Multi-source data flood inundation range analysis method based on evidence weight

Abstract

The invention provides a multi-source data flood inundation range analysis method based on evidence weight, and the method comprises the steps: data obtaining, multi-source data space overlay analysis, multi-source data inundation feature classification, inundation feature weight coefficient calculation, grid flood inundation logarithm posterior probability calculation, and grid flood inundation probability calculation after fusion. And determining a flood inundation probability threshold. According to the method, an evidence weight method is used as a multi-source data fusion weight calculation method, three kinds of submerging data of flood model simulation, remote sensing monitoring interpretation and socialized investigation are fused, the problems of unreasonable submerging range or incomplete coverage caused by errors, method defects and the like of a single data source are solved, and the analysis precision of the submerging range can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of flood disaster management and planning, and particularly relates to a method for analyzing flood inundation range based on multi-source data with evidence weight. Background Art

[0003] After a flood disaster occurs, quickly locating the disaster area and obtaining the flood inundation area are the basis for understanding the disaster impact, assessing the disaster losses, and carrying out flood control emergency and disaster relief work, providing key support for formulating disaster relief plans and allocating personnel and materials. Currently, there are various means for evaluating the flood inundation range, including flood model analysis, remote sensing image analysis, simplified GIS methods, etc.

[0004] The flood model can quickly conduct analysis and timely determine the flood inundation or water accumulation area in the modeling area. However, it has high requirements for modeling data. Especially in the case of lack of basic data, unclear current working conditions, and fuzzy future working condition development trends, the model calculation results often have large deviations. In addition, the flood model also relies on historical flood data to carry out parameter calibration work.

[0005] Remote sensing image analysis can interpret large-scale inundation range data. However, the acquisition of images is restricted by the satellite overpass frequency. At the same time, factors such as aerial cloud cover and ground crops will also cause large errors in the interpreted inundation range.

[0006] The simplified GIS analysis method is mainly based on DEM data and uses GIS for spatial analysis to obtain results. This method has deficiencies in terms of timeliness, accuracy, and the degree of reflection of working conditions.

[0007] In addition to the above technical means, multimedia photos, videos circulated on the Internet, and relevant discussions on Weibo can also provide certain auxiliary information on flood inundation. In addition, China has established a grid member disaster reporting system. After a flood occurs, grid members in various places will report key disaster-affected points in key areas, and the reported data content and format are relatively unified, which can be used as a powerful supplement to flood inundation information at each stage.

[0008] During the occurrence of a flood disaster, both technical means and socialized information reporting have laid the foundation for determining the inundation range. However, due to various factors such as the defects of its own methods and the limitations of external basic data, the single flood disaster source information has problems such as incomplete coverage of flood disaster information sources and large errors in the inundation range.

[0009] Therefore, there is an urgent need for an analysis method that can effectively integrate multiple flood inundation information sources, make up for the shortcoming of incomplete information source coverage, reduce the error of the inundation range, and thus obtain the inundation range analysis method that is most matched or closest to the actual situation in a timely and accurate manner when flood disasters occur. Summary of the Invention

[0010] Aiming at the problems existing in the prior art, the purpose of the present invention is to provide a multi-source data flood inundation range analysis method based on evidence weight. This method effectively integrates multiple flood inundation information sources through specific mathematical methods, strengthens the real area, reduces or weakens the false and incorrect areas, so as to obtain the inundation range that is most matched or closest to the actual situation, and can solve the problems of missing, incomplete or excessive error of the inundation range during the development of flood disasters, and provide more accurate data for flood control emergency rescue.

[0011] The purpose of the present invention is achieved through the following technical solutions:

[0012] The present invention provides a multi-source data flood inundation range analysis method based on evidence weight, including the following steps:

[0013] Step 1, data acquisition:

[0014] Determine the study area, and obtain the flood model simulation data, remote sensing inundation range data, social survey data and currently known flood inundation true value data of the study area.

