Rainfall similarity analysis method based on rainfall color map

Abstract

The application discloses a rain amount similarity analysis method based on rain amount color spot maps, and first proposes using rain amount color spot data to analyze rain amount similarity, solves the secrecy problem of traditional rain amount numerical data, and effectively combines rain amount and rain space information by using color and space information in the rain amount color spot data as search features. And according to the rain trend of historical rain amount, the current rain trend is judged, and data support is provided for flood control monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of rainfall analysis, in particular to a rainfall similarity analysis method based on a rainfall color map. BACKGROUND

[0002] The trend of rainfall directly affects the development of flood conditions. Wang Guoan once proposed in the book "Hydrological Theorems, Laws and Hypotheses" published by Yellow River Water Conservancy Press in 2022 that for two rainstorm floods in a specific river basin, if their flood processes are highly similar in time and space, the rainstorm and weather processes are also similar. Therefore, for a specific river basin, according to the current rainfall conditions, the most similar rainfall event is found from the historical rainfall, and the trend of the current rainfall is deduced according to the trend of the rainfall in the historical rainfall event, which can effectively deduce the evolution law of the current flood, and further improve the command and decision-making ability of the relevant personnel in flood control.

[0003] However, most of the current rainfall similarity analysis methods use some similarity analysis models to analyze numerical data such as hydrological stations, rainfall stations, daily rainfall, and rainfall summaries. To some extent, it can effectively calculate the similarity between different rainfall events. However, these numerical rainfall data feature indicators can only represent the rainfall at the location of the hydrological station or rainfall station, and cannot effectively express the spatial and geographical location information of the rainfall, so there are great limitations. Moreover, the numerical data used for rainfall similarity analysis is confidential and not easy to obtain for water conservancy flood control personnel, further limiting the ability of the water conservancy flood control field to accurately analyze the trend of rainfall similarity. SUMMARY

[0004] The present application aims to provide a rainfall similarity analysis method based on a rainfall color map, which solves the problem that the current rainfall similarity analysis cannot indicate the spatial and geographical location information of the rainfall, and the rainfall numerical data is confidential and not easy to obtain.

[0005] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0006] The rainfall similarity analysis method based on the rainfall color map according to the present application comprises the following steps:

[0007] S1, synthesis of the rainfall color map of the area to be analyzed;

[0008] A screen with a resolution of 640x480 is divided into 307200 grid points intersecting in the XY direction, the geographical position of the rainfall station to be analyzed is converted into grid coordinates, and according to the rainfall data of the rainfall station, 16 different colors are used to assign rainfall values to each grid point to form a rainfall color map; the rainfall color map contains geographical location information of the area to be analyzed, and at the same time, the rainfall at each geographical location is marked using colors.

[0009] S2, color block segmentation is performed on the rain amount color patch map;

[0010] The pixel intervals and weights of brown, grass color, green, blue, pink, and red are set, the threshold method is used for color block segmentation of the rain amount color patch map, and a binary mask feature map of each color block is obtained. The segmented color block contains the geographical location covered by the color block and the rainfall amount at the geographical location covered by the color block. The binary mask feature map only retains the geographical location information covered by the color block by removing the color information of the color block.

[0011] S3, using the mask method, the historical rain amount color patch map is divided into a plurality of historical rain amount blocks corresponding to the color blocks according to the binary mask feature map of the color block;

[0012] The historical rain amount color patch map is segmented by the geographical location information in the binary mask feature map to obtain a historical rain amount block corresponding to the geographical location information covered by the color block. The color on the historical rain amount block represents the historical rainfall amount of the region corresponding to the color block in the current rain amount color patch map to be analyzed.

[0013] S4, the color features of the historical rain amount block are extracted and saved separately;

[0014] S5, calculating the rain amount similarity;

[0015] S5.1, using a similarity calculation method to calculate the similarity between the color features of each historical rain amount block and the color features of the corresponding color block;

[0016] S5.2, according to the weight of the historical rain amount block corresponding to the color block, the weight is assigned to the similarity, and the product of the weight and the similarity is the weight similarity of the historical rain amount block;

[0017] S5.3, the weight similarities of the plurality of historical rain amount blocks segmented from the historical rain amount color patch map are added to obtain the final similarity of the historical rain amount color patch map;

[0018] S5.4, selecting the historical rain amount color patch map with the maximum final similarity as the reference of the rain amount color patch map;

[0019] Further, in S1, the screen is divided into 0 to 639 points in the X direction and 0 to 479 points in the Y direction.

