R) / (NIR + R + 0.16), which is more suitable for vegetation health diagnosis in sparse vegetation conditions. In the above formula, NIR is the near-infrared band reflectance, R is the red band reflectance, Green is the green band reflectance, and SWIR is the short-wave infrared band reflectance.
[0037] In some embodiments, radar features include dual-polarization backscattering coefficients and polarization decomposition parameters; wherein, the backscattering coefficients of VV and VH polarizations and the polarization decomposition parameters (such as volume scattering components) can be calculated, and radar features play a key role in penetrating vegetation to sense surface soil moisture.
[0038] In some embodiments, the time-series dynamic features include the maximum, minimum, mean, standard deviation, and phase (time of maximum and minimum values) of each feature index (such as optical features and radar features) over a time series, which can be calculated.
[0039] In some embodiments, groundwater characteristic factors include groundwater level depth and flooding frequency.
[0040] In some embodiments, the above-mentioned physical mechanism-based feature extraction of preprocessed multi-source remote sensing image data and groundwater level data yields physically aligned surface feature factors and groundwater feature factors, specifically including: Surface feature factors were constructed based on the preprocessed multi-source remote sensing image data; Based on the groundwater depth raster layer, groundwater characteristic factors are constructed according to the preprocessed groundwater level data. In some embodiments, preprocessed multi-source remote sensing image data, i.e., time-series optical and radar images, can be used to construct a time-series feature dataset, i.e., surface feature factors.
[0041] In some embodiments, the following spatially explicit hydrological feature factors can be constructed based on the groundwater depth raster layer generated above: Groundwater level depth: The vertical distance from the Earth's surface to the groundwater level; Flooding frequency: The proportion of days with a groundwater level ≤ 0 (i.e., the water level is higher than the ground surface) out of the total number of days.
[0042] In box 140, feature fusion is performed on surface feature factors and groundwater feature factors to obtain a fused feature vector.
[0043] In some embodiments, the surface feature factors generated above ( F RS ) and generated groundwater characteristic factors ( F GW Resampling is performed to ensure consistent pixel scale, and the pixels are then concatenated to form a fused feature vector. F Fusion =[ F RS , F GW ].
[0044] In box 150, the fused feature vector is input into the preset classifier, and the wetland classification result map is output. The preset physical rules are embedded as hard constraints in the decision-making process of the preset classifier. The preset physical rules are set based on the physical consistency relationship between groundwater depth and remote sensing features to force the correction of classification results.
[0045] In some embodiments, the preset classifier can be set based on the user's actual needs. When constructing samples, it can build three types of training samples for non-aquatic surfaces, such as wetlands, non-wetlands, and seasonal wetlands. Among them, the groundwater level depth being less than a set threshold (generally 0.2m) all year round serves as a mandatory criterion for wetland identification.
[0046] In some embodiments, the aforementioned preset physical rules include: When the groundwater depth is less than or equal to the preset minimum threshold and the normalized vegetation index is greater than the first threshold, the pixel is determined to be a wetland. When the groundwater depth is greater than the preset maximum threshold and the improved normalized water index is greater than the second threshold, the pixel is determined based on the radar backscattering coefficient. When the groundwater depth is less than the preset maximum threshold, the time difference between the peak time of the improved normalized water index and the valley time of the groundwater depth is less than or equal to the preset number of days, and the average value of the valley and peak values of the groundwater depth is greater than the third threshold, the humidity confidence of this pixel is set to high.
[0047] In some embodiments, the preset minimum threshold, preset maximum threshold, preset number of days, first threshold, second threshold, and third threshold can be set based on the user's actual needs, combined with experience and literature references, as well as the average depth of plant root distribution.
[0048] For example, you can set the preset minimum threshold to 0.2m, the preset maximum threshold to 2m, the preset number of days to 15 days, the first threshold to 0.2, the second threshold to 0.3, and the third threshold to a minimum set threshold, such as 0.2m.
[0049] In some embodiments, based on the threshold setting described above, the embedded logical judgment rules during the classifier decision-making process may include: Rule A (Mandatory Wetlands): If the groundwater depth of a certain pixel is... D GW ≤0.2 m If the groundwater level is extremely shallow and the NDVI of the pixel is greater than 0.2, even if the MNDWI shows a negative value (not water body), the pixel will be forcibly identified as a "permanent wetland". Rule B (Excluding False Wetlands): If a certain pixel's D GW If the groundwater level is greater than 2.0m, and the MNDWI of the pixel is greater than 0.3, it is necessary to make a judgment based on the radar backscattering coefficient. For example, if the radar co-polarization phase difference shows that the area is bare soil or buildings, then the pixel should be forcibly excluded from the wetland category. Rule C (Seasonal Adjustment): If a certain pixel's D GW <2.0m, for time-series images, extract the MNDWI peak time. T RS Groundwater depth valley value time T GWH (Highest water level), peak groundwater depth time T GWL (Lowest water level) and the average of the valley and peak values of groundwater depth A GW If | T RS - T GWH |≤15 and A GW >0.2 m If the confidence level is high, then the wetland confidence level of that pixel is set to high; otherwise, it is set to low.
