Water regimen disaster monitoring system and method

Abstract

The invention discloses a water regimen disaster monitoring system and method, and relates to the technical field of synthetic aperture imaging monitoring. The method comprises the following steps: marking a high-risk terrain shadow area and a medium-risk terrain attenuation area in a pre-collected main SAR image according to a preset risk threshold value; screening pixels marked as a high-risk terrain shadow area and a medium-risk terrain attenuation area in the main SAR image to obtain candidate terrain pseudo dark areas; extracting a crop row structure pseudo dark area from the main SAR image; according to a standby SAR image collected in advance, pixels in the main SAR image are marked as a visual angle sensitive pseudo dark area or a potential water body; pixels marked as candidate terrain pseudo dark areas, crop row structure pseudo dark areas, visual angle sensitive pseudo dark areas and potential water bodies in the main SAR image are subjected to preset elimination and retention operation, and then a determined water body and a to-be-verified area are obtained; reliable water body distribution information is provided for flood disaster relief, resource mismatching is avoided, and efficient development of disaster relief work is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of synthetic aperture imaging monitoring, and more particularly to a water regime disaster monitoring system and method. BACKGROUND

[0002] In the mountainous farmland area, when using synthetic aperture radar to carry out flood monitoring, complex terrain and crop planting environment are often faced. These areas are mostly undulating slopes, terraces, and also plant corn, grape and other strip-shaped crops. The row direction of the crops and the slope gradient of the terrain jointly constitute a special ground structure. The current technology for synthetic aperture radar flood monitoring is based on the side-looking geometric imaging characteristics of radar. By analyzing the backscattering intensity of the ground reflection, water bodies are identified. Generally, the area with low backscattering intensity is determined as the waterlogged area. Because the specular reflection of the water surface will weaken the radar echo, it appears as a dark tone.

[0003] However, the existing technology has obvious limitations in the mountainous farmland scene. When the slope direction of the mountain slope and terrace or the row direction of the strip-shaped crops is parallel to the radar line of sight, the shielding effect caused by the terrain undulation or the directional structure formed by the crop rows will greatly weaken the radar echo signal. These areas appear as large dark tones on the radar amplitude map, which is very similar to the actual water body or radar shadow.

[0004] Since the existing technology only uses low backscattering intensity as the core basis for identifying water bodies, it cannot distinguish between dark areas caused by terrain and crop structure and real waterlogged areas, resulting in a large number of misjudgments. Such misjudgments will directly reduce the reliability of the flood distribution map, causing rescue personnel to invest limited manpower, resources and other valuable resources into areas that are not submerged, while the places that are actually threatened by floods miss the best rescue opportunity due to insufficient attention. More importantly, the traditional method lacks quantitative analysis of the false positive rate with terrain slope and aspect, and does not reduce misjudgments through effective terrain correction methods, so this problem has existed for a long time.

[0005] In view of this, the present application provides a water regime disaster monitoring system and method to solve the above problems. SUMMARY

[0006] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned purposes, the present application provides the following technical scheme: a water regime disaster monitoring method, comprising:

[0007] According to the preset risk threshold, mark the high-risk terrain shadow area and the medium-risk terrain attenuation area in the pre-collected main SAR image;

[0008] From the pixels marked as high-risk terrain shadow area and medium-risk terrain attenuation area in the main SAR image, candidate terrain false dark areas are screened.

[0009] The pseudo-dark region of crop row structure was extracted from the main SAR image;

[0010] Based on the pre-acquired backup SAR images, the pixels in the main SAR images are marked as view-sensitive false dark areas or potential water bodies;

[0011] After performing preset removal and retention operations on the pixels marked as candidate terrain false dark areas, crop row structure false dark areas, view-sensitive false dark areas and potential water bodies in the main SAR image, the identified water bodies and areas to be verified are obtained.

[0012] The identified water bodies and areas to be verified were then validated to obtain a water body distribution map.

[0013] Furthermore, methods for marking high-risk terrain shadow areas and medium-risk terrain attenuation areas in pre-acquired master SAR images include:

[0014] Based on the slope and aspect of each pixel in the pre-calculated main SAR image, the relative angle between the slope and the radar line of sight is calculated in combination with SAR imaging geometric parameters; the SAR imaging geometric parameters include the incident angle and the radar azimuth angle.

[0015] The risk thresholds include the upper limit of the slope threshold, the lower limit of the relative angle threshold, and the middle limit of the slope threshold;

[0016] Pixels in the main SAR image that meet the conditions of having a slope greater than or equal to the upper limit of the slope threshold and a relative angle between the slope and the radar line of sight less than or equal to the lower limit of the relative angle threshold are marked as high-risk terrain shadow areas.

[0017] Pixels in the main SAR image that meet the conditions of having a slope between the middle and upper limits of the slope threshold and a relative angle between the slope and the radar line of sight that is less than or equal to the lower limit of the relative angle threshold are marked as medium-risk terrain attenuation zones.

[0018] Furthermore, methods for obtaining candidate terrain pseudo-dark areas include:

[0019] The main SAR image is corrected by a preset terrain radiometric correction model to obtain the corrected main SAR image and the corrected backscattering coefficients of each pixel.

[0020] The average value of the corrected backscattering coefficients of all pixels in the high-risk terrain shadow area and the medium-risk terrain attenuation area in the corrected main SAR image is denoted as the backscattering mean.

[0021] If the mean backscatter value of the corresponding area is lower than the set normal backscatter threshold, the corresponding area will be included in the candidate terrain pseudo-dark area.

[0022] Furthermore, methods for obtaining pseudo-dark areas in crop row structure include:

[0023] Multiple dark pixel sequences parallel to the crop row direction in the main SAR image were extracted by directional Gabor filtering.

[0024] Record the number of pixels in each dark fringe pixel sequence, combine it with the spatial resolution of the main SAR image, calculate the actual area corresponding to a single pixel, and then multiply the number of pixels in each dark fringe pixel sequence by the actual area of ​​a single pixel to obtain the linear dark fringe area of ​​each dark fringe pixel sequence.

[0025] The crop planting area in the main SAR image is divided into multiple analysis units according to a rectangular pixel window of a preset size, and the area of ​​each analysis unit is calculated.

[0026] Divide the area of ​​the linear dark lines by the area of ​​the region in the analysis unit to obtain the texture consistency index of each dark line pixel sequence;

[0027] If the texture consistency index of the dark pixel sequence is greater than the set pseudo-dark area threshold, and the average spacing of the dark pixel sequence within the analysis unit matches the pre-acquired crop row spacing, then the region corresponding to the analysis unit is marked as a pseudo-dark area of ​​the crop row structure.

[0028] Furthermore, the method for determining whether the average spacing of the dark texture pixel sequence within the analysis unit matches the pre-acquired crop row spacing includes:

[0029] Calculate the distance between corresponding points on the center lines of two adjacent linear dark stripes within the analysis unit, and calculate the average of all distances to obtain the average spacing of the dark stripes;

[0030] The average spacing of the dark ripples is compared with the pre-acquired crop row spacing. If the difference between the average spacing of the dark ripples and the pre-acquired crop row spacing is within the preset allowable error range, then the average spacing of the dark ripple pixel sequence in the analysis unit matches the pre-acquired crop row spacing.