[0020] Further, in S1, the rain amount value is assigned using the Kriging or inverse distance weighting method.

[0021] Further, in S2, the binary mask feature map only includes the geographical location information of the color block.

[0022] Further, the color feature in the step S4 is a color histogram and a color moment.

[0023] Further, the similarity calculation method in the step S5 is an Euclidean distance.

[0024] Further, the method is also used for calculating the rainfall field similarity, specifically: synthesizing the rainfall color patch map of the area to be analyzed according to hours or days; and calculating the similarity of the color patch in the historical rainfall color patch map and the color patch in the rainfall color patch map of the area to be analyzed according to the same time interval, and giving the similarity a weight according to the order of the time interval, and calculating the rainfall field similarity by weighting.

[0025] The advantage of the present application is that it is the first time to use the rainfall color patch data to analyze the rainfall similarity, solves the problem of the secrecy of the traditional rainfall numerical data, and effectively combines the rainfall and the rainfall spatial information by using the color and the spatial information in the rainfall color patch data as the retrieval features. And the rainfall trend of the historical rainfall is used to assist in judging the current rainfall trend, and provides data support for the flood monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 is the method flowchart of the present application.

[0027] Figure 2 is the rainfall color patch map to be analyzed in the method embodiment 2 of the present application.

[0028] Figure 3 is the color patch segmentation of the rainfall color patch map to be analyzed in the method embodiment 2 of the present application.

[0029] Figure 4 is the binary mask feature map of Figure 3 .

[0030] Figure 5 is the historical rainfall color patch map segmentation result in the embodiment 2 of the present application.

[0031] Figure 6 is the corresponding diagram of the color patch of the rainfall color patch map to be analyzed and the historical rainfall color patch in the embodiment 2 of the present application. DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the present application will be described clearly and completely below. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0033] Embodiment 1: The rain similarity analysis method based on the rain color map of the present application is described in detail, and the flowchart is shown in Figure 1 The specific steps include the following:

[0034] S1, the current rain color map of the area to be analyzed is synthesized;

[0035] The screen with a resolution of 640x480 is divided into 307200 grid points intersected in the XY direction, where the X direction is divided into 0 to 639 points, the Y direction is divided into 0 to 479 points, and there are a total of 307200 intersection points. The geographical position of each rain gauge station is converted into grid coordinates, and plotted on the screen to form a rain control point. Then 16 different colors are used to interpolate the rainfall of each grid point to form a rain color map. The rainfall of each grid point is determined by the Kriging or inverse distance weighting method according to the rainfall data of the measured rain gauge station. The rainfall is divided into 10mm, 25mm, 50mm, 100mm, 150mm, 200mm, 250mm, 300mm, etc. 16 grades, each grade corresponds to a color. The rain color map formed at this time contains not only the geographical position information of the area to be analyzed, but also uses color to mark the rainfall at each geographical position.

[0036] S2, color block segmentation is performed on the current rain color map of the area to be analyzed;

[0037] The pixel interval and weight of brown, green, green, blue, pink, and red are set, such as pink weight 5, blue weight 4, green weight 3, green weight 2, brown weight 1, etc. In fact, in practical application, the color commonly used in meteorological forecasting is usually used to represent the size of rainfall, and the color weight representing the rainfall in the main rain area is the largest, and the weight gradually decreases with the decrease of rainfall.

[0038] Then, the threshold method is used to segment the rain color map into brown, green, green, blue, pink, and red color blocks, each color block contains only one color, and different color blocks represent different rainfall. Then the binary mask feature map of each color block is obtained. Since the current rain color map of the area to be analyzed synthesized in step S1 contains the geographical position information of the area to be analyzed, and the rainfall at each geographical position is marked using color. Therefore, after color block segmentation of the rain color map, each color block contains the geographical position it covers and the rainfall at the geographical position it covers. The binary mask feature map only retains the geographical position information covered by the color block, and the color information of the color block is removed.

[0039] S3, using a mask method, the historical rainfall color patch map of the area to be analyzed is divided into several historical rainfall blocks corresponding to the color blocks according to the binary mask feature map of the color block in sequence; this step is to divide the historical rainfall color patch map of the area to be analyzed according to the geographical location information of the binary mask feature map of the color block, and obtain the historical rainfall block corresponding to the geographical location information covered by the color block. The color on the block represents the historical rainfall in the historical rainfall color patch map corresponding to the area of the color block in the current rainfall color patch map of the area to be analyzed.