[0015] Step 2, spatial overlay analysis of multi-source data:

[0016] Based on the grid of the flood model simulation data obtained in Step 1, overlay the remote sensing inundation range data and social survey data respectively, specifically including:

[0017] S21, spatial overlay analysis of remote sensing inundation range data and flood model simulation data

[0018] Cut the remote sensing flood inundation range map according to the grid size of the flood model, identify the inundation range blocks cut by the grid with grid numbers, and establish the association relationship between the remote sensing flood inundation range map and the grid to form a spatial layer of whether the remote sensing image is inundated or not indexed by grid numbers;

[0019] S22, spatial overlay analysis of social survey data and flood model simulation data

[0020] Overlay the social disaster situation data map and the flood model grid spatially, identify the grid numbers of the social disaster situation data, and establish the relationship between the social disaster situation data and the grid.

[0021] Step 3, inundation feature classification of multi-source data:

[0022] Classify the inundation characteristics of the flood model simulation data, remote sensing inundation range data, and social survey data obtained in Step 1 as classification evidence for flood inundation, specifically including:

[0023] S31, Classification of flood model simulation data

[0024] The data calculated by the flood model includes various information such as the inundation depth, inundation duration, and inundation velocity of each grid. Using the inundation depth of each grid as a characteristic index, the data is divided into several different levels to establish evidence information;

[0025] S32, Classification of remote sensing inundation range data

[0026] The remote sensing inundation range data refers to the flood inundation or waterlogging range of the research area obtained by remote sensing monitoring. Using whether it is inundated as a characteristic index, the data is divided into several different levels to establish evidence information;

[0027] S33, Classification of social survey data

[0028] The social survey data refers to the standard data reported by communities, villages, or grid workers. The data attributes include 7-level descriptions of the inundation points being above the knees, above the waist, above the chest, above the head, above the window, above the roof beam, and above the roof. The data is divided into several different levels to establish evidence information.

[0029] Step 4, Calculation of inundation characteristic weight coefficients:

[0030] Calculate the weight coefficients for different levels of inundation characteristics after classification in Step 3, including positive weights and negative weights The calculation formula is:

[0031]

[0032] In the formula, i is the inundation characteristic classification of multi-source data, A1 is the area that reaches a certain classification and is actually inundated in a certain inundation characteristic data, A2 is the total area of actual inundation corresponding to A1 but not at this classification, is the area that reaches a certain classification and is actually not inundated in a certain inundation characteristic data, is corresponding to and is the total area of not being inundated corresponding to this classification and not actually inundated.

[0033] Step 5, Calculation of the logarithmic posterior probability of grid flood inundation:

[0034] Taking the grids of the two-dimensional flood model as units, calculate the logarithmic posterior probability L of flood inundation for each grid. The formula is as follows:

[0035]

[0036] In the formula, P(D) refers to the prior probability of flood inundation in the area, and W i refers to the value taken according to all feature classifications based on whether the multi-source data attributes of the grid meet a certain feature classification. If the grid meets a certain special classification, the corresponding value is If not, the corresponding value is i is the inundation feature classification of the multi-source data, and n is the total number of inundation feature classifications.

[0037] Step 6, calculation of the flood inundation probability of the fused grid:

[0038] Taking the grid of the two-dimensional flood model as a unit, calculate the flood inundation probability of each grid. The formula is as follows:

[0039] P k = exp(L k ) / (1 + exp(L k ))

[0040] In the formula, P k refers to the flood inundation probability after the grid fuses the multi-source data, L k refers to the flood inundation log posterior probability of each grid, and k refers to the grid serial number of the two-dimensional flood model.

[0041] Step 7, determination of the flood inundation probability threshold:

[0042] Taking the social survey data obtained in Step 1 as verification points, compare the flood inundation probability map of each fused grid with the verification points, and take a certain inundation probability value as the flood inundation threshold. When the flood inundation situation of a certain proportion of verification points is consistent with the fused flood inundation range, it is considered that the inundation range meets the requirements, and select the grids with a flood inundation probability greater than or equal to this threshold as the flood inundation range;

[0043] During the flood development process, for the flood model simulation data, remote sensing inundation range data, and social survey data obtained in each stage, repeat the above Steps 1-7 for iterative analysis to obtain the flood inundation range in each time period, so as to realize the analysis of the flood inundation range in the research area.

