Rainfall target identification and correction method based on adaptive threshold

Abstract

The invention relates to the field of weather inspection and correction, and provides a rainfall target identification and correction method based on an adaptive threshold value in order to improve the overall precision and reliability of forecast inspection, and the method comprises the steps: dynamically adjusting the threshold value according to the distribution of the adaptive threshold value, and capturing rainfall systems with different intensities and scales. Meanwhile, the area ratio is used as an index for objectively quantifying whether target object recognition is reasonable or not, after different target pairs are recognized, a forecasting field is corrected through error information of the target pairs, the intensity and spatial distribution of rainfall forecasting are optimized, the overall precision and reliability of forecasting inspection can be improved, and the forecasting efficiency is improved. And the forecasting model can adapt to the change of a complex weather system.

Description

Technical Field

[0001] The invention relates to the field of weather inspection and correction, and in particular to a precipitation target recognition and correction method based on an adaptive threshold. Background Art

[0002] MODE (Method for Object-Based Diagnostic Evaluation, MODE) is an effective means for evaluating the performance of meteorological forecasts. While paying attention to traditional skill scoring, it also pays more attention to the differences between forecasts and actual conditions at different spatial scales. The MODE method is based on defining and calculating different attributes of objects, setting weight coefficients for different attributes, and using fuzzy logic algorithms to calculate the total benefit function of forecast performance to judge the overall performance of the forecast. There are clear boundaries for object distinction and recognition. Object recognition uses a convolution operator to determine the execution object, making the original data field smoother and more continuous. By controlling the range of meteorological element thresholds, the inspection target is filtered and the focus point strong weather targets are screened. Object attribute precipitation objects are assigned a series of spatial attributes after being identified, and the spatial attribute parameters describing the target monomers and the attributes of the geometric and spatial relationship between any target monomer and any target monomer in the actual field are calculated. Once the object attributes are determined, the attribute diagnostic quantity of the object pair is used to calculate the similarity of any two objects (from different fields). The similarity calculation uses a fuzzy logic algorithm, and all object attributes are input into it to calculate the total benefit function of the forecast performance.

[0003] The MODE method relies on a set fixed threshold to identify objects. The observation field and model respond differently to the threshold in different weather systems, and there are significant differences in intensity and morphology. For precipitation systems with weak observations or weak model forecasts, a single threshold may not be able to correctly identify these objects, resulting in incomplete or missed pattern recognition, or too many false positive recognitions in strong patterns, causing object shape distortion and size deviation. Summary of the invention

[0004] In order to improve the overall accuracy and reliability of forecast verification, this application provides a precipitation target recognition and correction method based on adaptive threshold.

[0005] The technical solution adopted by the present invention to solve the above problems is:

[0006] The precipitation target recognition and correction method based on adaptive threshold includes:

[0007] Step 1, preprocessing the original field to obtain a spatial field with a grid, wherein the original field includes an observation field and a forecast field;

[0008] Step 2, select the study area in the spatial field and calculate the precipitation thresholds of the observation field and the forecast field respectively;

[0009] Step 3: Based on the calculated precipitation threshold, the precipitation values ​​of the observation field and the forecast field are filtered to obtain a mask field. Specifically, the precipitation value less than the precipitation threshold is marked as 0, and the precipitation value greater than or equal to the precipitation threshold is marked as 1;

[0010] Step 4: In the filtered observation field and forecast field, connected targets are extracted and merged using the connected domain algorithm:

[0011] Identify connected targets in the observation and forecast fields that are greater than their respective precipitation thresholds;

[0012] Determine the neighborhood range near_dis of the connected target and the neighborhood ratio near_rate between the connected targets according to the size of the connected target; if the neighborhood ratio near_rate is 0, the connected target is not merged; if the neighborhood ratio near_rate is greater than or equal to the neighborhood ratio threshold, the two connected targets are merged to form a new connected target; repeat the above steps until the neighborhood ratio near_rate between all connected targets is less than the neighborhood ratio threshold, and the connected target at this time is the precipitation target;