[0031] Furthermore, methods for marking pixels in the main SAR image as view-sensitive false dark areas or potential water bodies include:

[0032] The SIFT feature matching algorithm is used to register the acquired primary SAR image with the backup SAR image;

[0033] After registration is completed, the difference in backscattering coefficients of corresponding pixels in the primary SAR image and the backup SAR image is calculated to obtain the backscattering difference.

[0034] If the backscattering coefficient of a pixel in the main SAR image is less than the preset low scattering threshold of the main image, and the backscattering difference of the corresponding pixel is greater than the preset high backscattering difference threshold, then the corresponding pixel is marked as a view-sensitive false dark area.

[0035] If the backscattering coefficient of a pixel in the main SAR image is less than the preset low scattering threshold of the main image, and the backscattering difference of the corresponding pixel is less than or equal to the preset low backscattering difference threshold, then the corresponding pixel is marked as a potential water body.

[0036] Furthermore, methods for identifying the water body and the area to be verified include:

[0037] Candidate water bodies are extracted from potential water bodies in the corrected main SAR image using a preset initial threshold method to obtain the initial water body region;

[0038] The initial water body areas in the corrected master SAR image are removed and retained according to the preset removal and retention rules;

[0039] Initial water bodies that were not removed after being excluded and did not meet the retention rules were marked as defined water bodies, while initial water bodies that met the retention rules were marked as areas to be verified.

[0040] Furthermore, the removal rules include a first removal rule and a second removal rule, specifically including:

[0041] The first rejection rule is to remove pixels that are simultaneously marked as candidate terrain false dark areas and view-sensitive false dark areas;

[0042] The second elimination rule is to set a connected region area threshold and a surrounding search radius for pixels that are only marked as pseudo-dark areas of crop row structure; when a pixel that is only marked as pseudo-dark areas of crop row structure satisfies that the connected region area is less than the connected region area threshold, and does not form a connection with an initial water body region whose connected region area is greater than the connected region area threshold within the surrounding search radius, the corresponding pixel is eliminated.

[0043] Furthermore, the retention rules include:

[0044] Pixels that meet the criteria of being labeled as candidate terrain pseudo-dark areas, view-sensitive pseudo-dark areas, or crop row structure pseudo-dark areas, but are located within a pre-acquired known water body preset distance threshold range.

[0045] A water disaster monitoring system, comprising:

[0046] The first marking module is used to mark high-risk terrain shadow areas and medium-risk terrain attenuation areas in the pre-acquired main SAR image according to a preset risk threshold.

[0047] The second labeling module is used to filter out candidate terrain false dark areas from pixels marked as high-risk terrain shadow areas and medium-risk terrain attenuation areas in the main SAR image.

[0048] The third labeling module is used to extract the crop row structure pseudo-dark areas from the main SAR image;

[0049] The fourth marking module is used to mark pixels in the main SAR image as view-sensitive false dark areas or potential water bodies based on pre-acquired backup SAR images.

[0050] The water body screening module is used to perform preset removal and retention operations on pixels marked as candidate terrain false dark areas, crop row structure false dark areas, view-sensitive false dark areas and potential water bodies in the main SAR image, so as to obtain the identified water bodies and areas to be verified.

[0051] The result verification module is used to verify the identified water body and the area to be verified, and to obtain the final water body distribution map.

[0052] Compared with the prior art, the technical effects and advantages of the water disaster monitoring system and method of the present invention are as follows:

[0053] This invention first marks high-risk terrain shadow areas and medium-risk terrain attenuation areas in the pre-acquired main SAR image based on a preset risk threshold. Then, candidate terrain pseudo-dark areas are obtained from the pixels of these marked areas. At the same time, crop row structure pseudo-dark areas are extracted from the main SAR image. Based on the pre-acquired backup SAR image, pixels in the main SAR image are marked as view-sensitive pseudo-dark areas or potential water bodies. Finally, preset elimination and retention operations are performed on the pixels marked as candidate terrain pseudo-dark areas, crop row structure pseudo-dark areas, view-sensitive pseudo-dark areas, and potential water bodies in the main SAR image to obtain the identified water bodies and areas to be verified. The results are further confirmed through optical image comparison, temporal comparison, and on-site sampling verification to generate the final water body distribution map.

[0054] This invention addresses the problems of existing technologies in SAR flood monitoring in mountainous farmland areas. These problems arise because terrain features such as slopes, terraces, and crop row structures weaken radar echoes, creating false dark areas. Low backscattering intensity alone cannot distinguish these false dark areas from real water bodies, leading to numerous misjudgments, reduced reliability of flood distribution maps, and wasted disaster relief resources. Furthermore, these technologies lack quantitative analysis and effective shape correction. The invention's advantages lie in its ability to accurately identify false dark areas caused by terrain and crop row structures. Through cross-validation of multi-view SAR images and multiple rounds of result verification, the accuracy of flood monitoring is improved. By using a data-driven approach to set various thresholds, it adapts to the complex surface environment of mountainous farmland, ensuring the authenticity of identified water bodies and the subsequent traceability of the areas to be verified. Ultimately, this provides reliable water distribution information for flood relief, avoids resource misallocation, and ensures efficient disaster relief efforts. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of a water disaster monitoring system according to an embodiment of the present invention;

[0056] Figure 2 This is a flowchart of a water disaster monitoring method according to an embodiment of the present invention;

[0057] Figure 3 This is a flowchart of a method for marking high-risk terrain shadow areas and medium-risk terrain attenuation areas in a pre-acquired main SAR image according to an embodiment of the present invention;

[0058] Figure 4 This is a flowchart illustrating the method for obtaining pseudo-dark areas in crop row structure according to an embodiment of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be described in detail, clearly, and completely below with reference to the accompanying drawings. It should be particularly noted that the specific embodiments described below are only for better illustrating and explaining the technical solutions of the present invention, and are intended to enable those skilled in the art to better understand and implement the present invention, and should not be construed as limiting the scope of protection of the present invention. Without departing from the spirit and substance of the present invention, those skilled in the art can modify, adjust, or make equivalent substitutions based on the content disclosed in the present invention, and these should all be considered within the scope of protection of the present invention.

[0060] Example 1

[0061] Please see Figure 1 As shown in the figure, this embodiment discloses a water disaster monitoring system, including a first marking module, a second marking module, a third marking module, a fourth marking module and a water body screening module. Each module is connected by wires and / or wirelessly to realize data transmission.

[0062] The first marking module is used to mark high-risk terrain shadow areas and medium-risk terrain attenuation areas in the pre-acquired main SAR image according to a preset risk threshold.

[0063] The primary SAR imagery is acquired using synthetic aperture radar (SAR) satellite data, such as that obtained via the Sentinel-1 satellite. The primary SAR imagery is selected from satellite data that meets the parameter requirements within 24 hours after the flood. In this embodiment, the primary SAR imagery serves as the core data source for flood monitoring. It must be imaged within 24 hours of the flood event, using C-band or X-band, including VV / VH dual polarization, with a resolution of at least 10 meters. Data from periods without rainfall is prioritized to reduce interference from surface changes and rainfall scattering.

