A method for optimizing the boundary position of the WRF model calculation area to minimize the boundary field forecast error

Abstract

The present invention discloses a method for optimizing the boundary position of a WRF model calculation region that minimizes boundary field forecast errors. The method comprises determining a forecast period and a global climate model; setting an optimization spatial range for the boundary position of the WRF model calculation region; analyzing the forecast errors of the global climate model for geopotential heights on different isobaric surfaces; and traversing the optimization spatial range to determine the optimal boundary position of the WRF model. Advantages include: overcoming the high subjectivity of conventional WRF model calculation region boundary position determination methods. By analyzing the forecast errors of the global climate model, a quantitative optimization method is used to find the boundary position of the calculation region that minimizes the global climate model boundary field forecast error. This method is quantitative, objective, and simple, and can provide a standardized process for the construction of the WRF model, facilitating the subsequent maintenance and promotion of the WRF model.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological forecasting, and in particular to a WRF model calculation area boundary position optimization method for minimizing boundary field forecast errors. Background Art

[0002] Numerical weather prediction (NWF) involves solving the fluid dynamics and thermodynamics equations that describe weather evolution based on atmospheric conditions, using large computers to perform numerical calculations under certain initial and boundary conditions. The WRF (Weather Research and Forecasting Model) is a unified mesoscale weather forecast model developed jointly by the National Center for Environmental Prediction (NCEP), the National Center for Atmospheric Research (NCAR), and several universities, research institutes, and business units. It is one of the most advanced NWF models worldwide and has a wide range of applications.

[0003] The WRF model does not directly generate meteorological data. Instead, it dynamically downscales the forecast boundary fields provided by the Global Climate Model (GCM) within a user-defined computational domain. This allows for higher-resolution, more refined, and more customized forecasts than the original GCM. The WRF model's computational domain is a rectangle bounded by two lines of longitude and two lines of latitude. Within this rectangle, WRF performs atmospheric numerical computations, while outside it, the GCM provides the meteorological boundary fields. Therefore, determining the boundary of the WRF model computational domain—that is, the coupling boundary between the WRF model and the GCM—is a key step in achieving accurate numerical weather forecasts. Currently, most methods for determining the boundary of the WRF model computational domain follow an empirical principle: placing the WRF model boundary in an area with relatively flat terrain and a relatively uniform vegetation type, encompassing the entire desired forecast area. It is generally believed that the GCM's forecast errors in these areas are relatively small, which can reduce the forecast errors transferred from the GCM to the WRF model, thereby improving the WRF model's meteorological forecast accuracy.

[0004] However, the aforementioned method for determining the WRF model boundary location is heavily subjective, relying primarily on manual judgment of the terrain's relief and the complexity of vegetation types. On the one hand, different modelers may select wildly different WRF boundary locations based on their subjective experience, hindering model standardization, maintenance, and widespread application. On the other hand, this empirical boundary location method cannot guarantee minimal forecast errors in the global climate model boundary field, hindering efforts to reduce the boundary error between WRF and GCMs and achieve accurate numerical weather forecasts. This shortcoming poses challenges to the application of numerical weather forecasts based on the WRF model. Summary of the Invention

[0005] The object of the present invention is to provide a method for optimizing the boundary position of a WRF model calculation area that minimizes the boundary field forecast error, thereby solving the above-mentioned problems existing in the prior art.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A WRF model calculation area boundary position optimization method for minimizing boundary field forecast errors includes the following steps:

[0008] S1. Set the optimization space range of the boundary position of the calculation area of ​​the WRF mode:

[0009] Based on the calculation area of ​​the WRF model, the minimum calculation area boundary of the WRF model is determined; the two sides of the minimum calculation area boundary in the longitude direction and the two sides in the latitude direction are expanded respectively, and the expanded calculation area is used as the optimization spatial range of the calculation area boundary position of the WRF model;

[0010] S2. Analyze the forecast errors of geopotential height on different isobaric surfaces by global climate models:

[0011] Based on the month of the forecast start date, within the optimization space, obtain the daily forecast geopotential height data of the global climate model for different isobaric surfaces within the forecast period and the daily geopotential height data of ERA5 for the corresponding isobaric surfaces in the same period. After uniformly interpolating the daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5 to the same spatial grid resolution, calculate the root mean square error of the potential height of each grid point within the optimization space.