[0013] Step 5: Set the neighborhood range cover_dis and the matching degree cover_rate based on the precipitation target, and match the precipitation target in the observation field with the precipitation target in the forecast field based on the neighborhood range cover_dis and the matching degree cover_rate;

[0014] Step 6: Calculate the area ratio of the precipitation target in the forecast field according to the matching result. The area ratio is the number of grid points with a pickling rate of 1 divided by the area of ​​a rectangular frame formed by the long axis and the short axis of the precipitation target as the side length;

[0015] Step 7: If the area ratio is less than the set area ratio threshold, adjust the neighborhood range near_dis and the neighborhood ratio threshold and repeat steps 4-6 until the area ratio is greater than the area ratio threshold;

[0016] Step 8: Calculate the spatial attribute errors of the target pair according to the matching results, including centroid position, intensity and area errors;

[0017] Step 9: Optimize and adjust according to the spatial attribute error.

[0018] Furthermore, the step 1 specifically includes:

[0019] Step 11: Preprocess the observation and forecast fields, including filling in missing data and interpolating them into grids with the same resolution;

[0020] Step 12: Determine a smoothing coefficient, and perform filtering on the observed field and the forecast field based on the smoothing coefficient.

[0021] Furthermore, the calculation method of the precipitation threshold in step 2 is:

[0022]

[0023] In the formula, m is the precipitation threshold, x is the sorted precipitation dataset, n is the number of data in the dataset, p is the percentile, ranging from 0 to 1, [np] means taking the integer part of np, and x is ([np]+1) Indicates the precipitation amount after np.

[0024] Furthermore, for the mesoscale precipitation system, the neighborhood range near_dis is 30-100km.

[0025] Furthermore, the neighborhood range near_dis is set to 50km, and the neighborhood ratio threshold is set to 0.2.

[0026] Furthermore, the step 5 is specifically as follows:

[0027] Set the neighborhood range cover_dis and matching degree cover_rate based on the merged precipitation target.

[0028] Calculate the matching degree cover_rate between the precipitation targets to be matched. If the matching degree cover_rate is greater than the set value of the matching degree cover_rate, the match is successful.

[0029] Match Where N a and N b Respectively represent the number of grid points of the forecast field precipitation target and the observation field precipitation target, n a and n b It is the number of grid points where precipitation targets are within each other's neighborhood range cover_dis.

[0030] Furthermore, the neighborhood range cover_dis is set to 50km, and the matching degree cover_rate setting value is set to 0.7.

[0031] Furthermore, in step 7, the area ratio threshold is set to 0.5.

[0032] Furthermore, the step 9 specifically includes:

[0033] According to the longitude and latitude errors of the centroid position, the forecast field is shifted in longitude and / or latitude;

[0034] The forecast field is increased or decreased by the corresponding intensity error, and the grid points with negative adjusted precipitation intensity are set to zero;

[0035] The area error is determined as the scaling factor, and the forecast field is spatially stretched or reduced based on the scaling factor to obtain the corrected precipitation field.

[0036] Compared with the prior art, the present invention has the following beneficial effects: more accurate spatial verification of the forecast field, dynamically adjusting the threshold according to the distribution of the adaptive threshold, capturing precipitation systems of different intensities and scales; at the same time, using the area ratio as an objective quantitative indicator of whether the target object identification is reasonable, after identifying different target pairs, correcting the forecast field through the error information of the target pair, and optimizing the intensity and spatial distribution of the precipitation forecast. The method of the present application can not only improve the overall accuracy and reliability of the forecast verification, but also enable the forecast model to adapt to the changes of complex weather systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of the precipitation target recognition and correction method based on adaptive threshold;

[0038] Figure 2 Schematic diagram of different percentile precipitation thresholds for different observation and forecast fields;

[0039] Figure 3 This is a schematic diagram of the neighborhood range;

[0040] Figure 4 Schematic diagram of a heavy rainfall event in the slope area of ​​Sichuan Province. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. 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.