[0064] In the process of identifying terrain shadows and slope-sensitive areas, the slope and aspect of each pixel in the main SAR image must first be calculated based on the digital elevation model (DEM). Because mountainous farmland areas have undulating slopes and terraces, these terrain undulations can block radar waves or cause radar echo attenuation, forming false dark areas similar to real water bodies. Slope quantifies the degree of surface tilt, intuitively reflecting the steepness of the terrain undulations, while aspect quantifies the direction of the surface tilt, clarifying the relative positional relationship between the terrain and the radar flight direction. Slope and aspect together constitute the basic parameters for determining whether terrain will interfere with SAR echoes, providing crucial evidence for subsequent identification of terrain-sensitive areas prone to false dark areas.

[0065] The slope and aspect of each pixel in the main SAR image need to be calculated first by obtaining the x-direction elevation gradient of each pixel using the second-order difference method with a 3×3 window. and elevation gradient in the y direction Then substitute the values ​​into the calculation formula for the slope. The method for calculating the slope is as follows:

[0066] ;

[0067] Specifically, the comprehensive gradient of the land surface tilt is obtained by taking the square root of the sum of the squares of the elevation gradients in the x and y directions of each pixel in the main SAR image. Then, through arctangent operation and radian angle conversion, the slope value of each pixel in the main SAR image is obtained.

[0068] The method for calculating slope aspect is as follows:

[0069] ;

[0070] Specifically, by performing arctangent 2 calculations on the elevation gradient in the y-direction and the elevation gradient in the negative x-direction of each pixel in the main SAR image, and combining this with radian angle conversion, the slope aspect value of each pixel in the main SAR image is obtained.

[0071] The methods for calculating slope and aspect are commonly used techniques in the field of geographic information science for extracting terrain parameters, and will not be elaborated on here.

[0072] After calculating the slope and aspect, the relative angle between the slope and the radar line of sight needs to be calculated using SAR imaging geometric parameters. These SAR imaging geometric parameters include the incident angle and radar azimuth angle, both extracted from the SAR image metadata file. Since slope and aspect only describe the tilt characteristics of the terrain itself, and the degree to which SAR echoes are affected by terrain is closely related to the radar's observation angle, terrain with the same slope and aspect will exhibit significantly different blocking or attenuation effects on radar waves under different incident angles or flight directions. Therefore, it is impossible to accurately determine whether false dark areas will form based solely on the terrain's own parameters. The relative angle between the slope and the radar line of sight quantifies the coupling relationship between them, intuitively reflecting the actual angle at which radar waves illuminate the slope, and thus accurately determining whether false dark areas are prone to occur under these terrain conditions. The formula for calculating the relative angle α between the slope and the radar line of sight is:

[0073] ;

[0074] In the formula, θ is the angle of incidence; This is the radar azimuth angle. The relative angle between the slope and the radar line of sight is derived based on trigonometric functions. By performing cosine calculation on the difference between the slope aspect and the radar azimuth angle, and combining the relationship between the slope and the incident angle, a correlation is established between terrain parameters and radar imaging geometric parameters. The magnitude of the relative angle between the slope and the radar line of sight directly corresponds to the actual angle between them. The smaller the relative angle, the closer the radar wave is to illuminating along the tangent of the slope. In this case, the uphill area is more likely to block the downhill area, or the reflection path of the radar wave on the slope increases, resulting in a significant reduction in radar echo intensity, and the blocking or attenuation effect is more prominent.

[0075] Finally, please see Figure 3 As shown, based on preset risk thresholds, high-risk terrain shadow areas and medium-risk terrain attenuation areas are marked in the main SAR image. The risk thresholds include an upper limit for slope, a lower limit for relative angle, and a middle limit for slope. Pixels in the main SAR image that satisfy a slope greater than or equal to the upper limit of the slope threshold, and a relative angle between the slope and the radar line of sight less than or equal to the lower limit of the relative angle threshold, are marked as high-risk terrain shadow areas. High-risk terrain shadow areas correspond to radar waves illuminating almost along the tangent direction of the slope, with a significant blocking effect of the upslope on the downslope, resulting in a significant weakening of the radar echo and easily forming strong false dark areas similar to real water bodies. Pixels in the main SAR image that satisfy a slope between the middle and upper limits of the slope threshold, and a relative angle between the slope and the radar line of sight less than or equal to the lower limit of the relative angle threshold, are marked as medium-risk terrain attenuation areas. Medium-risk terrain attenuation areas correspond to radar waves not illuminating completely along the tangent direction of the slope, but the slope inclination leads to an increased reflection path, resulting in a significant attenuation effect of the terrain on the radar waves and easily forming weak false dark areas similar to real water bodies.

[0076] The method for setting the risk threshold includes: constructing a reference sample set, which is derived from historical waterless period images from the same sensor and the same orbital direction; selecting stable non-water-accumulating pixels; calculating the slope and the relative angle between the slope and the radar line of sight for the corresponding time period to form a joint distribution baseline; the joint distribution baseline can clearly present the distribution characteristics of stable non-water-accumulating pixels in the two-dimensional space of the relative angle between the slope and the radar line of sight, providing a benchmark distribution range for subsequent differentiation between negative and positive classes, and clarifying the parameter boundaries of normal non-water-accumulating areas. Using the reference sample set as the first negative class sample and the confirmed shadow and decay samples in the target area as the first positive class sample, the upper bound, middle bound, and lower bound of the slope threshold are automatically obtained by minimizing the classification error rate or maximizing the Youden index. The risk threshold is mapped to the two-dimensional space of the joint distribution baseline. It is checked whether the first negative class samples are evenly distributed within the normal parameter range defined by the threshold, and at the same time, it is confirmed whether the first positive class samples are evenly distributed within the risk parameter range defined by the risk threshold. If the samples can be clearly separated by the threshold on the joint distribution baseline, and there are no large numbers of first negative class samples falling into the risk range and no large numbers of first positive samples falling into the normal range, then the risk threshold is determined to be reasonable and can be used for subsequent labeling of high-risk terrain shadow areas and medium-risk terrain decay areas. If there is a case of sample cross-distribution, the selection criteria for positive and negative class samples need to be readjusted or the criterion function needs to be optimized, such as by adding a regularization term. The risk threshold is then recalculated and verified in conjunction with the joint distribution baseline until the risk threshold can effectively distinguish between positive and negative class samples on the joint distribution baseline.

[0077] For example, in a hilly terraced area in southern China, when the slope is greater than 25° and the relative angle is less than 30°, the slope's shielding effect on radar waves reaches a significant level, which matches the topographic characteristics of a high-risk terrain shadow area. When the slope is between 15° and 25° and the relative angle is less than 20°, the slope's attenuation effect on radar waves reaches a significant level, which matches the topographic characteristics of a medium-risk terrain attenuation area. By using the numerical ranges corresponding to the above characteristics, high-risk terrain shadow areas and medium-risk terrain attenuation areas can be accurately marked.

[0078] The second labeling module is used to filter out candidate terrain false dark areas from pixels marked as high-risk terrain shadow areas and medium-risk terrain attenuation areas in the main SAR image.

