Flood water body identification method, system and device and storage medium

Abstract

The invention provides a flood water body recognition method, system and device and a storage medium, and belongs to the technical field of flood recognition, and the method comprises the steps: obtaining GRD image data of a target region within a set time period; acquiring a time sequence of grid points for the GRD image data pixel by pixel, and inputting the time sequence into the harmonic regression model to obtain predicted values of pixel points; performing abnormal image recognition according to a difference value between the observed value and the predicted value to obtain a binary flood abnormal image set corresponding to all the GRD images; and counting the precipitation amount of each abnormal flood image in the binary abnormal flood image set in a set time period before and after the set time period, comparing the precipitation amount with the average accumulated precipitation amount in the same time period, and retaining an area exceeding the average accumulated precipitation amount in the abnormal flood images to obtain a flood water body identification result of the target area. The stable identification of flood water bodies in different regional environments is realized, and the identification precision is high.

Description

Technical Field

[0001] The present invention belongs to the technical field of flood identification, and specifically relates to a method, system, device and storage medium for flood water body identification. Background Technique

[0002] As a natural phenomenon triggered by heavy rainfall, continuous rainstorms or long-term concentrated rainfall, flood disasters are one of the most frequent and destructive natural disasters globally. Therefore, timely and accurately monitoring and evaluating the flood inundation range is of great significance for disaster prevention and reduction. With the full coverage and rapid development of global satellites, it is possible to achieve all-weather disaster identification, break through the limitations of traditional flood disaster monitoring methods, improve monitoring efficiency, and save time and labor costs.

[0003] Sentinel-1 satellite carries a synthetic aperture radar to obtain the original SAR image data. The existing flood water body identification methods based on SAR image data mainly include visual interpretation, threshold segmentation, machine learning, and deep learning, etc. Visual interpretation is not only time-consuming and laborious, but also affected by the experience and knowledge level of interpreters, and it is difficult to guarantee the recognition accuracy and consistency. Threshold segmentation is simple and easy to implement, but under complex terrain and weather conditions, due to the interference of factors such as surface roughness and vegetation coverage, it is difficult to select the threshold, which affects the recognition accuracy. Machine learning requires a large number of training samples and computing resources, and the generalization ability of the model is limited. The training process of deep learning models is complex, consumes a large amount of computing resources, and the accuracy is easily affected by data quality and application scenarios. Therefore, there are certain limitations in the accuracy of the existing methods for flood water body identification. Summary of the Invention

[0004] In order to improve the accuracy of flood water body identification, the present invention provides a method, system, device and storage medium for flood water body identification.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A method for flood water body identification specifically includes the following steps:

[0007] Obtain the GRD image data of the target area within a set time period; obtain the observed value of the backscattering intensity of the pixel points within the set time period according to the GRD image data;

[0008] Obtain the time series of grid points for each pixel of the GRD image data; input the time series into a harmonic regression model to obtain the predicted value of the backscattering intensity of the pixel points;

[0009] Perform anomaly identification on the observed value and the predicted value according to the anomaly point identification criterion to obtain a binary flood anomaly image set corresponding to all GRD images within the set time period;

[0010] Statistically analyze the precipitation of each image in the binary flood anomaly image within a set time period, compare it with the average cumulative precipitation in the same time period, retain the binary flood anomaly images with precipitation exceeding the average cumulative precipitation, and obtain the recognition result of flood water bodies in the target area based on the retained binary flood anomaly images.

[0011] Preferably, the harmonic regression model is specifically:

[0012]

[0013] where x represents the Julian day, T represents the number of days in a year, a represents the coefficient of the overall value, b represents the coefficient of the trend change, and c, d represent the coefficients of the seasonal change. represents the predicted value at the Julian day x.

[0014] Preferably, the abnormal point recognition criterion is specifically: the difference between the observed value of the pixel and the model predicted value is less than the empirical threshold and lower than the minimum threshold calculated by the Pauta criterion method.