[0050] It should be noted that Rule A is based on the principle that the area is a typical "gleyed" wetland, shrub marsh, or forest marsh, with no surface water but saturated soil. Pure spectral methods may misclassify it as dryland woodland. This rule can resolve the following misidentification scenarios and improve identification accuracy: 1) Some wetlands have high vegetation cover, making the water bodies under the canopy invisible, thus misclassifying the wetlands as grasslands; 2) In large areas of discontinuous wetlands (such as national wetland parks and nature reserves), a small portion of the wetland surface is dry, so the wetland is mistakenly identified as grassland or bare land.
[0051] The principle of rule B is as follows: small bubbles formed by short-term heavy rainfall on bare land will cause the corresponding pixels in the remote sensing image to have a high MNDWI value. Based solely on the remote sensing image, the bare land may be misjudged as wetland or water body. In this case, combining SAR data can eliminate remote sensing misjudgments caused by "shadows" or "high water content of bare soil (after rain)".
[0052] Rule C is based on the principle that, given the significant differences in surface water and groundwater characteristics between the wet and dry seasons in some watersheds, single-period characteristics are insufficient to accurately determine whether a site is a wetland. This rule addresses this issue and improves identification accuracy by including: Seasonal wetlands dry out during the dry season, resulting in lower MNDWI values, which may even be negative. Furthermore, the groundwater level is buried at a greater depth, which does not meet the requirements of rules A and B. During the high-water season, both surface water and groundwater are abundant, resulting in a high MNDWI and shallow groundwater depth, which can be easily confused with permanent wetlands. MNDWI peak time T RS Time value of groundwater buried deep valley T GW The short interval condition limits the time, and the average water level during the wet and dry seasons is greater than the third threshold to distinguish it from permanent wetlands. The accuracy of seasonal wetland identification is improved by judging multiple time series.
[0053] In some embodiments, the preset classifier is a random forest classifier, used to run preset physical rules in the post-processing stage of the random forest classifier.
[0054] In some embodiments, groundwater data can be used as a "hard constraint" rather than merely as an input feature by introducing a "random forest classifier with prior knowledge constraints".
[0055] It should be noted that when running preset physical rules in the post-processing stage of the random forest classifier, the classification results obtained from the corresponding pre-processing stage can be adopted in a "one-to-one" form, that is, "if the pre-processing stage classification result is wetland, only rule A is run in the post-processing stage; if the pre-processing stage classification result is non-wetland, only rule B is run in the post-processing stage; if the pre-processing stage classification result is seasonal wetland, only rule C is run in the post-processing stage".
[0056] To further improve recognition accuracy, when running preset physical rules in the post-processing stage of the random forest classifier, the classification results obtained from the corresponding pre-processing stage can be adopted in a "one-to-many" form, that is, "if the classification result of the pre-processing stage is wetland, then rules A, B, and C are run in the post-processing stage; the classification results of seasonal wetlands and non-wetlands are executed in the same way."
[0057] Based on this, the fused feature vectors are input into the random forest classifier, and the above physical rules are run in the post-processing stage of the classifier to further optimize the classification results, and finally output the wetland classification result map.
[0058] In some embodiments, the above wetland classification result map includes a classification confidence layer to distinguish between wetlands, seasonal wetlands and non-wetlands.
[0059] For example, the fused feature vector is first used as input, and three types of samples are fed into the random forest classifier, which will distinguish between wetlands, non-wetlands, and seasonal wetlands. The classification results are further optimized and adjusted based on the above three rules. The classification result map is output, along with a classification confidence layer, to distinguish between wetlands, seasonal wetlands, and non-wetlands, making it easier for users to intuitively obtain the recognition results.