[0044] The beneficial effects of the present invention compared with the prior art are as follows:

[0045] 1. The present invention uses the evidence weight method as the calculation method for the weight of multi-source data fusion, fuses three types of inundation data: flood model simulation, remote sensing monitoring interpretation, and social survey, breaks through the problems of unreasonable inundation range or incomplete coverage caused by single data source due to errors, method defects, etc., and can improve the accuracy of inundation range analysis;

[0046] 2. The weight-of-evidence method is a statistical analysis method based on Bayes' law and probability uncertainty. The present invention sorts out the inundation characteristics of each flood data source, uses the weight-of-evidence method to establish the weight values of each inundation characteristic, represents the contribution degree of each inundation evidence to flood inundation prediction, synthesizes the elements that have important contribution value to the prediction result, and then obtains the flood inundation logarithmic posterior probability, which is used to represent the possibility of each grid cell in the study area being inundated. Further, the posterior probability of the fusion area is calculated to determine the inundation range that best matches or is closest to the actual situation, providing strong support for flood control emergency rescue and disaster relief. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The present invention will be further described below in conjunction with the drawings and embodiments:

[0048] Figure 1 is a schematic flow chart of the multi-source data flood inundation range analysis method based on the weight of evidence according to the present invention;

[0049] Figure 2 shows the basic situation of the flood detention and retention area and the hydrological processes of Dongcicun in the North Ditch River and Beihedian in the South Juma River;

[0050] Figure 3 shows the multi-source data of the maximum inundation range in the embodiment;

[0051] Figure 4 is a remote sensing monitoring map of the flood analysis grid superposition;

[0052] Figure 5 is a water depth grading map of the flood model simulation inundation range;

[0053] Figure 6 is a water depth grading map of the social survey points;

[0054] Figure 7 shows the flood inundation logarithmic posterior probability after multi-source data fusion;

[0055] Figure 8 is a comparison chart of the multi-source data fusion effect. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] To make the objectives, technical solutions and advantages of the present invention clearer, the following will further describe the disclosed embodiments of the present invention in detail with reference to the drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0057] Embodiment 1

[0058] As Figure 1As shown in the figure, this embodiment provides a method for analyzing the flood inundation range of multi-source data based on the weight of evidence, taking the Langgouwa flood detention and retarding area in the Haihe River Basin as an example area. Starting from July 31, 2023, flood diversion was initiated in the Langgouwa flood detention and retarding area, with a total of 7 flood diversions at Xiaoyingheng Dike, Xicicun (2 locations), Dongxinzhuang, Zhuzhuang, Dongwu, and Liyi Pukou. As the flood levels of the Beigou River and the South Juma River dropped, the flood detention area discharged floodwaters through the Dongmaying Pukou at 18:00 on August 6. By August 22, the floodwaters had basically receded, with only a small amount of standing water in local low-lying areas. During the flood diversion operation in Langgouwa, flood simulation, remote sensing monitoring, and social surveys were continuously carried out. This embodiment takes the fusion of the maximum flood inundation range during the flood diversion process as an example.

[0059] The method for analyzing the flood inundation range of multi-source data based on the weight of evidence includes the following steps:

[0060] Step 1, obtaining data:

[0061] Determine the study area and obtain the flood model simulation data, remote sensing inundation range data, social survey data, and currently known flood inundation true value data of the study area; among them, the true value data is the currently known flood inundation range obtained through actual measurement or other means and confirmed to be accurate.

[0062] The flood model simulation data is the flood inundation range of the study area calculated using a hydrodynamic model, including the inundation depth, inundation duration, and inundation velocity of each grid during the flood process, as well as the maximum inundation depth during each single simulation process of each grid. The data format is vector surface data. In this embodiment, to determine the flood evolution and recession during the flood diversion process, a one-dimensional and two-dimensional flood analysis model of the Baigou River, South Juma River, Daqing River, and flood detention and retarding area is established. In the one-dimensional model, the measured and predicted flow data of Dongcicun Station for the Baigou River and Beihedian Station for the South Juma River are used as boundary conditions ( Figure 2 ), and the lower boundary of the Daqing River adopts the water level - flow relationship; the modeling range extends north from Xiaoyingheng Dike, east to the right dike of the Baigou River, south to the left dike of the South Juma River, and west is bounded by exceeding the current designed flood detention level. The modeling area is approximately 523 km 2 , as Figure 3 shown, the area covered by the grid is the modeling range and also the study area mentioned in this embodiment. Using the measured and predicted hydrological processes of Dongcicun Station and Beihedian Station, the inundation range and water depth distribution of the flood detention and retarding area at different times can be continuously simulated.