[0015] Preferably, before obtaining the time series of grid points for each pixel of the GRD image data, preprocessing of the GRD image data is also included, which specifically includes the following steps:

[0016] Eliminate the areas in the GRD image data where the signal-to-noise ratio dB value is less than the set value;

[0017] Remove the mountain shadow, bare land, water body, ice and snow covered areas in the image;

[0018] Perform minimum value synthesis on the images obtained by ascending and descending orbits within a set time, and select the orbit with the largest number of images as the GRD image data set.

[0019] Preferably, it also includes outputting the recognition result of flood water bodies in the target area through the GEE cloud platform, and the GEE cloud platform can select the annual scale, seasonal scale and event scale for the recognition result of flood water bodies.

[0020] Preferably, it also includes using ordinary least squares (OLS) to estimate the coefficients of the harmonic regression model.

[0021] The present invention also provides a flood water body recognition system, which specifically includes:

[0022] An image acquisition module, configured to acquire GRD image data of a target area within a set time period; obtain the observed value of the backscattering intensity of pixel points within the set time period according to the GRD image data;

[0023] A model processing module for obtaining the time series of grid points pixel by pixel from the GRD image data; inputting the time series into a harmonic regression model to obtain the predicted value of the backscattering intensity of the pixel point;

[0024] An outlier recognition module for performing outlier recognition on the observed value and the predicted value according to the outlier recognition criterion to obtain a set of binary flood anomaly images corresponding to all GRD images within a set time period;

[0025] A flood recognition module for statistically calculating the precipitation of each image in the binary flood anomaly images within a set time period, comparing it with the average cumulative precipitation in the same period, retaining the binary flood anomaly images with precipitation exceeding the average cumulative precipitation, and obtaining the flood water body recognition result of the target area according to the retained binary flood anomaly images.

[0026] The present invention also provides a computer device, including a memory, a processor, and a computer program stored on the memory, and the processor executes the computer program to implement the steps in the above-mentioned flood water body recognition method.

[0027] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is loaded by a processor, it can execute the steps in the above-mentioned flood water body recognition method.

[0028] The flood water body recognition method provided by the present invention has the following beneficial effects:

[0029] The present invention obtains the GRD image data of the target area within a set time period, obtains the observed value of the backscattering intensity of the pixel point within the set time period and the time series of grid points obtained pixel by pixel from the GRD image data, and based on the time series, uses the coefficients of the harmonic regression model to obtain the predicted value of the backscattering intensity of the pixel point. The harmonic regression model can flexibly fit the periodic fluctuations in the image, improving the accuracy of the predicted value. According to the outlier recognition criterion, the difference between the observed value and the predicted value is used for outlier image recognition, effectively distinguishing flood water bodies from non-flood water bodies. By statistically calculating the precipitation of each image in the flood anomaly images within a set time period and comparing it with the average cumulative precipitation in the same period, the flood anomaly images, referring to the precipitation deviation factor directly related to flood generation, enhance the reliability of flood water body recognition in the target area. At the same time, it can realize the stable recognition of flood water bodies in different regional environments, with high universality. Description of the Drawings

[0030] To more clearly illustrate the embodiments of the present invention and their design solutions, the following will briefly introduce the accompanying drawings required for this embodiment. The accompanying drawings in the following description are only partial embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0031] Figure 1 It is a flowchart of a method for identifying flood water bodies in an embodiment of the present invention.

[0032] Figure 2 It is a comparison chart of flood events occurring in Area A in an embodiment of the present invention. Among them, Figure 2 (a) is the SAR image of Area A on October 31, 2023; Figure 2 (b) is the SAR image of Area A on November 10, 2023.

[0033] Figure 3 It is the identification result of flood water bodies in Area A in an embodiment of the present invention. Among them, Figure 3 (a) is the monitoring result map of emergency management services; Figure 3 (b) is the flood inundation range map drawn in a standardized manner; Figure 3 (c) is the flood mapping generated by the method used in the present invention.