[0079] A topographic radiometric correction model based on the DEM (Digital Image Model) is adopted, such as the RVOG model. This topographic radiometric correction model is a commonly used technique in the field of geographic information and remote sensing to eliminate the influence of terrain undulations on the backscattering of the main SAR image. Its core principle is to establish a correlation between terrain elevation and radar wave propagation path, correcting the radar wave incident angle deviation caused by slope inclination. It can meet the correction needs of mountainous farmland areas without modifying the model structure or calculation logic, effectively reducing the spurious backscattering changes caused by terrain undulations, and providing an accurate backscattering data foundation for subsequent false dark area identification.

[0080] The influence of terrain undulation on backscattering in the main SAR image is eliminated by using a terrain radiometric correction model, resulting in the corrected main SAR image and the corrected backscattering coefficients for each pixel. The correction process follows the formula:

[0081] ;

[0082] In the formula, The corrected backscattering coefficient is used to represent the quantified value of the ability of each pixel in the main SAR image to scatter radar waves after topographic radiometric correction. θ' is the raw backscattering coefficient, used to represent the quantified value of the scattering ability of each pixel in the main SAR image to radar waves, without topographic radiometric correction; θ' is the slope incidence angle. The backscattering coefficient is a core physical parameter in remote sensing used to describe the radar wave reflection characteristics of ground objects. Essentially, it is the ratio of the intensity of the echo signal received by the radar sensor to the intensity of the radar wave incident on the ground object's surface, usually expressed in decibels (dB). The magnitude of the backscattering coefficient directly reflects the ground object's scattering ability to radar waves: a higher value indicates stronger scattering, such as rough surfaces like bare land and vegetation canopies, which typically have high backscattering coefficients; a lower value indicates weaker scattering, such as calm water bodies where, due to specular reflection, most radar waves are reflected away from the radar sensor's receiving direction, producing only very weak echoes, thus resulting in a very low backscattering coefficient. In SAR flood monitoring, the backscattering coefficient is a key indicator for distinguishing between water bodies and non-water bodies. However, factors such as topographic relief and crop structure can interfere with the true value of the backscattering coefficient. Correction and filtering are necessary to eliminate interference and ensure that it accurately reflects the actual scattering characteristics of ground features. The slope incidence angle is the angle between the radar wave incident direction and the slope normal. It is obtained by using slope gradient to characterize the slope inclination, and the difference between slope aspect and radar azimuth angle to reflect the relative relationship between the slope orientation and the radar flight direction. Substituting the slope gradient (characterizing the slope inclination) and the difference between slope aspect and radar azimuth angle into the angle synthesis formula yields the actual incidence angle of the radar wave relative to the slope normal, i.e., the slope incidence angle.

[0083] A secondary screening process is performed on the corrected master SAR image, targeting the marked high-risk terrain shadow areas and medium-risk terrain attenuation areas. The screening process first requires calculating the average backscattering coefficient of all pixels within the high-risk terrain shadow areas and medium-risk terrain attenuation areas after correction, denoted as the backscattering mean. Despite terrain radiometric correction, some high-risk terrain shadow areas and medium-risk terrain attenuation areas may still exhibit residual low-scattering characteristics due to terrain structures such as multiple reflections from steep slopes and partial obstruction by terraced fields. These characteristics cannot be completely eliminated by a single correction; therefore, it is necessary to calculate the backscattering mean through mean statistics to determine whether the overall backscattering level of the corresponding area remains within the low-scattering range that could easily be misidentified as water.

[0084] The criteria for determining whether the overall backscattering level of a corresponding area is still within the low-scattering range that is easily misidentified as water include: if the average backscattering value of the corresponding area is lower than the set normal backscattering threshold, then the corresponding area is included in the candidate topographic pseudo-dark area. The normal backscattering threshold is set based on SAR images of the study area acquired by the same sensor during the same time period before the flood (during the period without water bodies). These SAR images reflect the normal surface scattering state of the study area when there is no flood interference. Stable, non-waterlogged areas such as flat bare land and uncultivated farmland without topographic interference or crop structure interference are selected from the SAR images of the pre-flood water-free period. The backscattering coefficients of these areas are extracted, and abnormal extreme values ​​caused by sensor noise and atmospheric interference are removed. The average backscattering level of the normal surface in the corresponding area is obtained through statistical calculation. This average backscattering level is the normal backscattering threshold.

[0085] Areas with backscattering below the normal backscattering threshold need to be included in the candidate topographic pseudo-dark areas because the backscattering coefficient of real flood water bodies is lower than that of normal surfaces, which is determined by the specular reflection characteristics of water bodies. However, if high-risk topographic shadow areas and medium-risk topographic attenuation areas are still below the normal backscattering threshold after correction, it indicates that the backscattering characteristics of the corresponding areas are highly similar to those of real flood water bodies. If the corresponding areas are not included in the candidate topographic pseudo-dark areas, they are very likely to be misclassified as real flood water bodies during the subsequent water body extraction process, leading to deviations in flood monitoring results. Therefore, it is necessary to screen out candidate topographic pseudo-dark areas to provide a basis for further distinguishing pseudo-dark areas from real water bodies.

[0086] The third labeling module is used to extract the crop row structure pseudo-dark areas from the main SAR image.

[0087] Please see Figure 4 As shown, methods for extracting false dark areas of crop row structure from main SAR images include:

[0088] Directional Gabber filtering is applied to the corrected master SAR image. In the corrected synthetic aperture radar image, the pseudo-dark areas formed by crop rows exhibit a linear distribution consistent with the crop row direction. Meanwhile, random speckle noise, isolated low-scattering points, and other noise or non-target features have chaotic directions. Directional Gabber filtering can strongly respond to features in the corrected master SAR image that align with the set crop row direction, while suppressing interference signals from other directions. This highlights features related to the crop row direction, laying the foundation for accurate extraction of dark fringes pixel sequences. When setting the directional Gabber filtering parameters, the filtering direction must be consistent with the crop row direction data, and the frequency must be dynamically adjusted according to the crop row spacing and SAR image resolution. The adjustment principle is to match the filtering frequency with the spatial distribution frequency of the crop rows in the image; that is, the spatial period corresponding to the filtering frequency should be similar to the pixel spacing of the crop row spacing in the image. This ensures that the filtering can accurately capture the linear structure formed by the crop rows, avoiding the omission of sparse crop row features due to excessively high frequencies, or the incorrect extraction of wide low-scattering regions outside the crop rows due to excessively low frequencies. Directional Gabor filtering is used to extract dark pixel sequences parallel to the crop row direction. The specific extraction process is as follows: Directional Gabor filtering enhances the grayscale changes of pixels in the same direction as the filtering direction and suppresses grayscale changes in other directions. Then, a grayscale threshold is set, which is determined based on the average grayscale value of the normal non-crop row area in the filtered image. It is usually taken as 60% to 70% of the average grayscale value of the normal non-crop row area. Pixels with grayscale values ​​below the grayscale threshold after filtering are initially screened out. Then, connectivity analysis is performed on the initially screened low grayscale pixels. Taking the dark pixel as the center, the continuous low grayscale pixels within the surrounding 3×3 or 5×5 pixel window are grouped into the same region. The surrounding region is defined as a square range with a side length of 3 or 5 pixels centered on a single low grayscale pixel. This ensures that only continuously distributed low grayscale pixel sequences are extracted and isolated low grayscale noise points are removed. Finally, a dark pixel sequence parallel to the crop row direction is obtained. The dark pixel sequence is a continuous low-grayscale line-shaped area parallel to the direction of crop planting rows in the corrected main SAR image, formed by the directional scattering and blocking of radar waves by the crop row structure, resulting in weakened radar echoes. The grayscale value of this area is lower than that of the surrounding non-crop row areas, and it has clear directional consistency and spatial continuity.