[0042] Take a heavy rainfall event in the slope area of ​​Sichuan Province as an example. Figure 1 As shown in FIG. 1 , the precipitation target recognition and correction method based on the adaptive threshold includes:

[0043] Step 1: Preprocess the original field f(x, y) to obtain a gridded spatial field C(x, y), wherein the original field includes an observation field and a forecast field.

[0044] Preprocessing includes filling the missing data of the observation field and the forecast field and interpolating them into the grid with the same resolution; determining the smoothing coefficient and filtering the observation field and the forecast field based on the smoothing coefficient. The smoothing coefficient can be selected according to the actual process, for example, the smoothing coefficient is 4 times the number of grid points, which is not limited here.

[0045] Step 2: Select the study area in the spatial field C(x,y) and calculate the precipitation thresholds of the observation field and the forecast field respectively. The calculation method of the precipitation threshold is:

[0046]

[0047] In the formula, m is the precipitation threshold, x is the sorted precipitation dataset, n is the number of data in the dataset, and p is the percentile, ranging from 0 to 1. The larger the p value, the more weak precipitation values ​​are filtered out; the identification target is generally heavy precipitation, and the p value can generally be 75%. [np] means taking the integer part of np, x ([np]+1) Indicates the precipitation amount after np. The precipitation threshold is adjusted dynamically according to the precipitation amount, and the captured precipitation system is more in line with actual needs.

[0048] Assume that the sorted precipitation data set is x = [1, 2, 3, 4, 5]. If p is 0.8, n×p = 4, then m = 0.5×(x (4) +x (5) )=0.5×(4+5)=4.5;If p is 0.7, n×p=3.5, then m=x (4) =4. Figure 2 The threshold results for different percentiles in different cases are given.

[0049] Step 3: Based on the calculated precipitation threshold, the precipitation values ​​of the observation field and the forecast field are filtered to obtain the mask field M(X,Y). Specifically, precipitation values ​​less than the precipitation threshold are marked as 0, and precipitation values ​​greater than or equal to the precipitation threshold are marked as 1.

[0050] If connected targets are directly extracted and merged on the mask field, some precipitation information of the original field will be lost. In order to reduce information loss, the original observation field and the original forecast field can be assigned values ​​respectively through the obtained mask field to obtain the reconstructed field F(x,y)=M(x,y)f(x,y), and connected targets can be extracted and merged on the reconstructed field.

[0051] Step 4: In the filtered observation field and forecast field, connected targets are extracted and merged using the connected domain algorithm:

[0052] Identify connected targets in the observation and forecast fields that are greater than their respective precipitation thresholds;

[0053] According to the size of the connected target, the neighborhood range near_dis of the connected target and the neighborhood ratio near_rate between the connected targets are determined; the neighborhood range near_dis refers to the range formed by the outer contour of the connected target extending outward by a distance D1, such as Figure 3 Dotted line shows the area ratio N aand N b Respectively represent the number of grid points of the forecast field precipitation target and the observation field precipitation target, n a and n b It is the number of grid points where precipitation targets are within each other's neighborhood range near_dis.

[0054] If the neighborhood ratio near_rate is 0, the connected target is not merged; if the neighborhood ratio near_rate is greater than or equal to the neighborhood ratio threshold, the two connected targets are merged to form a new connected target; repeat the above steps until the neighborhood ratio near_rate between all connected targets is less than the neighborhood ratio threshold, and the connected target at this time is the precipitation target.

[0055] The specific values ​​of the distance D1 and the neighborhood ratio threshold can be set according to actual needs. In this embodiment, the distance D1 is set to 50km and the neighborhood ratio threshold is set to 0.2. According to previous statistics, the distance D1 for medium-scale targets (20 to 200km) can be set to 30-100km; the smaller the neighborhood ratio threshold, the smaller the merging probability. Through continuous debugging, the selection is made according to the effect.

[0056] Step 5: Set the neighborhood range cover_dis and the matching degree cover_rate, and match the precipitation targets in the observation field with the precipitation targets in the forecast field based on the neighborhood range cover_dis and the matching degree cover_rate.