[0034] Figure 4 It is the flood verification result map in an embodiment of the present invention. Detailed implementation manners

[0035] In order to enable those skilled in the art to better understand the technical solutions of the present invention and be able to implement them, the present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention.

[0036] Embodiment

[0037] The Level-1 products of Sentinel-1A include SLC (Single Look Complex) and GRD (Ground Range Detected). The GRD data is obtained by performing multi-look processing, georegistration, etc. on the Sentinel-1 SAR raw data.

[0038] The present invention provides a method for identifying flood water bodies. As Figure 1 shown, on the GEE cloud platform, it specifically includes the following steps:

[0039] S1. Obtain the ground range images GRD (VH band) of Sentinel-1 from 2015 to the present, and traverse single GRD images for preprocessing, which specifically includes the following steps:

[0040] (1) Generate a new edge file (crop an 8-km buffer from the actual extent of the image), and exclude areas on the image where the SNR dB value is less than -40.

[0041] (2) Remove mountain shadows; currently, the data is sourced from publicly available datasets on GEE: Slope (WWF / HydroSHEDS / 03VFDEM), HAND (users / gena / GlobalHAND / 30m / hand-1000). The slope threshold is determined to be 5°, and the HAND threshold is determined to be 5 m to mask the shadow areas in flat regions where the vertical distance from the nearest water system is greater than 5 m.

[0042] (3) Remove bare land, water, ice, and snow regions; global land use data (ESA / WorldCover / v200).

[0043] (4) Image selection. Mosaic the images acquired within a 6-day range for ascending and descending orbits (minimum value composition), and select the orbit with the largest number of images as the image set to ensure the most comprehensive and representative coverage of flood events. This helps to retain the features most relevant to floods while reducing redundant data.

[0044] S2. Perform anomaly detection based on the preprocessed GRD image data.

[0045] S21. Obtain the time series of grid points for each pixel of the preprocessed GRD image, input the time series into the harmonic regression model, and use ordinary least squares (OLS) to estimate the model coefficients to capture the dynamic changes of the surface. The model includes seasonal, trend, and constant term components.

[0046]

[0047] Among them, x represents the Julian day, T represents the number of days in a year (T = 365), a represents the coefficient of the overall value, b represents the coefficient of the trend change, c and d represent the coefficients of the seasonal change, and y(x) represents the predicted value at Julian day x.

[0048] S22. Use the difference between the actual observed value of each pixel in the pre - processed SAR image and the predicted value output by the harmonic regression model as the basis for identifying anomalies. The observed value of each pixel is the backscatter intensity time series of this point retrieved within a certain time range. After a flood occurs, the scatter intensity of the pixel will suddenly decrease, that is, the observed value decreases, and the difference from the predicted value will be large. Flood identification is carried out through this phenomenon. When considering the difference between the observed value and the predicted value as an anomaly, the following two specific criteria need to be met: a) less than the empirically set predefined threshold of - 5; b) lower than the minimum threshold calculated using the Pauta criterion method. Each GRD image in the image collection has a corresponding binary flood anomaly image (0: normal point, 1: anomaly point).

[0049] S23. For each image identified with anomalies, the ERA5P cumulative precipitation for 10 days before and after this image was recalculated, and these values were compared with the average cumulative precipitation over the years during the same period as the reference background value. The areas in the flood anomaly image where the cumulative precipitation exceeded the corresponding reference background precipitation value were retained, and the remaining areas were ignored. Through this refinement process, it was ensured that only the anomalies directly related to significant precipitation deviations were considered in the analysis as the final flood identification results.

[0050] S24. Use the connectivity function to obtain the target objects, count the number (using 8 - neighborhood statistical analysis), and remove discrete points with less than 6. Removing small, unconnected flood areas helps to improve the overall accuracy and reliability of the flood detection process.

[0051] S25. After separately processing the GRD image data detected in ascending and descending orbits according to S21 to S24, they are merged.

