Precipitation forecast error correction optimization method and system based on complex terrain

Abstract

The invention discloses a precipitation forecast error correction optimization method and system based on a complex terrain, and solves the technical problem of low accuracy of precipitation forecast in the complex terrain in the prior art. Comprising the following steps: acquiring meteorological observation data and numerical mode forecast data of a target area; interpolating the grid point rainfall forecast data of the numerical mode to an observation station through a bilinear interpolation method, and classifying according to a preset rainfall grade division threshold value; calculating the forecast accuracy score of each mode at different time, different rainfall levels and different altitude partitions based on a TS scoring formula; adopting a multi-mode rainfall grading TS scoring weight integration method to correct a forecast result; further optimizing a forecast result by adopting a terrain height correction method; and outputting a corrected rainfall forecast result, and evaluating a correction effect based on the TS score. According to the invention, through refined classification and multi-dimensional evaluation, refined diagnosis and correction of forecast errors are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of climate prediction, and particularly relates to a precipitation forecast error correction and optimization method and system based on complex terrain. Background Art

[0002] With global climate change, the frequency and intensity of weather events are increasing continuously. With the influence of global climate change, the characteristics of "warming and humidification" of the climate are gradually emerging, which has a profound impact on regional water resources, ecological systems and agricultural production. Accurate precipitation forecasts can help relevant departments take timely response measures to reduce losses caused by natural disasters. In recent years, with the development of numerical weather prediction models, the accuracy of precipitation forecasts has been significantly improved, but there are still certain limitations, especially in complex terrain areas. Therefore, correcting precipitation forecast models and improving the accuracy of forecast and warning have become important topics in the meteorological community.

[0003] Currently, the development trend of precipitation forecast models mainly focuses on improving the resolution of the models, improving the parameterization of physical processes, and developing multi-model ensemble forecasting techniques. High-resolution models can better capture the details of terrain and atmospheric processes, thereby improving the accuracy of forecasts. In addition, multi-model ensemble forecasting techniques improve the stability and reliability of forecasts by synthesizing the forecast results of multiple models. The development of these technologies provides new directions for improving precipitation forecast models. Numerical forecasting is the core of quantitative precipitation prediction, but due to the chaotic nature of the system, errors in the internal dynamics and physical processes of the model, and errors in the initial field, errors inevitably exist in the forecasts. Since precipitation is a non-linear physical process, its prediction difficulty is more significant than that of other meteorological elements. At the same time, with the continuous upgrading and optimization of numerical models, the characteristics of their forecast errors are also evolving. Therefore, it is particularly important to conduct subjective and objective evaluations of the precipitation forecasts directly output by the models, which helps to make targeted corrections based on the evaluation results and thus effectively improve the accuracy of precipitation forecasts.

[0004] Kizilsu Kirgiz Autonomous Prefecture is located on the northwestern border of China. Due to its unique geographical location and complex climate characteristics, the accuracy of summer precipitation forecasts is crucial for local agricultural production, water resource management and disaster prevention and mitigation work. However, due to the complexity of the terrain and climate system, precipitation forecasting in Kizilsu Kirgiz Autonomous Prefecture has always been a challenge. The present invention aims to evaluate the accuracy of summer precipitation simulations in Kizilsu Kirgiz Autonomous Prefecture and compare the performance of two precipitation models, namely the Regional Medium-Scale Numerical Forecast Product of the National Meteorological Center (CMA_MESO) and the Global Weather Forecast Model of the National Meteorological Center (CMA_GFS). Summary of the Invention

[0005] The purpose of the present invention is to provide a precipitation forecast error correction and optimization method and system based on complex terrain, so as to solve the technical problem of low accuracy of precipitation forecasting in complex terrain in the prior art.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] In the first aspect, the present invention provides a precipitation forecast error correction and optimization method based on complex terrain, including the following steps:

[0008] Step 1: Obtain meteorological observation data and numerical model forecast data of the target area, and the numerical model at least includes the CMA_MESO regional model and the CMA_GFS global model;

[0009] Step 2: Interpolate the grid precipitation forecast data of the numerical model to the observation stations by the bilinear interpolation method, and classify them according to the threshold values divided by the preset precipitation levels;

[0010] Step 3: Calculate the forecast accuracy scores of each model at different times, different precipitation levels, and different altitude zones based on the TS scoring formula;

[0011] Step 4: Use the multi-model precipitation grading TS scoring weight integration method to correct the forecast results;

[0012] Step 5: Further optimize the forecast results by using the terrain height correction method, specifically, by analyzing the difference between the station altitude and the true terrain height at each grid point position, the precipitation forecast field is corrected;

[0013] Step 6: Output the corrected precipitation forecast results, and evaluate the correction effect based on the TS score.