[0089] The core characteristic of crop row structure pseudo-dark areas is the high directional consistency and spatial continuity of linear dark streaks within the region. In contrast, low-grayscale areas of real water bodies lack obvious linear textures, and the linear features of other pseudo-dark areas, such as terrain shadows, are chaotic. Therefore, a texture consistency index is calculated to quantify the distribution density and regularity of linear dark streaks, distinguishing crop row structure pseudo-dark areas from other low-scattering regions. Furthermore, extracting linear dark streaks alone cannot determine whether the linear features within a region are dominated by the crop row structure; a small number of random linear interferences may be incorrectly extracted. Further verification using the texture consistency index is needed to ensure that only regions dominated by the crop row structure are included as pseudo-dark area candidates. The formula for calculating the texture consistency index (CI) is:

[0090] CI = ASIZE / SIZE × 100%;

[0091] In the formula, ASIZE is the linear dark fringe area of ​​each dark fringe pixel sequence, used to represent the actual area occupied by the extracted dark fringe pixel sequence. It is obtained by: performing pixel-level statistics on each dark fringe pixel sequence, recording the number of pixels contained in each dark fringe pixel sequence, and calculating the actual area corresponding to a single pixel of each dark fringe pixel sequence in combination with the spatial resolution of the SAR image, which is the square of the spatial resolution. Then, multiply the number of dark fringe pixels in each dark fringe pixel sequence by the actual area of ​​a single pixel to obtain the linear dark fringe area of ​​each dark fringe pixel sequence. SIZE is the region area, used to represent the actual area of ​​the analysis unit selected when calculating the texture consistency index. The analysis unit needs to cover a certain range of crop planting area, usually set as a rectangular pixel window of a fixed size, such as 50×50 pixels. The number of window pixels is converted into the actual area according to the spatial resolution of the SAR image, which is the region area.

[0092] When determining whether a region is marked as a false dark area in the crop row structure, the texture consistency index (CI) must be greater than a set false dark area threshold. The false dark area threshold is set as follows: Select normal surface areas within the study area, such as flat bare land without crop rows or cultivated land during the non-planting season. Calculate the texture consistency index (CI) of the normal surface area and obtain the 95th quantile of the CI. Use this 95th quantile as the false dark area threshold. If the CI of the analysis unit is greater than the false dark area threshold, it indicates that the distribution density of linear dark lines in the corresponding area is higher than that of the normal surface, which matches the texture characteristics of a false dark area in the crop row structure. Furthermore, the average spacing of the dark lines must match the known crop row spacing. The matching method is as follows: First, calculate the distance between corresponding points on the center lines of two adjacent linear dark lines, and then calculate the average of all adjacent distances to obtain the average spacing of the dark lines. Then, compare the average spacing of the dark lines with the pre-obtained crop row spacing. If the difference between the average spacing of the dark lines and the known crop row spacing is within a preset allowable error range, it is considered a match. The error range is determined based on the spatial resolution of the SAR image and the regularity of crop planting, and can accommodate normal fluctuations in planting spacing. The analysis unit regions that meet the above two conditions are marked as pseudo-dark areas of crop row structure.

[0093] For areas where crop planting row orientation information has not been obtained in advance, the Fourier spectrum peak direction of the corrected synthetic aperture radar image is calculated using a sliding window. The size of the sliding window needs to be quantitatively adjusted based on the crop row spacing and the spatial resolution of the main SAR image. The specific calculation method is as follows: the side length of the sliding window is equal to the crop row spacing multiplied by a coefficient of 2 to 3, and then divided by the spatial resolution of the main SAR image. The window size calculated by this formula can ensure that the window contains at least 2 to 3 crop row periods. Including 2 to 3 crop row periods can fully present the linear dark stripe features formed by the crop rows within the window, providing sufficient samples for the subsequent Fourier transform to accurately capture the gray-scale change pattern, avoiding incomplete row orientation features due to an excessively small window, or excessive interference from non-crop rows due to an excessively large window. For example, if the spatial resolution of the main SAR image is 10 meters and the row spacing of crops in a certain area is 50 centimeters, the sliding window side length calculated using the formula is (50 cm × 2) ÷ 10 meters = 0.1 meters ÷ 10 meters = 0.01 pixels. This result is not accurate. The units need to be standardized before recalculation. Converting the 50 cm row spacing to 0.5 meters and recalculating yields (0.5 meters × 2) ÷ 10 meters = 0.1 pixels. This is still too small and the coefficient needs to be adjusted to 20 to 30, i.e., (0.5... (0.5m × 20) ÷ 10m = 1 pixel, (0.5m × 30) ÷ 10m = 1.5 pixels, rounded to 2×2 pixels; if the crop row spacing is 1 meter and the spatial resolution is 10 meters, then (1m × 2) ÷ 10m = 0.2 pixels, rounded to 1×1 pixel, (1m × 3) ÷ 10m = 0.3 pixels, rounded to 1×1 pixel. In practical applications, the coefficient can be adjusted appropriately according to the ratio of crop row spacing to resolution to ensure that the window can cover the effective row features.

[0094] A two-dimensional Fourier transform is performed on the pixel grayscale values ​​of the corrected main SAR image within the sliding window, converting the grayscale distribution in the spatial domain into a spectral distribution in the frequency domain. The direction corresponding to the highest energy peak in the Fourier spectrum is the direction of the greatest grayscale change within the window. Linear dark fringes formed by crop rows cause frequent grayscale changes along the direction perpendicular to the dark fringes in the spatial domain, which manifests as spectral peaks in the corresponding direction in the frequency domain. By locating the position of the highest energy peak in the spectrum, the angle corresponding to the highest energy peak position is read as the dominant linear texture direction, i.e., the potential crop row direction. After identifying the dominant linear texture direction, the texture consistency index (CI) calculation and determination process is repeated to allow areas without crop row direction data to complete linear dark fringe extraction, CI calculation, and matching determination of the average dark fringe spacing with the crop row spacing through the automatically identified dominant linear texture direction. This avoids the omission of false dark areas in the crop row structure in areas without crop row direction data. Finally, by repeating the process, the pseudo-dark areas of crop row structure in the areas without crop row direction data are obtained and merged with the pseudo-dark areas of crop row structure in the areas with crop row direction data to form a pseudo-dark area of ​​crop row structure covering the entire study area.

[0095] The fourth labeling module is used to label pixels in the main SAR image as view-sensitive false dark areas or potential water bodies based on pre-acquired backup SAR images.