[0060] According to the embodiments of this disclosure, the following technical effects are achieved: By acquiring and preprocessing multi-source remote sensing image data and groundwater level data, and then extracting features from the preprocessed multi-source remote sensing image data and groundwater level data based on physical mechanisms, physically aligned surface feature factors and groundwater feature factors are obtained. These surface feature factors and groundwater feature factors are then fused to obtain a fused feature vector. This fused feature vector is input into a pre-defined classifier, which outputs a wetland classification result map. Pre-defined physical rules are embedded as hard constraints in the decision-making process of the pre-defined classifier. These rules are set based on the physical consistency relationship between groundwater depth and remote sensing features to forcibly correct the classification results. Based on this, multi-source remote sensing data can be integrated, and machine learning algorithms can be used to introduce groundwater level constraints, that is, the causal control effect of groundwater level is introduced into the recognition model. This effectively solves the problem of diverse and difficult-to-identify non-water body wetland features in remote sensing wetland identification, thereby improving the accuracy of wetland identification.
[0061] In some embodiments, the above method further includes: Accuracy verification of wetland classification results maps was performed based on independent field survey points. Morphological opening and closing operations were used to optimize the boundaries of the wetland classification map based on the groundwater depth contour lines; the groundwater depth contour lines were extracted based on the groundwater depth raster layer.
[0062] In some embodiments, a confusion matrix can be constructed using independent field survey points (GPS records) to calculate overall accuracy, Kappa coefficient, and user accuracy, thereby verifying the accuracy of the wetland classification result map.
[0063] In some embodiments, morphological opening and closing operations can be used for post-processing of the "salt and pepper noise" in the classification results. In particular, the downstream boundary of the wetland is forcibly cut using the groundwater depth contour lines generated above, ensuring that the wetland is logically consistent with the boundary of the topographic catchment area, thus completing the boundary smoothing process.
[0064] Figure 2 A flowchart is shown for another wetland identification method based on groundwater depth and multi-source remote sensing data according to an embodiment of the present disclosure.
[0065] like Figure 2 As shown, wetland identification methods based on groundwater depth and multi-source remote sensing data may also include: Step 1: Data Acquisition and Preprocessing: Acquisition and preprocessing of multi-source remote sensing image data, and acquisition and interpolation of groundwater level data; Step 2: Feature extraction based on physical mechanisms: extraction of surface feature factors and extraction of groundwater feature factors; Step 3: Multi-source data fusion and physical constraint classification: feature fusion and intelligent classification based on physical consistency constraints; Step 4: Accuracy verification and boundary optimization: accuracy verification and boundary smoothing.
[0066] In summary, unlike previous methods that extract ground image features through remote sensing, this disclosure proposes a wetland identification method that combines surface and subsurface data and utilizes multi-source remote sensing data, thus solving the problem of some wetlands being easily missed from a physical mechanism perspective.
[0067] It should be noted that this disclosure can also be applied to the fields of water resources investigation, natural resources investigation and monitoring, and geographic information technology, such as land use type investigation, wetland park protection, ecological environment protection and planning geographic information services, etc.
[0068] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this disclosure is not limited to the described order of actions, because according to this disclosure, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this disclosure.
[0069] The above is an introduction to the method embodiments. The following describes the solution described in this disclosure further through device embodiments.
[0070] Figure 3 A block diagram of a wetland identification device 300 based on groundwater depth and multi-source remote sensing data according to an embodiment of the present disclosure is shown. Figure 3 As shown, the device 300 includes: The acquisition module 310 is used to acquire multi-source remote sensing image data and groundwater level data; Preprocessing module 320 is used to preprocess multi-source remote sensing image data and groundwater level data; The extraction module 330 is used to extract features from the preprocessed multi-source remote sensing image data and groundwater level data based on physical mechanisms, and obtain physically aligned surface feature factors and groundwater feature factors. The fusion module 340 is used to fuse surface feature factors and groundwater feature factors to obtain a fused feature vector. The generation module 350 is used to input the fused feature vector into the preset classifier and output the wetland classification result map. The preset physical rules are embedded as hard constraints in the decision-making process of the preset classifier. The preset physical rules are set based on the physical consistency relationship between groundwater depth and remote sensing features to force the correction of classification results.
[0071] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the described module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0072] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.
[0073] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.
[0074] Figure 4A block diagram of an exemplary electronic device 400 capable of implementing embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.
[0075] Electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in ROM 402 or a computer program loaded into RAM 403 from storage unit 408. RAM 403 can also store various programs and data required for the operation of electronic device 400. The computing unit 401, ROM 402, and RAM 403 are interconnected via bus 404. I / O interface 405 is also connected to bus 404.
[0076] Multiple components in electronic device 400 are connected to I / O interface 405, including: input unit 406, such as keyboard, mouse, etc.; output unit 407, such as various types of displays, speakers, etc.; storage unit 408, such as disk, optical disk, etc.; and communication unit 409, such as network card, modem, wireless transceiver, etc. Communication unit 409 allows electronic device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0077] The computing unit 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of method 100 described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform method 100 by any other suitable means (e.g., by means of firmware).