[0040] S4, extract and save the color features of each historical rainfall block separately; the color features can be color histogram and color moment, etc., which are used for subsequent calculation of the similarity between historical rainfall and the geographical location rainfall covered by the color block to be analyzed.

[0041] S5, calculate the rainfall similarity; specifically including the following steps:

[0042] S5.1, calculate the similarity between the color features of each historical rainfall block and the color features of the corresponding color block using the Euclidean distance similarity calculation method;

[0043] S5.2, assign weights to the similarity according to the weights of the color blocks corresponding to the historical rainfall block, and the product of the weight and the similarity is the weight similarity of the historical rainfall block;

[0044] S5.3, add the weight similarities of the several historical rainfall blocks divided from the historical rainfall color patch map to obtain the final similarity between the historical rainfall color patch map of the area to be analyzed and the current rainfall color patch map of the area to be analyzed;

[0045] S5.4, select the historical rainfall color patch map with the maximum final similarity as the reference of the rainfall color patch map;

[0046] The above method can also be used to calculate the rainfall event similarity, specifically: synthesize the rainfall color patch map of a certain rainfall event in the area to be analyzed by hour or by day; calculate the similarity between the historical rainfall block in the historical rainfall color patch map and the color block in the rainfall color patch map of the area to be analyzed in the same time interval in sequence, and assign weights to the similarity according to the order of the time interval to calculate the similarity of the rainfall event. For example, the rainfall color patch map of the first hour or the first day of a certain rainfall event in the area to be analyzed is calculated according to steps S2 to S5 to calculate the similarity between the historical rainfall block in the historical rainfall color patch map of the first hour or the first day of a certain rainfall event and the color block in the rainfall color patch map of the first hour or the first day of a certain rainfall event in the area to be analyzed; then the rainfall color patch map of the second hour or the second day of a certain rainfall event in the area to be analyzed is calculated according to steps S2 to S5 to calculate the similarity between the historical rainfall block in the historical rainfall color patch map of the second hour or the second day of a certain rainfall event and the color block in the rainfall color patch map of the second hour or the second day of a certain rainfall event in the area to be analyzed; and so on.

[0047] Then, the similarity weight is given according to the order of time interval, such as the rain map similarity weight value of the first hour or the first day is 2, the rain map similarity weight of the second hour or the second day is 1.5, the rain map similarity weight of the third hour or the third day is 1, and the weight of the following is 1, etc. The similarity of the time interval of the historical rainfall is weighted and summed up, and the similarity of the current rainfall to be analyzed and the current rainfall to be analyzed can be obtained.

[0048] Embodiment 2: For the rain color spot map to be analyzed as shown in Figure 2 , the rain similarity analysis and application process using the method described in the application are as follows:

[0049] The current rain color spot map as shown in Figure 2 is divided into brown, green, green, blue, pink and red color blocks, and six color block maps of various shapes as shown in Figure 2 are extracted from Figure 3 . Each color block map includes only one color. At this time, the color in the color block map represents the rainfall, and the shape of the color block map contains the geographical location information covered by the rainfall to be analyzed.

[0050] Next, the binary mask feature map of the above color block map is obtained as shown in Figure 4 , and as can be seen in Figure 4 , the color information is removed from the binary mask feature map of the six color block maps, and only the shape of the color block map is retained, that is, only the geographical location information covered by the color block map is included.

[0051] Then, the binary mask feature map of the color block map is taken in turn, and the historical rain color spot map is divided into a plurality of historical rain blocks corresponding to the color blocks, as shown in Figure 5 . As can be seen in Figure 5 , the shape of the historical rain block corresponds to the color block, indicating that the geographical location covered by the historical rain block is completely consistent with the color block to which it corresponds, and the historical rain block includes more than one color, that is, the rainfall of the historical rain block and the color block to which it corresponds is different.

[0052] Then, the Euclidean distance similarity calculation method can be used to calculate the rainfall similarity of the historical rain block and the color block to which it corresponds. Specifically:

[0053] The color features of the historical rain blocks divided from the historical rain color spot map are extracted in turn, and are represented by color histogram and color moment, etc. Then, according to the Euclidean distance method, the similarity of the color features of each historical rain block and the color features of the corresponding color block is calculated in turn, as shown in Figure 6 ; and as can be seen in Figure 6 .The left side of the middle is the color block segmented from the rain color patch to be analyzed, and the right side is the historical rain patch corresponding to the color block on the left side segmented from the historical rain color patch. The similarity between each historical rain patch and the corresponding color block is calculated in turn by the Euclidean distance method.