[0014] Further, in the step 3, the classification criteria for different times, different precipitation levels, and different altitude zones are as follows:

[0015] The research time is from June to August, and it is limited to the 24h, 48h, and 72h forecasts of the precipitation amount from 20:00 to 20:00;

[0016] The precipitation levels are divided into four levels: light rain: 0.1mm / d, moderate rain: 6mm / d, heavy rain: 12mm / d, and rainstorm: 24mm / d;

[0017] The target area is divided into mountainous areas: altitude ≥ 2500m, shallow mountainous areas: 1400m ≤ altitude ≤ 2500m, and plain areas: altitude ≤ 1400m according to altitude;

[0018] Through the above classification criteria, the 24h, 48h, and 72h precipitation inspection and evaluation and forecast correction are carried out for each area and different precipitation levels.

[0019] Further, in step 3, the TS scoring formula is as follows:

[0020]

[0021] In the formula, hits represents the number of stations where both the forecast and the observation occur, False represents the number of stations where the forecast occurs but the observation does not occur, and misses represents the number of stations where the forecast does not occur but the observation occurs.

[0022] Further, the multi - model precipitation classification TS scoring weight integration method includes the following steps:

[0023] For each precipitation level i, calculate the original precipitation forecast values of the CMA_MESO regional model and the CMA_GFS global model respectively;

[0024] According to the TS scores of each model, weight their forecast values;

[0025] Fuse the forecast results of the two models to generate a more accurate revised precipitation forecast value, and improve the quality of the overall forecast by means of weight allocation. The specific formula is as follows:

[0026]

[0027] In the formula, tp_meso(i) represents the original precipitation forecast value of the CMA_MESO model for a certain precipitation level i; tp_gfs(i) represents the original precipitation forecast value of the CMA_GFS model for a certain precipitation level i; TS_meso(i) represents the TS score of the CMA_MESO model for a certain precipitation level i; TS_gfs(i) represents the TS score of the CMA_GFS model for a certain precipitation level i; cal_tp(i) represents the revised precipitation forecast value.

[0028] Further, the terrain height correction method corrects the forecasts of the CMA_MESO model and the CMA_GFS model by analyzing the influence of terrain height on precipitation distribution, and calculates using the following formula:

[0029] adj_tp(j) = tp(j) * exp[b * (sta_h(j) - fcst_h(j))];

[0030] In the formula, j represents the index of the observation station at the correction time step; b is the terrain correction parameter, which is a constant used to control the influence degree of terrain height change on precipitation adjustment; tp(j) represents the original forecast precipitation value; adj_tp(j) represents the corrected forecast precipitation value. This formula is used to adjust the forecast precipitation tp(j) according to the terrain height difference between the observation station sta_h(j) and the forecast station fcst_h(j).

[0031] In a second aspect, the present invention provides a precipitation forecast error correction and optimization system based on complex terrain, including:

[0032] A data acquisition module for acquiring meteorological observation data and numerical model forecast data;

[0033] An interpolation and classification module for implementing data interpolation and precipitation level division;

[0034] A TS score calculation module for generating TS scores of each model in different regions;

[0035] A correction and optimization module, including sub-module one and sub-module two, which are respectively used to perform multi-model precipitation classification TS score weight integration correction and perform terrain height difference adjustment;

[0036] A result output and evaluation module for outputting the correction result and generating an evaluation report.

[0037] Furthermore, the system further includes a visualization module for displaying the comparison of TS scores of precipitation forecasts before and after correction and the terrain height distribution map.

[0038] In a third aspect, the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory. When the processor executes the program, it implements the steps of the precipitation forecast error correction and optimization method based on complex terrain.

[0039] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps of the precipitation forecast error correction and optimization method based on complex terrain.

[0040] Based on the above technical solutions, the embodiments of the present invention can at least produce the following technical effects:

[0041] An optimization method for precipitation forecast error correction based on complex terrain provided by the present invention corrects the forecast results through the application of a multi-model precipitation hierarchical TS score weight integration method and a terrain height correction method. By comparing and analyzing two precipitation models, CMA_MESO and CMA_GFS, their respective advantages and disadvantages are found, and a more effective correction method is explored to improve the accuracy of summer precipitation forecasts in Kizilsu Kirgiz Autonomous Prefecture. By analyzing the characteristics of different model products through the present invention, it is expected to better utilize the forecast results of numerical models, improve the forecasters' ability to control numerical models, and ultimately achieve the effect of improving the forecast accuracy rate, which can provide a scientific basis and technical support for precipitation forecasts in Kizilsu Kirgiz Autonomous Prefecture and similar regions, and contribute to improving the service quality and efficiency of meteorological forecasts. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on the structures shown in these drawings.