[0063] The remote sensing inundation range data is the flood inundation or waterlogging range obtained by interpreting remote sensing images, and the data format is vector surface data. In this embodiment, satellite remote sensing monitoring data during the flood detention of the flood detention area is collected, and the remote sensing flood inundation ranges at each stage are combined to obtain the maximum flood inundation range monitored by remote sensing, as Figure 3 shown.

[0064] The social survey data refers to the data obtained through surveys, etc., and the data format is vector point data. In this embodiment, the social survey data in the northern and southern parts of the flood detention and retention area is collected, with a total of 164 points. The main attribute reflecting the flood inundation is the depth of the survey points, such as Figure 3 shown.

[0065] Collect the data layer of the currently known flood inundation range during the flood diversion process in the flood detention and retention area as the current inundation truth value, which is vector polygon data.

[0066] Step 2, spatial overlay analysis of multi-source data:

[0067] Based on the grid of the flood model simulation data obtained in Step 1, overlay the remote sensing inundation range data and the social survey data with it respectively. Specifically, it includes:

[0068] S21, spatial overlay analysis of remote sensing inundation range data and flood model simulation data

[0069] Overlay the remote sensing flood inundation range map (i.e., the maximum flood diversion inundation range monitored by remote sensing in this embodiment) with the grid of the flood model simulation data. The remote sensing flood inundation range map is cut by the edges of each grid and divided into blocks not larger than the original grid size. Identify each block and assign the grid number where it is located, and establish the association relationship between the remote sensing flood inundation range map and the grid to form a spatial layer of whether the remote sensing image is inundated indexed by the grid sequence number, as shown in Figure 4 .

[0070] S22, spatial overlay analysis of social survey data and flood model simulation data

[0071] Select a certain proportion of survey points for data fusion, and the remaining survey points are used for selecting the inundation probability threshold after fusion and verifying the inundation range. In this embodiment, 80% of the survey points are taken for fusion, and the remaining 20% of the points are used for verification. Overlay the social survey data used for fusion with the flood model grid to analyze the grid number where the social survey data points are located.

[0072] Step 3, classification of inundation characteristics of multi-source data:

[0073] Classify the inundation characteristics of the flood model simulation data, remote sensing inundation range data, and social survey data obtained in Step 1 as classification evidence for flood inundation. Specifically, it includes:

[0074] S31, classification of flood model simulation data

[0075] The data calculated by the flood model includes various types of information such as the inundation depth, inundation duration, and inundation velocity of each grid. Taking the inundation depth of each grid as a characteristic index, the data is divided into several different levels to establish evidence information. Specifically, taking the inundation depth of each grid as a characteristic index, according to the inundation depth h < 0.05; 0.05 ≤ h < 0.5; 0.5 ≤ h < 1; 1 ≤ h < 2; 2 ≤ h < 3 and h ≥ 3 (unit: m), it is divided into 6 flood inundation levels, as shown in Figure 5 .

[0076] S32, Classification of Remote Sensing Inundation Range Data

[0077] Remote sensing inundation range data refers to the flood inundation or water accumulation range of the study area obtained by remote sensing monitoring. Taking whether it is inundated as a characteristic index, the data is divided into several different levels to establish evidence information. Specifically, in this embodiment, after the remote sensing flood inundation range map is overlaid with the grid, it is divided into two levels: inundated and non-inundated. The grid inundation attribute of the area covered by the inundation range is assigned a value of 1, indicating that the remote sensing monitoring believes that this area is inundated by the flood; the grid inundation attribute of the area not covered is assigned a value of 0, indicating that the remote sensing monitoring believes that this area is not inundated by the flood; as Figure 4 shown in the figure, the block grid attributes of the area outside the remote sensing inundation range are assigned a value of 0.