[0054] According to the established weight, such as pink weight 5, blue weight 4, green weight 3, grass color weight 2, brown weight 1, etc., or the main rain area weight 5, the secondary rain area 4, etc., the weight similarity of each historical rain patch is calculated.

[0055] Weight similarity = weight * the rainfall similarity of the historical rain patch and the corresponding color block.

[0056] After obtaining the weight similarity of the historical rain patch segmented from the historical rain color patch, all the data are added, that is, the final similarity of the historical rain color patch and the rain color patch to be analyzed.

[0057] In the flood season, the trend of rain will directly affect the development of flood, and the accurate prediction of the development trend of rainfall will help the flood control supervisors to make timely flood control deployment. The rain color patch, as a kind of data representing the distribution of regional rainfall, directly and intuitively represents the rainfall information in the form of spatial color, and the visual information contained is simple and effective. The similarity of rainfall at two time points can be evaluated by comparing the color distribution of the same region in different rain patches.

[0058] The method described in the application is usually repeated in use to perform the above process, analyze multiple historical rain color patches in the database, obtain the final similarity of each historical rain color patch and the rain color patch to be analyzed, and take the historical rain color patch with the highest final similarity as the reference. The flood control supervisors can judge the change trend of the current rainfall according to the subsequent change trend of the most similar historical rain color patch, and provide rainfall prediction information for flood control work.

Claims

1. A method for rainfall similarity analysis based on rainfall color patch maps, characterized in that: Includes the following steps: S1, Rainfall color patch map synthesis for the area to be analyzed; The screen with a resolution of 640×480 was divided into 307,200 grid points intersecting in the X and Y directions. The geographical locations of the measured rain gauge stations in the area to be analyzed were converted into grid point coordinates. Based on the rainfall data from the measured rain gauge stations, 16 different colors were used to assign rainfall values ​​to each grid point, forming a rainfall color map. The rainfall color map not only includes the geographical location information of the area to be analyzed, but also uses colors to mark the rainfall at each geographical location. S2, perform color block segmentation on the rainfall color patch map; We set pixel ranges and weights for brown, grass, green, blue, pink, and red, and used a thresholding method to segment the rainfall color patch map into color blocks, obtaining a binary mask feature map for each color block. Each color block contains both the geographical location it covers and the rainfall at that location. The binary mask feature map of the color block only retains the geographical location information it covers. S3, using the masking method, the historical rainfall color patch map is divided into several historical rainfall map blocks that correspond one-to-one with the color blocks according to the binary mask feature map of the color blocks. S4, extract and save the color features of the historical rainfall tiles separately; S5, calculate rainfall similarity; S5.1, Use the similarity calculation method to calculate the similarity between the color features of each historical rainfall map patch and the color features of its corresponding color patch; S5.2, assign weights to the similarity based on the weights of the color blocks corresponding to the historical rainfall map tiles, and the product of the weights and the similarity is the weighted similarity of the historical rainfall map tiles; S5.3, add up the weighted similarity of several historical rainfall patches segmented from the historical rainfall color map to obtain the final similarity of the historical rainfall color map; S5.4 Select the historical rainfall color patch map with the highest final similarity as the reference for the rainfall color patch map.

2. The rainfall similarity analysis method based on rainfall color patch map according to claim 1, characterized in that: In step S1, the screen is divided into 0 to 639 points in the X direction and 0 to 479 points in the Y direction.

3. The rainfall similarity analysis method based on rainfall color patch map according to claim 1, characterized in that: The rainfall value assigned in step S1 is obtained using the Kriging or inverse distance weighting method.

4. The rainfall similarity analysis method based on rainfall color patch map according to claim 1, characterized in that: The color features mentioned in step S4 are the color histogram and color moments.

5. The rainfall similarity analysis method based on rainfall color patch map according to claim 1, characterized in that: The similarity calculation method described in step S5 is Euclidean distance.

6. The rainfall similarity analysis method based on rainfall color patch map according to any one of claims 1-5, characterized in that: It is also used to calculate the similarity of rainfall events, specifically: synthesizing a rainfall color map of the area to be analyzed by hour or day; calculating the similarity between the historical rainfall patches in the historical rainfall color map and the color patches in the rainfall color map of the area to be analyzed according to the same time interval, and assigning the weight of the similarity according to the order of the time interval, and calculating the similarity of rainfall events by weight.

Images