[0052] S3. Image export to obtain the flood water body identification results for the target area. On the GEE cloud platform, three types can be selected: annual scale, seasonal scale, and event scale, which can meet the diverse needs of users whether for long - term trend analysis, seasonal change research, or specific event monitoring.

[0053] Taking the flood event scale in Area A in November 2023 as an example:

[0054] Figure 2 is the Sentinel SAR image obtained within the time range of the flood event, which intuitively reflects the inundation range during the flood. The flood water body has a very low BCSG, which helps to quickly and accurately locate the flood inundation area.

[0055] Figure 3 is the flood water body identification result, Figure 3 (a) of which is the emergency management service monitoring result map (EMSR), Figure 3Part (b) is a flood inundation area map (Z - score) drawn in a standardized manner. Figure 3 Part (c) is a display of the flood mapping (HRM) generated by the method used in the present invention. It can be seen that the flood water body identification results generated by this product are relatively consistent with the spatial distribution of EMSR.

[0056] Figure 4 This is the verification of random point results. By randomly selecting 1000 verification points and using EMSR as the verification benchmark, Z - score and HRM are tested. If the point falls within the flood range defined by EMSR, it is regarded as a flood point and marked as 1; conversely, if the point is outside the EMSR range, it is regarded as a non - flood point and marked as 0. To measure the accuracy of classification, the key parameter of overall accuracy is adopted. The overall accuracy calculated according to the Z - score method is 0.74, which means that among the 1000 randomly selected points, the flood status (whether it is a flood point or a non - flood point) of 74% of the points is correctly classified. However, when using the model proposed in this patent for flood water body classification, the overall accuracy is increased to 0.78, that is, the flood status of 78% of the measured points is correctly classified. Compared with the Z - score method, the overall accuracy of the method of this patent is increased by 5 percentage points, and this improvement is statistically significant, indicating that the model of this patent has superior performance in flood water body identification compared with traditional methods.

[0057] The present invention also provides a flood water body identification system, specifically including:

[0058] An image acquisition module, used to acquire GRD image data of the target area within a set time period; and obtain the observed values of the backscattering intensity of pixel points within the set time period according to the GRD image data.

[0059] A model processing module, used to obtain the time series of grid points for each pixel of the GRD image data; input the time series into the harmonic regression model to obtain the predicted values of the backscattering intensity of pixel points.

[0060] An outlier identification module, used to perform outlier identification on the observed values and predicted values according to the outlier identification criterion to obtain a set of binary flood anomaly images corresponding to all GRD images within the set time period.

[0061] A flood identification module, used to count the precipitation in each image in the binary flood anomaly images within the set time period, compare it with the average cumulative precipitation in the same period, retain the binary flood anomaly images with precipitation exceeding the average cumulative precipitation, and obtain the flood water body identification result of the target area according to the retained binary flood anomaly images.

[0062] Each module in the above-mentioned flood water body recognition system can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above modules can be embedded in the processor of a computer device in hardware form or be independent of it, or stored in the memory of the computer device in software form so that the processor can call and execute the operations corresponding to each of the above modules.

[0063] The present invention also provides a computer device, including a memory, a processor, and a computer program stored on the memory. The processor executes the computer program to implement the steps in an embodiment of a flood water body recognition method. For the specific implementation method, reference can be made to the method embodiment, which will not be elaborated here.

[0064] Furthermore, the present invention also provides a non-transitory computer-readable storage medium containing instructions, on which a computer program is stored. For example, a memory containing instructions. The above instructions can be executed by the processor of a computer device to complete the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc. When the computer program is executed by the processor, it can implement the steps in an embodiment of a flood water body recognition method. For the specific implementation method, reference can be made to the method embodiment, which will not be elaborated here.

[0065] Those skilled in the art should understand that the embodiments of the present invention can provide a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0066] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0067] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements in the process Figure 1 one process or a plurality of processes and / or blocks Figure 1 the functions specified in one block or a plurality of blocks.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, such that a series of operational steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in Figure 1 one process or a plurality of processes and / or blocks Figure 1 one block or a plurality of blocks.