[0078] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0079] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0080] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0081] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including voice input, speech input, or tactile input).
[0082] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.
[0083] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.
[0084] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.
[0085] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.
Claims
1. A wetland identification method based on groundwater depth and multi-source remote sensing data, characterized in that, include: Acquire multi-source remote sensing image data and groundwater level data; The multi-source remote sensing image data and groundwater level data are preprocessed; Based on physical mechanisms, feature extraction was performed on preprocessed multi-source remote sensing image data and groundwater level data to obtain physically aligned surface feature factors and groundwater feature factors. The surface feature factors and groundwater feature factors are fused to obtain a fused feature vector; The fused feature vector is input into a preset classifier, which outputs a wetland classification result map. The preset physical rules are embedded as hard constraints in the decision-making process of the preset classifier. The preset physical rules are set based on the physical consistency relationship between groundwater depth and remote sensing features to force correction of the classification results.
2. The method according to claim 1, characterized in that, The multi-source remote sensing image data includes optical image data, radar image data, and topographic image data; the groundwater level data includes groundwater depth. The preprocessing of the multi-source remote sensing image data and groundwater level data includes: Initial preprocessing is performed on the multi-source remote sensing image data and groundwater level data respectively; Based on the topographic humidity index in the preprocessed multi-source remote sensing image data, spatial interpolation is performed on the groundwater depth to generate a groundwater depth raster layer with the same spatial resolution as the multi-source remote sensing image data.
3. The method according to claim 1, characterized in that, The physical mechanism-based feature extraction of preprocessed multi-source remote sensing image data and groundwater level data yields physically aligned surface feature factors and groundwater feature factors, including: Surface feature factors were constructed based on the preprocessed multi-source remote sensing image data; Based on the groundwater depth raster layer, groundwater feature factors are constructed according to the preprocessed groundwater level data. The surface characteristic factors include optical characteristics, radar characteristics, and temporal dynamic characteristics; the optical characteristics include the normalized vegetation index, the improved normalized water index, and the modified soil-regulated vegetation index; the radar characteristics include the dual-polarization backscattering coefficient and polarization decomposition parameters; the temporal dynamic characteristics include the maximum, minimum, mean, standard deviation, and phase of each characteristic index over time; and the groundwater characteristic factors include the groundwater level depth and flooding frequency.
4. The method according to claim 3, characterized in that, The preset physical rules include: When the groundwater depth is less than or equal to the preset minimum threshold and the normalized vegetation index is greater than the first threshold, the pixel is determined to be a wetland. When the groundwater depth is greater than the preset maximum threshold and the improved normalized water index is greater than the second threshold, the pixel is determined based on the radar backscattering coefficient. When the groundwater depth is less than the preset maximum threshold, the time difference between the peak time of the improved normalized water index and the valley time of the groundwater depth is less than or equal to the preset number of days, and the average value of the valley and peak values of the groundwater depth is greater than the third threshold, the humidity confidence of this pixel is set to high.
5. The method according to claim 4, characterized in that, The preset classifier is a random forest classifier, which is used to run the preset physical rules in the post-processing stage of the random forest classifier.
6. The method according to any one of claims 2 to 5, characterized in that, The method further includes: The accuracy of the wetland classification results map was verified based on independent field survey points. Morphological opening and closing operations are used to optimize the boundary of the wetland classification result map based on the groundwater depth contour lines; the groundwater depth contour lines are extracted based on the groundwater depth raster layer.
7. The method according to any one of claims 1 to 5, characterized in that, The wetland classification result map includes a classification confidence layer, which is used to distinguish between wetlands, seasonal wetlands and non-wetlands.
8. A wetland identification device based on groundwater depth and multi-source remote sensing data, characterized in that, include: The acquisition module is used to acquire multi-source remote sensing image data and groundwater level data; The preprocessing module is used to preprocess the multi-source remote sensing image data and groundwater level data; The extraction module is used to extract features from preprocessed multi-source remote sensing image data and groundwater level data based on physical mechanisms, and obtain physically aligned surface feature factors and groundwater feature factors. The fusion module is used to perform feature fusion on the surface feature factors and groundwater feature factors to obtain a fused feature vector. The generation module is used to input the fused feature vector into a preset classifier and output a wetland classification result map; the preset physical rules are embedded as hard constraints in the decision-making process of the preset classifier; the preset physical rules are set based on the physical consistency relationship between groundwater depth and remote sensing features to force correction of the classification results.
9. An electronic device, characterized in that, include: At least one processor; as well as A memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in any one of claims 1-7.
10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-7.