[0078] S33, Classification of Social Survey Data

[0079] Social survey data refers to the standard data reported by community, village, or grid workers. The data attributes include 7-level descriptions of the inundation points passing over the knees, over the waist, over the chest, over the head, over the window, over the roof beam, and over the roof. The water depths correspond to h < 0.4; 0.4 ≤ h < 1.0; 1.0 ≤ h < 1.5; 1.5 ≤ h < 1.8; 1.8 ≤ h < 2.0; 2.0 ≤ h < 3.0 and h ≥ 3.0 (unit: m), and are divided into 7 inundation levels, as Figure 6 described.

[0080] Step 4, Calculation of Inundation Feature Weight Coefficient:

[0081] For the inundation features of different levels after classification in Step 3 (including 6 levels of flood model simulation data, 2 levels of remote sensing inundation range data, and 7 levels of social survey data), weight coefficient calculation is performed, including positive weights and negative weights

[0082] S41, Random Sampling of the Study Area and Calculation of Inundation Prior Probability

[0083] Randomly sample N grids in the study area, and based on the flood inundation information of the currently known true flood inundation data, count the inundation situations of these N sampling points, and calculate the prior probability of flood inundation. The formula is as follows:

[0084] P(D) = A 淹 / A 总

[0085] In the formula, P(D) refers to the prior probability of flood inundation in the area, A 淹 refers to the total inundated area of the sampled grids, and A 总 refers to the total area of the sampled grids.

[0086] In this embodiment, the model grid area is used as the evaluation area, and a random sample is taken at 3% of the total number of grids. The inundation situation of the true inundation before fusion is counted, and the prior probability of flood inundation in the study area is calculated, which is approximately 48%.

[0087] S42. Statistically count the positive information of flood inundation

[0088] In the N sampled grids, for the flood model simulation data, remote sensing inundation range data, and social survey data, calculate the area A where the corresponding inundation characteristics of each type of data reach a certain classification level and are actually inundated 1淹j_i to calculate the total actual inundated area Sum_A of flood in a certain type of data 1淹j The formula is as follows:

[0089]

[0090] In the formula, i refers to the inundation characteristic classification of multi-source data, j refers to the data type, and m is the number of characteristic classifications of a certain type of data in multi-source data;

[0091] Calculate the total area A where inundation actually occurs but is not at this classification level 2淹j_i The formula is as follows:

[0092] A 2淹j_i = Sum_A 1淹j - A 1淹j_i

[0093] The calculation results are shown in the following table:

[0094]

[0095]

[0096] S43. Statistically count the negative information of flood inundation

[0097] In N sampling grids, for the flood model simulation data, remote sensing inundation range data, and social survey data, calculate the area where the corresponding inundation characteristics of each type of data reach a certain classification level and are actually not inundated. Thus, calculate the total area where the flood is actually not inundated in a certain type of data. The formula is as follows:

[0098]

[0099] Calculate the total area where inundation actually does not occur for non - this classification level. The formula is as follows:

[0100]

[0101] The calculation results are shown in the following table:

[0102]

[0103] S44. Calculate the positive and negative weights of 15 characteristic classification levels for 3 types of data, that is, and the values of. The calculation formula is:

[0104]

[0105]

[0106] In the formula, i is the inundation characteristic classification level of multi - source data, A1 is the area that reaches a certain classification level and is actually inundated in a certain inundation characteristic data, A2 is the total area where inundation actually occurs for non - this classification level corresponding to A1, is the area that reaches a certain classification level and is actually not inundated in a certain inundation characteristic data, is for the total area where inundation actually does not occur for non - this classification level corresponding to.

[0107] The calculation results are shown in the following table:

[0108]

[0109] Step 5. Calculate the logarithmic posterior probability of grid flood inundation:

[0110] Taking the grids of the two - dimensional flood model as units, calculate the logarithmic posterior probability L of flood inundation for each grid. The calculation results are as Figure 7 shown. The formula is as follows:

[0111]

[0112] In the formula, P(D) refers to the prior probability of flood inundation in the region, W iThe pointer grades all features and takes values according to whether the multi-source data attributes of the grid meet a certain feature grade. If they meet a certain special grade, the corresponding value is If not, the corresponding value is i is the inundation feature grade of the multi-source data, and n is the total number of inundation feature grades.