[0096] Backup SAR imagery is acquired using SAR satellite data. This backup SAR imagery is selected from satellite data with different vertical orbits or polarizations within the same contemporaneous period, based on the orbit and polarization parameters of the primary SAR imagery. As a data source for comparison and verification, the backup SAR imagery must be contemporaneous with the primary SAR imagery, for example, with a time difference of less than or equal to 3 days, and exhibiting either an ascending or descending orbit intersection or different polarization patterns. Its resolution must be the same as or higher than that of the primary SAR imagery. Backup SAR imagery provides scattering data from different viewpoints or polarizations, used to distinguish viewpoint-sensitive false dark areas from real water bodies.

[0097] It should be noted that the primary SAR image and the backup SAR image used in all steps of the fourth labeling module have been corrected by the topographic radiometric correction model, and the backup SAR image uses the same correction model and the same parameters as the primary SAR image.

[0098] Due to differences in imaging trajectory and time, the pixel positions of the same ground feature do not coincide between the primary and backup SAR images. Directly calculating the backscatter difference between corresponding pixels would lead to distorted results because the pixels do not represent the same ground feature, making it impossible to accurately distinguish between false dark areas and real water bodies. Therefore, registration between the primary and backup SAR images is necessary. Registration unifies the spatial coordinates of the primary and backup SAR images, ensuring that each pair of pixels involved in the calculation corresponds to the same ground feature, providing an accurate spatial matching basis for backscatter difference calculation. The registration method uses the SIFT feature matching algorithm. First, stable feature points such as edges and corners are extracted from the two images. The spatial correspondence between the two images is established through the grayscale and shape similarity of these feature points. Then, affine transformation is used to spatially correct the backup SAR image, adjusting its pixel positions to match the spatial coordinates of the primary SAR image.

[0099] After registration, the difference in backscattering coefficients of corresponding pixels in the primary SAR image and the backup SAR image is calculated to obtain the backscattering difference. Due to the view-dependent nature of false dark areas caused by terrain shadows, crop row structures, etc., the backscattering intensity changes depending on the angle of view of the radar wave illumination and reflection path when the radar orbits of the primary and backup SAR images are perpendicular. In contrast, due to the specular reflection characteristics of real water bodies, most radar waves are reflected out of the sensor's receiving direction regardless of the radar viewpoint, resulting in consistently low backscattering intensity. The difference in backscattering intensity between pixels of the same water body in the two images is small. This difference can be quantified using the backscattering difference, allowing for a preliminary distinction between false dark areas and real water bodies based on their fundamental difference in scattering.

[0100] Pre-set scattering thresholds, including the low scattering threshold for the main image. High threshold for backscattering difference Low threshold for backscattering difference The method for setting the scattering threshold is as follows: Select SAR images of the study area during the 30-day period before the flood, ensuring they are in the same orbit and polarization as the main SAR image. Extract the backscattering coefficients of stable, non-waterlogged areas such as flat bare land and uncultivated farmland. Calculate the mean value of the backscattering coefficients of stable, non-waterlogged areas. with standard deviation ,Will Set as -2× ,make sure It can accurately define low-scattering areas and exclude accidental low-scattering pixels caused by noise on normal surfaces. A reference sample set is constructed, with normal surface pixels in waterless images as the second negative class, pixels with known topographic shadows or crop row structure pseudo-dark areas as positive class 1, and pixels with known real water bodies as positive class 2. Positive class 1 and positive class 2 samples constitute the second positive class. Substituting the reference sample set into the classification model, the high and low thresholds for backscattering difference are automatically calculated using the criteria of minimizing the classification error rate or maximizing the Youden index. The Youden index is an indicator that measures the effectiveness of a classification model in distinguishing between different classes of samples. It is calculated by subtracting 1 from the sum of sensitivity and specificity. Sensitivity refers to the proportion of samples correctly identified by the classification model in the second positive class, and specificity refers to the proportion of samples correctly identified by the classification model in the second negative class. The Youden index ranges from 0 to 1. A higher value indicates that the classification model is better at distinguishing between different classes of samples. By maximizing the Youden index, the thresholds that allow the second positive class sample to be optimally distinguished from the second negative class sample, and from the positive class 1 sample to the positive class 2 sample can be found. The high threshold of backscattering difference is the threshold for distinguishing between the positive class 1 sample and the second negative class sample, and from the positive class 2 sample, while the low threshold of backscattering difference is the threshold for distinguishing between the positive class 2 sample and the second negative class sample, and from the positive class 1 sample.

[0101] If the backscattering coefficient of a pixel in the main SAR image is less than And the backscattering difference of the corresponding pixel is greater than In this case, all such pixels are marked as view-sensitive false dark areas because the backscattering coefficient of the corresponding pixel in the main image is less than [value missing]. This only indicates that its scattering intensity is low, which could be a pseudo-dark area or a real water body; while the backscattering difference of the corresponding pixel is greater than This indicates that the backscattering coefficient of the corresponding pixel in the backup SAR image is higher than that in the main image, meaning that the scattering intensity of the same ground object varies greatly under different viewing angles. This is consistent with the core characteristics of view-sensitive false dark areas, namely, the low scattering of the false dark area is caused by terrain occlusion or crop row directional scattering under specific viewing angles. For example, when the main image is in a descending orbit and the backup image is in an ascending orbit, the occlusion or directional scattering effect disappears and the backscattering intensity rebounds. Therefore, it is determined to be a view-sensitive false dark area.

[0102] If the backscattering coefficient of a pixel in the main SAR image is less than the preset low scattering threshold for the main image, and the backscattering difference of the corresponding pixel is less than or equal to... If a pixel exhibits low scattering in both the primary and backup SAR images, it is labeled as a potential water body. Since the corresponding pixel in the primary SAR image shows low scattering, and the backscattering difference is less than or equal to T3, it indicates that the backscattering coefficient of the corresponding pixel in the backup SAR image is similar to that in the primary image. This means that the same ground feature maintains low scattering intensity under different viewing angles, consistent with the core characteristics of a real water body. The low scattering of a real water body is determined by specular reflection and is independent of the radar viewing angle. Regardless of the viewing angle, specular reflection results in extremely weak radar echoes, and the backscattering intensity remains consistently low and stable; therefore, it is identified as a potential water body.

[0103] The water body screening module is used to perform preset removal and retention operations on pixels marked as candidate terrain false dark areas, crop row structure false dark areas, view-sensitive false dark areas and potential water bodies in the main SAR image, so as to obtain the identified water bodies and areas to be verified.

[0104] Candidate water bodies are extracted from the corrected main SAR image using an initial threshold method: an adaptive threshold is set, which is calculated using the Otsu algorithm. The Otsu algorithm performs statistical analysis on the corrected backscattering coefficients of all pixels in the corrected main SAR image and automatically finds the gray value that maximizes the variance between potential water bodies and non-water bodies, which is the adaptive threshold. The adaptive threshold can effectively distinguish between low-scattering and high-scattering regions in the corrected main SAR image. Regions with corrected backscattering coefficients less than the adaptive threshold are marked as initial water body regions.

[0105] The marked regions in the corrected master SAR image are removed and retained according to preset removal and retention rules. The preset removal and retention rules are as follows:

[0106] The first elimination rule is to remove areas that are simultaneously marked as candidate terrain pseudo-dark areas and view-sensitive pseudo-dark areas. This is because candidate terrain pseudo-dark areas are terrain interference areas that still have residual low scattering characteristics after terrain radiometric correction, while view-sensitive pseudo-dark areas are view-dependent pseudo-dark areas confirmed by multi-view cross-validation. Areas that are simultaneously marked as candidate terrain pseudo-dark areas and view-sensitive pseudo-dark areas have pseudo-dark area characteristics caused by both terrain and view, making them highly credible as pseudo-dark areas and likely to not contain real water bodies. Therefore, they are directly removed to avoid misjudgment.