[0057] Based on the merged precipitation target, the neighborhood range cover_dis and the matching degree cover_rate setting value are set. The neighborhood range cover_dis is the distance deviation allowed when calculating the overlap of the observation and forecast targets, that is, the range formed by extending the outer contour of the target outward by a distance of D2. The target in this study is a mesoscale, D2 is set to 50km, and the matching degree cover_rate setting value is selected as 0.7. It is generally believed that a match is successful when it is above 0.7. It can also be set to other values ​​according to actual needs;

[0058] Calculate the matching degree cover_rate between the precipitation targets to be matched. If the matching degree cover_rate is greater than the set value of the matching degree cover_rate, the match is successful.

[0059] Match Where N a and N b Respectively represent the number of grid points of the forecast field precipitation target and the observation field precipitation target, n a and n b It is the number of grid points where precipitation targets are within each other's neighborhood range cover_dis.

[0060] Step 6: Calculate the area ratio of the precipitation target in the forecast field based on the matching result. The area ratio is the number of grid points with a pickling rate of 1 divided by the area of ​​a rectangular frame with the major axis and minor axis of the precipitation target as the side lengths.

[0061] Step 7: If the area ratio is less than the set area ratio threshold, adjust the neighborhood range near_dis and the neighborhood ratio threshold and repeat steps 4-6 until the area ratio is greater than the area ratio threshold to complete the precipitation target identification. In this embodiment, the area ratio threshold is set to 0.5, which is considered to be a good way to identify the target. It can also be set to other values, which are not limited here.

[0062] Traditional recognition methods often rely on subjective judgment, while this application provides a more quantitative evaluation standard through area ratio, which can effectively measure whether the object of pattern recognition is consistent with actual observations, more objectively evaluate the forecast performance, and avoid the deviation caused by relying solely on subjective judgment.

[0063] Step 8: Calculate the spatial attribute errors of the target pair based on the matching results, including centroid position, intensity and area errors, where the area error is the ratio of the precipitation target area between the forecast field and the observation field;

[0064] Step 9: Optimize and adjust according to spatial attribute errors:

[0065] According to the longitude and latitude errors of the centroid position, the forecast field is shifted in longitude and / or latitude;

[0066] The forecast field is increased or decreased by the corresponding intensity error, and the grid points with negative adjusted precipitation intensity are set to zero;

[0067] The area error is determined as the scaling factor, and the forecast field is spatially stretched or reduced based on the scaling factor to obtain the corrected precipitation field.

[0068] Through the above all-round correction of precipitation fields, the consistency between the forecast field and the actual observation field in terms of spatial position, precipitation intensity and coverage can be significantly improved. The corrected results more accurately reflect the actual precipitation situation, especially in heavy precipitation events, and can effectively reduce the problems of position deviation, intensity mismatch and too large or too small area in the forecast. This method is not only suitable for the precise correction of individual precipitation events, but can also be extended to the same type of heavy precipitation events, systematically improving the accuracy and reliability of the forecast, and providing more valuable reference forecast products for disaster warning.

[0069] Figure 4 Schematic diagram of a heavy rainfall event in the slope area of ​​Sichuan Province: (a) is the observation field, (b) is the FY4B satellite precipitation product, and (c) is the corrected precipitation field, unit: millimeter (mm).