[0113] Step 6, calculation of the flood inundation probability after fusion:

[0114] Taking the grids of the two-dimensional flood model as units, calculate the flood inundation probability of each grid. The formula is as follows:

[0115] P k =exp(L k ) / (1+exp(L k ))

[0116] In the formula, P k refers to the flood inundation probability after the grid fuses multi-source data, L k refers to the flood inundation log posterior probability of each grid, and k refers to the grid serial number of the two-dimensional flood model.

[0117] Step 7, determine the flood inundation probability threshold:

[0118] Taking the social survey data obtained in Step 1 as verification points, compare the flood inundation probability map of each grid after fusion with the verification points, and take a certain inundation probability value as the flood inundation threshold. When the flood inundation situation of a certain proportion of verification points is consistent with the flood inundation range after fusion, it is considered that the inundation range meets the requirements, and select the grids with the inundation probability greater than or equal to this threshold as the flood inundation range.

[0119] During the flood development process, for the flood model simulation data, remote sensing inundation range data, and social survey data obtained in each stage, repeat the above Steps 1-7 for iterative analysis to obtain the flood inundation range of each time period, so as to realize the analysis of the flood inundation range in the research area.

[0120] Compare the flood survey points with the inundation ranges under different inundation probabilities, take P>20% as the threshold of flood inundation, and generate a flood inundation layer, as Figure 8 shown.

[0121] Finally, it should be noted that the above is only used to illustrate the technical solution of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred arrangement, those of ordinary skill in the art should understand that the technical solution of the present invention (such as the application of various formulas, the sequence of steps, etc.) can be modified or equivalently replaced without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A flood inundation range analysis method based on multi-source data based on evidence weight, characterized in that: The method comprises the following steps: Step 1, get the data: Determine the study area, obtain flood model simulation data, remote sensing inundation range data, social survey data and currently known flood inundation true value data of the study area; Step 2: Multi-source data spatial overlay analysis: Taking the grid of flood model simulation data obtained in step 1 as the benchmark, remote sensing flooding range data and social survey data are superimposed on it, including: S21, Spatial overlay analysis of remote sensing inundation range data and flood model simulation data The remote sensing flood inundation range map is cut according to the grid size of the flood model, and the grid numbers are used to identify the inundation range blocks cut out by the grids, so as to establish the association between the remote sensing flood inundation range map and the grids, and form a spatial layer of whether the remote sensing image is flooded or not, with the grid numbers as indexes; S22, Spatial overlay analysis of social survey data and flood model simulation data The social disaster data map is spatially superimposed with the flood model grid, the grid sequence number of the social disaster data is marked, and the relationship between the social disaster data and the grid is established; Step 3: Multi-source data flood feature classification: The flood model simulation data, remote sensing flood range data, and social survey data obtained in step 1 are classified into different levels of flood characteristics as evidence for flood classification, including: S31, Classification of flood model simulation data The data calculated by the flood model include various information such as the flood depth, flood duration and flood flow rate of each grid. The flood depth of each grid is used as a characteristic indicator to divide the data into several different levels to establish evidence information. S32, Classification of Remotely Sensed Inundation Extent Data Remote sensing inundation range data refers to the flood inundation or waterlogging range of the research area obtained by remote sensing monitoring. Whether it is inundated is used as a characteristic indicator, and the data is divided into several different levels to establish evidence information; S33, Classification of Social Survey Data Social survey data refers to standard data reported by communities, villages or grid workers. The data attributes include seven levels of description: flood points above the knees, above the waist, above the chest, above the head, above the windows, above the beams and above the roof. The data are divided into several different levels to establish evidence information. Step 4, calculation of flooding characteristic weight coefficient: Calculate the weight coefficients for the different levels of flooding characteristics after classification in step 3, including positive weights and negative weights The calculation formula is: In the formula, i is the flooding characteristic classification of multi-source data, A1 is the area that reaches a certain classification in a certain flooding characteristic data and is actually flooded, and A2 is the total area that is actually flooded but not in this classification corresponding to A1. It is the area that reaches a certain level in a certain flood characteristic data and is not actually flooded. For The corresponding total area that is not classified and is not actually flooded; Step 5, calculation of the logarithmic posterior probability of grid flooding: Taking the grid of the two-dimensional flood model as the unit, the logarithmic posterior probability L of flood inundation in each grid is calculated. The formula is as follows: Where P(D) refers to the prior probability of flooding in the region, W i Refers to the classification of all features. The value is determined based on whether the multi-source data attributes of the grid meet a certain feature classification. If it meets a certain special classification, the corresponding value is If it does not meet the corresponding value i is the flooding characteristic classification of multi-source data, and n is the total number of flooding characteristic classifications; Step 6: Calculation of flood inundation probability of the fused grid: Taking the grid of the two-dimensional flood model as the unit, the flood inundation probability of each grid is calculated. The formula is as follows: P k =exp(L k ) / (1+exp(L k )) Where P k Refers to the flood inundation probability after the grid integrates multi-source data, L k refers to the logarithmic posterior probability of flood inundation in each grid, and k refers to the grid number of the two-dimensional flood model; Step 7, determine the flood inundation probability threshold: The social survey data obtained in step 1 is used as the verification point, and the flood inundation probability map of each grid after fusion is compared with the verification point. A certain inundation probability value is taken as the flood inundation threshold. When the flood inundation situation of a certain proportion of verification points is consistent with the flood inundation range after fusion, it is considered that the inundation range meets the requirements, and the grids with inundation probability greater than or equal to the threshold are selected as the flood inundation range; During the development of the flood, the above steps 1-7 are repeated for the flood model simulation data, remote sensing inundation range data, and social survey data obtained at each stage, and an iterative analysis is performed to obtain the flood inundation range in each time period, thereby realizing the flood inundation range analysis of the study area.