[0043] Figure 1 It is a distribution map of the terrain and meteorological stations in Kizilsu Kirgiz Autonomous Prefecture according to an embodiment of the present invention;

[0044] Figure 2 It is a comparison chart of TS scores of 24-hour precipitation forecasts in Kizilsu Kirgiz Autonomous Prefecture under different models according to an embodiment of the present invention;

[0045] Figure 3 It is a comparison chart of TS scores of 48-hour precipitation forecasts in Kizilsu Kirgiz Autonomous Prefecture under different models according to an embodiment of the present invention;

[0046] Figure 4 It is a comparison chart of TS scores of 72-hour precipitation forecasts in Kizilsu Kirgiz Autonomous Prefecture under different models according to an embodiment of the present invention.

[0047] In the figure, a1, a2, and a3 respectively represent the plain area, the shallow mountain area, and the mountain area. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The technical solutions in the embodiments of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in this field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0049] The present invention uses the summer precipitation observation data and model forecast data of Kizilsu Kirgiz Autonomous Prefecture. The details are as follows:

[0050] (1) Station observation data: The observation data comes from the China Meteorological Administration. The precipitation data of 296 automatic stations in Kezhou are selected as the actual precipitation and observation station altitude data for calculating the model precipitation deviation and the precipitation terrain correction method; the daily 24:00 cumulative precipitation (00UTC) observation data from June to August 2022-2023 (the observation abnormal stations have been removed as the actual situation, and the data has been quality controlled and has a certain quality assurance) are selected to test the correction effect. The station distribution is as follows: Figure 1 shown.

[0051] (2) Model data: The data used in the present invention are precipitation forecast products of the CMA regional model CMA_MESO and the CMA global model CMA_GFS and the corresponding precipitation real-time data. CMA-MESO and CMA_GFS are two meteorological numerical forecast models used by the China Meteorological Administration. They differ in resolution, coverage, horizontal resolution, number of vertical layers, model top height and forecast range during 2022-2023. The CMA-MESO model has a higher resolution, usually 0.1° to 0.09°, which is suitable for China and surrounding areas, while the CMA_GFS model has a higher resolution, usually 0.09°, which is suitable for weather forecasts worldwide. Both models use the ARAKAWA-C grid, with multiple layers of resolution in the vertical direction, and the model top height is set at about 30km to include most of the activities in the troposphere and stratosphere. The forecast range covers most of China and other Asian regions, providing support for tasks such as weather forecasting, climate research, disaster warning and climate monitoring. In practical applications, these models will be combined with other numerical models and forecasting tools, as well as the experience of forecasters, to provide the most accurate weather forecasts. The forecast range captured in this report is (70-82°E, 34.125-46.125°N).

[0052] The precipitation forecast error correction and optimization method based on complex terrain in the present invention selects the precipitation forecast data of CMA_MESO and CMA_GFS and the terrain height forecast data to calculate the model precipitation deviation; selects the 24-hour cumulative precipitation (00 UTC) forecast products output by CMA_MESO and CMA_GFS from June to August 2022 - 2023 as the basic model products for the 24 - 72h time limit 24h precipitation correction and evaluation research. For the resolution of the above data, the horizontal resolution of CMA_MESO is 3 km; for CMA_GFS, the horizontal resolution is 25 km in 2022 and 12.5 km in 2023; when evaluating, the grid data of the model is uniformly interpolated to the ground observation stations by the bilinear interpolation method, which specifically includes the following steps:

[0053] Step 1: Obtain the meteorological observation data and numerical model forecast data of the target area, and the numerical model includes at least the CMA_MESO regional model and the CMA_GFS global model.

[0054] Step 2: Interpolate the grid precipitation forecast data of the numerical model to the observation stations by the bilinear interpolation method and classify them according to the preset precipitation level division threshold.

[0055] Bilinear Interpolation is a technique for interpolation in a two-dimensional space, usually applied in fields such as image processing and computer graphics. Its basic idea is to estimate the value of the target point using the weighted average of the surrounding four known points. Among them, calculating the weighted average is based on the distance between the target point and the adjacent points to calculate a value to obtain the final value of the target point.