Claims

1. A precipitation target recognition and correction method based on adaptive threshold, characterized in that: include: Step 1, preprocessing the original field to obtain a spatial field with a grid, wherein the original field includes an observation field and a forecast field; Step 2, select the study area in the spatial field and calculate the precipitation thresholds of the observation field and the forecast field respectively; Step 3: Based on the calculated precipitation threshold, the precipitation values ​​of the observation field and the forecast field are filtered to obtain a mask field. Specifically, the precipitation value less than the precipitation threshold is marked as 0, and the precipitation value greater than or equal to the precipitation threshold is marked as 1; Step 4: In the filtered observation field and forecast field, connected targets are extracted and merged using the connected domain algorithm: Identify connected targets in the observation and forecast fields that are greater than their respective precipitation thresholds; Determine the neighborhood range near_dis of the connected target and the neighborhood ratio near_rate between the connected targets according to the size of the connected target; if the neighborhood ratio near_rate is 0, the connected target is not merged; if the neighborhood ratio near_rate is greater than or equal to the neighborhood ratio threshold, the two connected targets are merged to form a new connected target; repeat the above steps until the neighborhood ratio near_rate between all connected targets is less than the neighborhood ratio threshold, and the connected target at this time is the precipitation target; Step 5: Set the neighborhood range cover_dis and the matching degree cover_rate based on the precipitation target, and match the precipitation target in the observation field with the precipitation target in the forecast field based on the neighborhood range cover_dis and the matching degree cover_rate; Step 6: Calculate the area ratio of the precipitation target in the forecast field according to the matching result. The area ratio is the number of grid points with a pickling rate of 1 divided by the area of ​​a rectangular frame formed by the long axis and the short axis of the precipitation target as the side length; Step 7: If the area ratio is less than the set area ratio threshold, adjust the neighborhood range near_dis and the neighborhood ratio threshold and repeat steps 4-6 until the area ratio is greater than the area ratio threshold; Step 8: Calculate the spatial attribute errors of the target pair according to the matching results, including centroid position, intensity and area errors; Step 9: Optimize and adjust according to the spatial attribute error.

2. The precipitation target recognition and correction method based on adaptive threshold according to claim 1 is characterized in that: The step 1 specifically includes: Step 11: Preprocess the observation and forecast fields, including filling in missing data and interpolating them into grids with the same resolution; Step 12: Determine a smoothing coefficient, and perform filtering on the observed field and the forecast field based on the smoothing coefficient.

3. The precipitation target recognition and correction method based on adaptive threshold according to claim 1 is characterized in that: The calculation method of the precipitation threshold in step 2 is: In the formula, m is the precipitation threshold, x is the sorted precipitation dataset, n is the number of data in the dataset, p is the percentile, ranging from 0 to 1, [np] means taking the integer part of np, and x ([np]+1) Indicates the precipitation amount after np.

4. The precipitation target recognition and correction method based on adaptive threshold according to claim 1 is characterized in that: For mesoscale precipitation systems, the neighborhood range near_dis is 30-100km.

5. The precipitation target recognition and correction method based on adaptive threshold according to claim 4 is characterized in that: The neighborhood range near_dis is 50km, and the neighborhood ratio threshold is 0.

2.

6. The precipitation target recognition and correction method based on adaptive threshold according to claim 1 is characterized in that: The step 5 is specifically as follows: Set the neighborhood range cover_dis and matching degree cover_rate based on the merged precipitation target. Calculate the matching degree cover_rate between the precipitation targets to be matched. If the matching degree cover_rate is greater than the set value of the matching degree cover_rate, the match is successful. Match Where N a and N b Respectively represent the number of grid points of the forecast field precipitation target and the observation field precipitation target, n a and n b It is the number of grid points where precipitation targets are within each other's neighborhood range cover_dis.

7. The precipitation target recognition and correction method based on adaptive threshold according to claim 6 is characterized in that: The neighborhood range cover_dis is set to 50km, and the matching degree cover_rate setting value is set to 0.

7.

8. The precipitation target recognition and correction method based on adaptive threshold according to claim 1 is characterized in that: In step 7, the area ratio threshold is set to 0.

5.

9. The precipitation target recognition and correction method based on adaptive threshold according to any one of claims 1 to 8, characterized in that: The step 9 specifically includes: According to the longitude and latitude errors of the centroid position, the forecast field is shifted in longitude and / or latitude; The forecast field is increased or decreased by the corresponding intensity error, and the grid points with negative adjusted precipitation intensity are set to zero; The area error is determined as the scaling factor, and the forecast field is spatially stretched or reduced based on the scaling factor to obtain the corrected precipitation field.

Images