2. The method according to claim 1, characterized in that In step 3, S31, when grading the flood model simulation data, the flood depth grading standard of each grid is divided into h<0.05; 0.05≤h<0.5; 0.5≤h<1; 1≤h<2; 2≤h<3 and h≥3, which is divided into 6 levels in total; In step 3, S32, when the remote sensing flood range data is classified, after the remote sensing flood range map is superimposed on the grid, the grid flood attribute of the area covered by the flood range is assigned a value of 1, indicating that the remote sensing monitoring believes that the area is flooded; the grid flood attribute of the uncovered area is assigned a value of 0, indicating that the remote sensing monitoring believes that the area is not flooded; In step 3 S33, when the social survey data is classified, the flooding point is described in 7 levels: over the knee, over the waist, over the chest, over the head, over the window, over the beam and over the roof, and the water depths correspond to h<0.4, 0.4≤h<1.0, 1.0≤h<1.5, 1.5≤h<1.8, 1.8≤h<2.0, 2.0≤h<3.0 and h≥3.0 respectively.

3. The method according to claim 1, characterized in that In step 4 S41, the weight coefficient calculation specifically includes: S41, random sampling of the study area and calculation of flooding probability N grids are randomly sampled in the study area, and the flooding situation of these N sampling points is statistically calculated based on the flooding information of the currently known true flooding data, and the prior probability of regional flooding is calculated. The formula is as follows: P(D)=A 淹 / A 总 Where P(D) refers to the prior probability of flooding in the region, A 淹 Refers to the total flooded area of ​​the sampling grid, A 总 Refers to the total area of ​​the sampling grid; S42, statistics on positive information of flooding In N sampling grids, for flood model simulation data, remote sensing inundation range data, and social survey data, calculate the area A where the corresponding inundation characteristics of each type of data reach a certain level and are actually inundated 1淹j_i , thereby calculating the total area actually flooded by floods in a certain type of data Sum_A 1淹j , the formula is as follows: In the formula, i refers to the flooding feature classification of multi-source data, j refers to the data type, and m is the number of feature classifications of a certain type of data in the multi-source data; Calculate the total area A that is not classified and actually flooded 2淹j_i , the formula is as follows: A 2淹j_i =Sum_A 1淹j -A 1淹j_i S43, statistics of negative flood information In N sampling grids, for flood model simulation data, remote sensing inundation range data, and social survey data, calculate the area where the inundation characteristics of each type of data reach a certain level and are not actually inundated. Thus, the total area not actually flooded by flood in a certain type of data can be calculated. The formula is as follows: Calculate the total area that is not classified and is not actually flooded The formula is as follows:

Images