[0056] Step 3: Calculate the forecast accuracy scores of each model at different times, different precipitation levels, and different altitude zones based on the TS scoring formula.

[0057] The research time spans from June to August 2022 to 2023, and is limited to the 24h, 48h, and 72h forecasts of the precipitation from 20:00 to 20:00.

[0058] The precipitation levels are divided into four levels: light rain: 0.1 mm / d, moderate rain: 6 mm / d, heavy rain: 12 mm / d, and rainstorm: 24 mm / d.

[0059] Kezhou is divided into three regions according to altitude: plain area, shallow mountain area, and mountain area. The altitude ≥ 2500 m is the mountain area, the interval of 1400 m ≤ altitude ≤ 2500 m is the shallow mountain area, and the altitude ≤ 1400 m is the plain area.

[0060] Through the above classification criteria, conduct 24h, 48h, and 72h precipitation inspection and evaluation and forecast correction for each region and different precipitation levels.

[0061] The TS scoring formula is as follows:

[0062]

[0063] In the formula, hits represents the number of stations where both the forecast and the actual situation occur, False represents the number of stations where the forecast occurs but the actual situation does not occur, and misses represents the number of stations where the forecast does not occur but the actual situation occurs.

[0064] Step 4: Use the multi-model precipitation classification TS scoring weight integration method to correct the forecast results, including the following steps:

[0065] For each precipitation level i, calculate the original precipitation forecast values of the CMA_MESO regional model and the CMA_GFS global model respectively;

[0066] According to the TS scores of each model, weight their forecast values;

[0067] Fuse the forecast results of the two models to generate a more accurate corrected precipitation forecast value, and improve the quality of the overall forecast through weight allocation. The specific formula is as follows:

[0068]

[0069] In the formula, tp_meso(i) represents the original precipitation forecast value of the CMA_MESO model for a certain precipitation level i; tp_gfs(i) represents the original precipitation forecast value of the CMA_GFS model for a certain precipitation level i; TS_meso(i) represents the TS score of the CMA_MESO model for a certain precipitation level i; TS_gfs(i) represents the TS score of the CMA_GFS model for a certain precipitation level i; cal_tp(i) represents the corrected precipitation forecast value.

[0070] The advantage of this method is that it can integrate the respective advantages of the two models and optimize the forecast results through statistical means. In practical applications, especially in areas lacking high-performance computing resources, the multi-model precipitation classification TS scoring weight integration method is an efficient and practical precipitation forecast correction tool. Through refined statistical processing, it can significantly improve the accuracy of precipitation forecasts.

[0071] Step 5: Use the terrain height correction method to further optimize the forecast results, specifically by analyzing the difference between the station elevation and the true terrain height at each grid point position to correct the precipitation forecast field.

[0072] Specifically, the terrain height correction method takes into account the influence of terrain on precipitation forecasting. By analyzing the impact of terrain height on precipitation distribution, this method corrects the forecasts of the CMA-MESO and CMA_GFS models. On this basis, this study adopted the terrain height correction method to eliminate and correct the influence of terrain errors. The terrain height correction method is a precipitation forecasting error correction method based on the difference between the actual terrain height and the forecast terrain height. By analyzing the difference between the "station elevation" at each grid point and the true terrain height, the precipitation forecast field is corrected. Through the application of the terrain height correction method, this study can better understand and predict the differences in precipitation indices in regions with different altitude gradients. The terrain height correction method can improve the accuracy of model forecasts and reduce forecast errors to a certain extent. This process is calculated using the following formula:

[0073] adj_tp(j) = tp(j) * exp[b * (sta_h(j) - fc5t_h(j))];

[0074] In the formula, j represents the index of the observation station at the correction time step; b is the terrain correction parameter, which is a constant used to control the degree of influence of terrain height changes on precipitation adjustment; tp(j) represents the original forecast precipitation value; adj_tp(j) represents the corrected forecast precipitation value. This formula is used to adjust the forecast precipitation amount tp(j) according to the terrain height difference between the observation station sta_h(j) and the forecast station fcst_h(j).

[0075] Step 6: Output the corrected precipitation forecast results and evaluate the correction effect based on the TS score.

[0076] In this invention, the TS score test is conducted on the precipitation forecasts of the CMA_MESO and CMA_GFS models. The results show that there are differences in the forecast performance of the two models under different precipitation thresholds. By adopting the multi-model precipitation classification TS score weight integration method and the terrain height correction method, the precipitation forecast errors of the CMA-MESO and CMA_GFS models are corrected. Finally, this study conducts an application evaluation on the corrected precipitation forecasts. By comparing the actual observed data and the corrected forecast data, the corrected forecast effect is evaluated.