[0107] The second rejection rule: For areas marked only as false dark areas of crop row structures, a connected domain area threshold and a surrounding search radius are set. The connected domain area threshold is determined by statistically analyzing the minimum water body area in historical real flood events in the study area, ensuring that the connected domain area threshold can exclude small-area low-scattering areas that are not water bodies. The surrounding search radius is set according to the spatial resolution of the corrected main SAR image, usually 3 to 5 pixels, to cover the possible continuous water body extension range around the area. When a pixel marked only as a false dark area of ​​crop row structures satisfies the condition that the connected domain area is less than the connected domain area threshold, and does not form a connection with an initial water body area greater than the connected domain area threshold within the surrounding search radius, it is determined that there is no continuous water body extension around it, and the corresponding pixel is rejected. These areas are rejected because the low-scattering characteristics of areas marked only as false dark areas of crop row structures may be caused by crop row structures, and the characteristics of small area and no continuous water body extension indicate that they lack the spatial continuity of real water bodies, and are more likely to be false dark areas of crop row structures rather than real flood water bodies. Therefore, they are rejected to reduce the false alarm rate.

[0108] Retention Rules: A distance threshold is set, determined by statistically analyzing the flood spread range of known rivers, reservoirs, and other water bodies in the study area during historical flood events. This ensures the distance threshold covers potential flood extension areas around known water bodies. Pixels marked as candidate terrain pseudo-dark areas, viewpoint-sensitive pseudo-dark areas, or crop row structure pseudo-dark areas, but located within the pre-acquired distance threshold range of known water bodies, are temporarily retained and marked as areas to be verified. The reason for retaining these areas is that known water bodies easily spread to surrounding areas during floods, forming small areas of water accumulation. These areas may be marked due to proximity to terrain interference areas or crop planting areas, but are actually real flood water bodies. Directly removing them would lead to missed detection of real flood areas. Therefore, they are temporarily retained and marked as areas to be verified, and further confirmation of whether these areas are real flood water bodies will be made through on-site sampling or optical image verification.

[0109] The initial water bodies that are not removed after being removed and do not meet the verification conditions in the retention rules are marked as determined water bodies. The final output is the determined water bodies and the areas to be verified.

[0110] The result verification module is used to verify the identified water body and the area to be verified, and to obtain the final water body distribution map.

[0111] The authenticity of water bodies and areas to be verified is conducted using cloud-free optical imagery taken during the same period as the flood. The selected optical imagery must meet the requirements of a short time interval between the flood occurrence and the actual flood, and the absence of cloud cover to ensure that the imagery accurately reflects the surface conditions at the time of the flood. Commonly used optical imagery includes Landsat-8 and Sentinel-2 images. The verification process uses the water body index as the core indicator. The calculation of the water body index is based on the green band reflectance and near-infrared band reflectance values ​​of the optical imagery. It leverages the characteristic that water bodies have strong green band reflectance and weak near-infrared band reflectance, while soil and vegetation have strong near-infrared band reflectance. Normalization calculations highlight water body characteristics. The core logic is to quantify the surface water body attributes through differences in band reflectance.

[0112] When the water index of a certain area is greater than the set water index judgment threshold, it indicates that the surface of the area is mainly composed of water, and it is confirmed as a real water body. If the water index is less than or equal to the water index judgment threshold, it is not possible to accurately distinguish based on the water index alone. It is necessary to further verify this by combining the polarization ratio of SAR imagery. This is because the polarization ratio characteristics of real water bodies differ from those of high-humidity soils. The polarization ratio can help determine whether the area is a real water body. For example, in a plain farmland area, if the water index of a certain area is greater than 0.3, it is directly confirmed as a real water body. If the water index is between 0.1 and 0.3, and the polarization ratio of the SAR imagery is greater than 0.8, it is also confirmed as a real water body. Otherwise, it is determined to be high-humidity soil.

[0113] Historical SAR images prior to the flood were selected as the benchmark for comparison. To ensure the effectiveness of the comparison, the historical SAR images needed to maintain the same orbital direction and polarization as the main SAR image, avoiding deviations in scattering characteristics caused by differences in imaging geometric parameters, and ensuring the comparability of water scattering information between the two images. First, the historical water area of ​​the study area was extracted from the historical SAR images, i.e., the area of ​​perennial rivers and reservoirs that existed stably in the area before the flood. Then, the current water area of ​​the corresponding area was extracted from the identified water bodies and the area to be verified. The newly added water area was obtained by calculating the difference between the current water area and the historical water area. The newly added water area is the current water area minus the historical water area. If the difference is negative, it indicates that there is no newly added water in the area.

[0114] The water change rate is calculated as the ratio of the newly added water area to the historical water area. This quantifies the degree of change in the water area and the area to be verified relative to its historical state. When the water change rate exceeds a set threshold, it indicates that the water area in the corresponding region has increased compared to before the flood, consistent with the characteristics of water expansion caused by flooding, and is therefore prioritized as a flood-inundated area. If the water change rate is less than or equal to the threshold, it indicates that the water area in the corresponding region has changed relatively little, possibly due to natural fluctuations in historical water levels or misjudgment based on high-humidity soil. Further analysis based on the regional land use type is required. For example, if the land use type in the region is cultivated land, high-humidity soil is likely to appear after irrigation, requiring on-site sampling to verify whether it is a genuine flood water body, thus obtaining the final water distribution map. For instance, if the historical water area of ​​a certain region is 2 square kilometers, and the current water area in the region and the area to be verified is determined to be 3.5 square kilometers, with a newly added water area of ​​1.5 square kilometers, the water change rate is 75%, exceeding the set threshold, and is therefore prioritized as a flood-inundated area.

[0115] Example 2

[0116] Please see Figure 2 As shown in the figure, this embodiment provides a method for monitoring water-related disasters, including:

[0117] Based on the preset risk threshold, high-risk terrain shadow areas and medium-risk terrain attenuation areas are marked in the pre-acquired main SAR images;

[0118] Candidate terrain false dark areas are obtained from the pixels marked as high-risk terrain shadow areas and medium-risk terrain attenuation areas in the main SAR image;

[0119] The pseudo-dark region of crop row structure was extracted from the main SAR image;

[0120] Based on the pre-acquired backup SAR images, the pixels in the main SAR images are marked as view-sensitive false dark areas or potential water bodies;

[0121] After performing preset removal and retention operations on the pixels marked as candidate terrain false dark areas, crop row structure false dark areas, view-sensitive false dark areas and potential water bodies in the main SAR image, the identified water bodies and areas to be verified are obtained.

[0122] The identified water bodies and areas to be verified were then validated to obtain a water body distribution map.