[0069] It should be noted that the above-described specific embodiments may enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present specification and embodiments have described the present invention in detail, those skilled in the art should understand that the present invention may still be modified or equivalently replaced; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered by the protection scope of the patent of the present invention. Any reference numerals in the claims should not be construed as limiting the claimed claims. Any simple variations or equivalent replacements of technical solutions that can be obviously obtained by any person skilled in the art within the technical scope disclosed by the present invention fall within the protection scope of the present invention.

Claims

1. A method for identifying flood water bodies, characterized in that, It includes the following steps: Obtain the GRD image data of the target area within a set time period; obtain the observed values of the backscattering intensity of pixel points within the set time period according to the GRD image data; Obtain the time series of grid points pixel by pixel for the GRD image data; input the time series into the harmonic regression model to obtain the predicted values of the backscattering intensity of pixel points; Perform anomaly identification on the observed values and predicted values according to the anomaly point identification criterion to obtain a set of binary flood anomaly images corresponding to all GRD images within the set time period; Statistically calculate the precipitation of each image in the binary flood anomaly images within the set time period, compare it with the average cumulative precipitation in the same period, retain the binary flood anomaly images with precipitation exceeding the average cumulative precipitation, and obtain the flood water body identification result of the target area according to the retained binary flood anomaly images.

2. The method for identifying flood water bodies according to claim 1, characterized in that The harmonic regression model is specifically: Among them, x represents the Julian day, T represents the number of days in each year, a represents the coefficient of the overall value, b represents the coefficient of the trend change, and c, d represent the coefficients of the seasonal change. represents the predicted value at the Julian day x.

3. The method for identifying flood water bodies according to claim 1, wherein The anomaly point identification criterion is specifically: the difference between the observed value and the model predicted value of a pixel is less than the empirical threshold and lower than the minimum threshold calculated using the Pauta criterion method.

4. A method for identifying flood water bodies according to claim 1, characterized in that, Before obtaining the time series of grid points pixel by pixel for the GRD image data, it also includes preprocessing the GRD image data, specifically including the following steps: Eliminate the areas in the GRD image data where the signal-to-noise ratio dB value is less than the set value; Remove the mountain shadow, bare land, water body, ice and snow covered areas in the image; Perform minimum value synthesis on the images obtained in the ascending and descending orbits within the set time, and select the orbit with the largest number of images as the GRD image data set.

5. A method for identifying flood water bodies according to claim 1, characterized in that, It also includes outputting the flood water body identification result of the target area through the GEE cloud platform, and the GEE cloud platform can select the annual scale, seasonal scale and event scale for the flood water body identification result.

6. The method for identifying flood water bodies according to claim 2, characterized in that, It also includes using ordinary least squares (OLS) to estimate the coefficients of the harmonic regression model.

7. A flood water body identification system, characterized in that, It includes: An image acquisition module for obtaining the GRD image data of the target area within a set time period; Obtaining the observed values of the backscattering intensity of pixel points within the set time period according to the GRD image data; A model processing module for obtaining the time series of grid points pixel by pixel for the GRD image data; inputting the time series into the harmonic regression model to obtain the predicted values of the backscattering intensity of pixel points; An anomaly point identification module for performing anomaly identification on the observed values and predicted values according to the anomaly point identification criterion to obtain a set of binary flood anomaly images corresponding to all GRD images within the set time period; A flood identification module for statistically calculating the precipitation of each image in the binary flood anomaly images within the set time period, comparing it with the average cumulative precipitation in the same period, retaining the binary flood anomaly images with precipitation exceeding the average cumulative precipitation, and obtaining the flood water body identification result of the target area according to the retained binary flood anomaly images.

8. A computer device, comprising a memory, a processor, and a computer program stored on the memory, characterized in that The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is loaded by the processor, it can execute the steps of the method according to any one of claims 1 to 6.

Images