[0077] In this embodiment, the TS scores of 24H, 48H, and 72H are compared:

[0078] 24H: As Figure 2As shown in the figure, the comparison of the TS scores of the CMA_MESO model and the CMA_GFS model before correction for predicting precipitation in Kezhou shows that the performance of CMA_MESO is better than that of the CMA_GFS model. For different regions, the mountainous area performs the best, followed by the shallow mountainous area, and the plain area is the last (a1 < a2 < a3). For different magnitudes, light rain performs the best, moderate rain is the second, heavy rain is worse, and heavy rainstorm is the worst, with the highest being only 0.01785 (the heavy rainstorm forecast for the plain area by the CMA_MESO model). After correction, the correction effect of CMA_GFS is better than that of CMA_MESO. Among them, the correction effect of the shallow mountainous area is the best, followed by the high mountainous area, and the plain area is the worst. Moderate rain and heavy rain are the best, and light rain and heavy rainstorm are the second. The best correction effect is the correction method using the terrain height correction method for heavy rain in the plain area, with a 23% increase.

[0079] 48H: The comparison of the TS scores of the CMA_MESO model and the CMA_GFS model before correction for predicting precipitation in Kezhou shows that the performance of CMA_MESO is better than that of the CMA_GFS model. For different regions, the mountainous area and the shallow mountainous area perform the best, and the plain area is the second. The highest is only 0.05803 (the heavy rainstorm forecast for the mountainous area by the CMA_MESO model). For different magnitudes, light rain performs the best, moderate rain is the second, heavy rain is worse, and heavy rainstorm is the worst, with the highest being only 0.13405. After correction, the correction effect of CMA_GFS is better than that of CMA_MESO. Among them, the correction effect of the mountainous area is the best, followed by the shallow mountainous area, and the plain area is the worst. Moderate rain and heavy rain are the best, and light rain and heavy rainstorm are the second. The best correction effect is the correction method using the terrain height correction method for heavy rainstorm in the mountainous area by the CMA_MESO model, with a 58.9% increase, and the correction method using the weight integration method for heavy rainstorm in the plain area by the CMA_MESO model, with a 41.6% increase.

[0080] 72H: The comparison of the TS scores of the CMA_MESO model and the CMA_GFS model before correction for predicting precipitation in Kezhou shows that in the forecast of light rain, the performance of CMA_GFS is better than that of the CMA_MESO model, while in the forecast of other magnitudes, the performance of CMA_MESO is better than that of the CMA_GFS model. For different regions, the mountainous area and the shallow mountainous area perform the best, and the plain area is the second. The highest is only 0.303 (the heavy rainstorm forecast for the plain area by the CMA_MESO model). For different magnitudes, light rain performs the best, moderate rain is the second, heavy rain is worse, and heavy rainstorm is the worst, with the highest being only 0.303 (the heavy rainstorm forecast for the plain area by the CMA_MESO model). After correction, the correction effect of CMA_GFS is better than that of CMA_MESO. Among them, the correction effect of the plain area is the best, followed by the mountainous area and the shallow mountainous area. Moderate rain and heavy rain are the best, and light rain and heavy rainstorm are the second. The best correction effect is the correction method using the weight integration method for heavy rainstorm in the mountainous area by the CMA_MESO model, with a 51.2% increase.

[0081] Evaluation of correction methods:

[0082] Comparing the terrain height correction method and the weight integration method, the weight integration method shows relatively stable performance. Evaluation of the performance of the weight integration method: The proportion of the correction effect being 0 - 10% is 50%, and the proportion of the correction effect being 10% - 50% is 50%. Evaluation of the performance of the terrain height correction method: The proportion of the correction effect being 0 - 10% is 81%, and the proportion of the correction effect being 10% - 50% is 12%.

[0083] The above has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of the present invention claimed is defined by the appended claims and their equivalents.