[0123] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

[0124] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A water regime disaster monitoring method, characterized by, The method comprises the following steps: According to the preset risk threshold, mark the high-risk terrain shadow area and the medium-risk terrain attenuation area in the pre-acquired main SAR image; From the pixels marked as high-risk terrain shadow area and medium-risk terrain attenuation area in the main SAR image, screen to obtain candidate terrain pseudo-dark area; Extract the crop row structure pseudo-dark area from the main SAR image; According to the pre-acquired backup SAR image, mark the pixels in the main SAR image as view angle sensitive pseudo-dark area or potential water body; After the pixels marked as candidate terrain pseudo-dark area, crop row structure pseudo-dark area, view angle sensitive pseudo-dark area and potential water body in the main SAR image are subjected to preset rejection and retention operation, the determined water body and the to-be-verified area are obtained; Verify the determined water body and the to-be-verified area to obtain the water body distribution map.

2. The water disaster monitoring method of claim 1, wherein, The method for marking the high-risk terrain shadow area and the medium-risk terrain attenuation area in the pre-acquired main SAR image comprises the following steps: Based on the pre-calculated slope and slope direction of each pixel of the main SAR image, the relative angle between the slope surface and the radar line of sight is calculated combined with the SAR imaging geometric parameters; the SAR imaging geometric parameters include the incidence angle and the radar azimuth angle; The risk threshold includes the upper slope threshold, the lower relative angle threshold and the middle slope threshold; The pixels in the main SAR image that satisfy the conditions of slope greater than or equal to the upper slope threshold and the relative angle between the slope surface and the radar line of sight less than or equal to the lower relative angle threshold are marked as high-risk terrain shadow area; The pixels in the main SAR image that satisfy the conditions of slope between the middle slope threshold and the upper slope threshold and the relative angle between the slope surface and the radar line of sight less than or equal to the lower relative angle threshold are marked as medium-risk terrain attenuation area.

3. The water disaster monitoring method of claim 1, wherein, The method for obtaining the candidate terrain pseudo-dark area comprises the following steps: Correct the main SAR image through the preset terrain radiation correction model to obtain the corrected main SAR image and the corrected backscattering coefficient corresponding to each pixel; Calculate the average value of the corrected backscattering coefficient of all pixels in the high-risk terrain shadow area and the medium-risk terrain attenuation area in the corrected main SAR image, and mark it as backscattering average value; If the backscattering average value of the corresponding area is lower than the set normal backscattering threshold, the corresponding area is included in the candidate terrain pseudo-dark area.

4. The water disaster monitoring method of claim 1, wherein, The method for obtaining the crop row structure pseudo-dark area comprises the following steps: Extract a plurality of dark line pixel sequences parallel to the crop row direction in the main SAR image through directional Gaber filtering; Record the pixel number of each dark line pixel sequence, calculate the actual area corresponding to a single pixel combined with the spatial resolution of the main SAR image, and then multiply the pixel number of each dark line pixel sequence by the actual area of a single pixel to obtain the linear dark line area of each dark line pixel sequence; Divide the crop planting area in the main SAR image into a plurality of analysis units according to a preset size of a rectangular pixel window, and calculate the area of each analysis unit; Divide the linear dark line area by the area of the analysis unit where it is located to obtain the texture consistency index of each dark line pixel sequence; If the texture consistency index of the dark line pixel sequence is greater than the preset false dark area threshold, and the average interval of the dark line pixel sequence in the analysis unit matches the pre-acquired crop row spacing, the region corresponding to the analysis unit is marked as a crop row structure false dark area.

5. The water disaster monitoring method of claim 4, wherein, The method for determining whether the average interval of the dark line pixel sequence in the analysis unit matches the pre-acquired crop row spacing comprises: calculating the distance between corresponding points on the center lines of adjacent two linear dark lines in the analysis unit, and calculating the average value of all distances to obtain a dark line average interval; comparing the dark line average interval with the pre-acquired crop row spacing, if the difference between the dark line average interval and the pre-acquired crop row spacing is within a preset allowable error range, then the average interval of the dark line pixel sequence in the analysis unit matches the pre-acquired crop row spacing.

6. The water disaster monitoring method of claim 1, wherein, The method for marking pixels in the main SAR image as view angle sensitive false dark areas or potential water bodies comprises: using a SIFT feature matching algorithm to register the collected main SAR image and the backup SAR image; after registration, calculating the difference between the backscattering coefficients of corresponding pixels in the main SAR image and the backup SAR image to obtain a backscattering difference value; if the backscattering coefficient of a pixel in the main SAR image is less than a preset main image low scattering threshold, and the backscattering difference value of the corresponding pixel is greater than a preset backscattering difference value high threshold, then the corresponding pixel is marked as a view angle sensitive false dark area; if the backscattering coefficient of a pixel in the main SAR image is less than a preset main image low scattering threshold, and the backscattering difference value of the corresponding pixel is less than or equal to a preset backscattering difference value low threshold, then the corresponding pixel is marked as a potential water body.

7. The water disaster monitoring method of claim 3, wherein, The method for obtaining a determined water body and a to-be-verified region comprises: extracting a candidate water body from the potential water body in the corrected main SAR image using a preset initial threshold method to obtain an initial water body region; performing elimination and reservation on the initial water body region in the corrected main SAR image according to preset elimination rules and reservation rules; marking the initial water body region that is not removed after elimination and does not satisfy the reservation rules as a determined water body, and marking the initial water body region that satisfies the reservation rules as a to-be-verified region.

8. The water disaster monitoring method of claim 7, wherein, The elimination rules include a first elimination rule and a second elimination rule, and specifically include: the first elimination rule is to remove pixels that are simultaneously marked as candidate terrain false dark areas and view angle sensitive false dark areas; the second elimination rule is to set a connected domain area threshold and a peripheral search radius for pixels that are only marked as crop row structure false dark areas; when the pixels that are only marked as crop row structure false dark areas satisfy the connected domain area being less than the connected domain area threshold, and not forming a connection with an initial water body region having a connected domain area greater than the connected domain area threshold in the initial water body region obtained in the range of the peripheral search radius, the corresponding pixels are removed.

9. The water disaster monitoring method of claim 7, wherein, The reservation rules include: pixels that satisfy being marked as candidate terrain false dark areas, view angle sensitive false dark areas or crop row structure false dark areas, but are located within a preset distance threshold range of a known water body.

10. A water regime disaster monitoring system for implementing the water regime disaster monitoring method according to any one of claims 1-9, characterized in that, The method comprises: a first marking module, configured to mark high-risk terrain shadow areas and medium-risk terrain attenuation areas in a pre-acquired main SAR image according to a preset risk threshold; a second marking module, configured to screen candidate terrain false dark areas from pixels marked as high-risk terrain shadow areas and medium-risk terrain attenuation areas in the main SAR image; a third marking module, configured to extract crop row structure false dark areas from the main SAR image; a fourth marking module, configured to mark pixels in the main SAR image as angle-sensitive false dark areas or potential water bodies according to a pre-acquired backup SAR image; a water body screening module, configured to obtain determined water bodies and to-be-verified areas by performing preset rejection and retention operations on pixels marked as candidate terrain false dark areas, crop row structure false dark areas, angle-sensitive false dark areas and potential water bodies in the main SAR image; a result verification module, configured to verify the determined water bodies and the to-be-verified areas to obtain a final water body distribution map.

Images