Claims

1. An optimized method for correcting precipitation forecast errors under complex terrain, characterized in that, It includes the following steps: Step 1: Obtain meteorological observation data and numerical model forecast data of the target area. The numerical models at least include the CMA_MESO regional model and the CMA_GFS global model; Step 2: Interpolate the grid precipitation forecast data of the numerical model to the observation stations by the bilinear interpolation method and classify them according to the preset precipitation level division thresholds; Step 3: Calculate the forecast accuracy scores of each model at different times, different precipitation levels, and different altitude zones based on the TS scoring formula; Step 4: Use the multi-model precipitation classification TS scoring weight integration method to correct the forecast results; Step 5: Further optimize the forecast results by using the terrain height correction method. Specifically, the precipitation forecast field is corrected by analyzing the difference between the station altitude and the true terrain height at each grid point position; Step 6: Output the corrected precipitation forecast results and evaluate the correction effect based on the TS score.

2. The precipitation forecast error correction and optimization method based on complex terrain according to claim 1, wherein In Step 3, the classification criteria for different times, different precipitation levels, and different altitude zones are as follows: The research time is from June to August, and the 24h, 48h, and 72h forecasts of the precipitation amount from 20:00 to 20:00 are limited; The precipitation levels are divided into four levels: light rain: 0.1mm / d, moderate rain: 6mm / d, heavy rain: 12mm / d, and rainstorm: 24mm / d; The target area is divided into mountainous areas by altitude: altitude ≥ 2500m, shallow mountainous areas: 1400m ≤ altitude ≤ 2500m, and plain areas: altitude ≤ 1400m; Through the above classification criteria, the 24h, 48h, and 72h precipitation inspection and evaluation and forecast correction are carried out for each area and different precipitation levels.

3. The precipitation forecast error correction and optimization method based on complex terrain according to claim 1, characterized in that In Step 3, the TS scoring formula is: In the formula, hits represents the number of stations where both the forecast and the actual situation occur, False represents the number of stations where the forecast occurs but the actual situation does not occur, and misses represents the number of stations where the forecast does not occur but the actual situation occurs.

4. The precipitation forecast error correction optimization method based on complex terrain according to claim 1, characterized in that The multi-model precipitation classification TS scoring weight integration method includes the following steps: For each precipitation level i, calculate the original precipitation forecast values of the CMA_MESO regional model and the CMA_GFS global model respectively; According to the TS scores of each model, weight their forecast values; Fuse the forecast results of the two models to generate a more accurate corrected precipitation forecast value, and improve the overall forecast quality through weight distribution. The specific formula is as follows: In the formula, i is the precipitation level index, tp_meso(i) represents the original precipitation forecast value of the CMA_MESO model; tp_gfs(i) represents the original precipitation forecast value of the CMA_GFS model for a certain precipitation level i; TS_meso(i) represents the TS score of the CMA_MESO model for a certain precipitation level i; TS_gfs(i) represents the TS score of the CMA_GFS model for a certain precipitation level i; cal_tp(i) represents the corrected precipitation forecast value.

5. The precipitation forecast error correction and optimization method based on complex terrain according to claim 1, characterized in that The terrain height correction method corrects the forecasts of the CMA_MESO model and the CMA_GFS model by analyzing the influence of terrain height on precipitation distribution, and is calculated using the following formula: adj_tp(j) = tp(j) * exp[b * (sta_h(j) - fcst_h(j))]; In the formula, j represents the index of the observation station at the correction time step; b is the terrain correction parameter, which is a constant and used to control the influence degree of the terrain height change on the precipitation adjustment; tp(j) represents the original forecast precipitation value; adj_tp(j) represents the corrected forecast precipitation value. This formula is used to adjust the forecast precipitation amount tp(j) according to the terrain height difference between the observation station sta_h(j) and the forecast station fcst_h(j).

6. A precipitation forecast error correction and optimization system based on complex terrain, characterized in that, It includes: A data acquisition module, which is used to acquire meteorological observation data and numerical model forecast data; An interpolation and classification module, which is used to implement data interpolation and precipitation grade division; A TS score calculation module, which is used to generate the TS scores of each model in different regions; A correction and optimization module, which includes sub-module one and sub-module two, and is respectively used to perform the integrated correction of the TS score weights of the multi-model precipitation classification and to perform the terrain height difference adjustment; A result output and evaluation module, which is used to output the correction result and generate an evaluation report.

7. The precipitation forecast error correction and optimization system based on complex terrain according to claim 6, characterized in that The system further includes a visualization module, which is used to display the comparison of the TS scores of the precipitation forecasts before and after correction and the terrain height distribution map.

8. An electronic device, comprising a memory, a processor, and a computer program stored on the memory, characterized in that When the processor executes the program, it implements the steps of the precipitation forecast error correction and optimization method based on complex terrain described in any one of claims 1-5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the precipitation forecast error correction and optimization method based on complex terrain described in any one of claims 1-5.

Images