[0012] S3. Traverse the optimization space to determine the optimal boundary position of the WRF model:

[0013] By expanding the grid points in the longitude and latitude directions of the minimum calculation area of ​​the WRF model, all possible boundary positions of the calculation area within the optimization space are traversed. The root mean square error of the potential height of each grid point within the optimization space is calculated based on the average root mean square error of all grid points at each boundary position, and the optimal boundary position of the WRF model calculation area is determined based on the calculation results.

[0014] Preferably, step S1 is specifically as follows: the calculation area of ​​the WRF model is a rectangle surrounded by two longitude lines and two latitude lines, which includes the entire forecast area; the minimum calculation area boundary of the WRF model can be determined according to the calculation area of ​​the WRF model; the longitudinal length of at least one minimum calculation area is expanded on both sides of the longitude direction of the minimum calculation area boundary; the latitudinal length of at least one minimum calculation area is expanded on both sides of the latitude direction of the minimum calculation area boundary; and the expanded calculation area is used as the optimal spatial range for the calculation area boundary position of the WRF model.

[0015] Preferably, step S2 specifically includes the following contents:

[0016] S21. For the month M in which the forecast start date falls, based on the optimal spatial range of the WRF calculation region boundary position, download the forecast results of the global climate model for the month M of multiple historical years for each start date of the month, extract the daily forecast geopotential height data of the global climate model for different isobaric surfaces within the forecast period; download the daily geopotential height data of ERA5 for the corresponding isobaric surfaces during the same historical period, and use it as the true value of the geopotential height to calculate the forecast error of the global climate model;

[0017] S22. Use bilinear interpolation to uniformly interpolate the daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5 to the same spatial grid resolution;

[0018] S23, calculating the root mean square error of the geopotential height of each grid point within the optimization space based on the interpolated daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5;

[0019]

[0020] Among them, RMSE (i,j) is the root mean square error of the potential height of each grid point; A is the optimization space range; (i, j) is each grid point in the optimization space range; is the geopotential height forecast result of the i, j grid point and the k isobaric surface on the tth reporting date and the tth day; is the geopotential height of the i-th and j-th grid points and the k-th isobaric surface provided by ERA5 on the same date; t0 represents each day within the forecast period; T is the length of the forecast period; t is each reporting start date within the M-th month; N is the number of reporting start dates; k is each isobaric surface; S is the number of isobaric surfaces.

[0021] Preferably, step S3 is specifically as follows: within the optimization space range of the calculation area boundary position of the WRF mode, with the minimum calculation area as the starting area, by expanding the grid points in the longitude and latitude directions, traversing all possible calculation area boundary positions, and calculating the average root mean square error of all grid points on each boundary position based on the root mean square error of the potential height of each grid point in the optimization space range, until the maximum optimization space range is reached, the calculation area boundary position with the smallest average root mean square error is the optimal boundary position of the WRF mode calculation area;

[0022]

[0023] Among them, RMSE m is the average value of the root mean square error of each grid point; B is the set of grid points at the boundary of the current calculation area; ng is the number of grid points.

[0024] Preferably, before step S1, the method further includes: S0, determining the forecast period and the global climate model: setting the forecast period length of the numerical precipitation forecast according to the forecast requirements; and selecting the global climate model that drives the WRF model according to the forecast requirements and data availability.

[0025] Preferably, the length of the forecast period is between 1-7 days, inclusive.

[0026] The beneficial effects of the present invention are as follows: 1. The present invention overcomes the problem of high subjectivity in the traditional WRF model calculation area boundary determination method. By analyzing the forecast error of the global climate model, the present invention adopts a quantitative optimization method to find the calculation area boundary position that minimizes the global climate model boundary field forecast error. It has the characteristics of being quantitative, objective, and simple, and can provide a standardized process for the construction of the WRF model. At the same time, it is easy to maintain and promote the application of the WRF model in the later stage. 2. By quantitatively optimizing the boundary position of the WRF model calculation area, the present invention can minimize the forecast error of the boundary field provided by the global climate model, thereby reducing the forecast error transmitted by the global climate model to the WRF model, and ultimately improving the numerical meteorological forecast accuracy of the WRF model. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 1 is a flow chart of a method according to an embodiment of the present invention;

[0028] Figure 23 is a spatial distribution diagram of the root mean square error of the geopotential height field of the 200hPa, 500hPa and 850hPa isobaric surfaces in an embodiment of the present invention (where the black box is the optimal boundary position of the calculation area). DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0030] Example 1

[0031] like Figure 1 As shown, in this embodiment, the method for determining the boundary position of the current WRF model calculation area has a strong subjective experience, which is not conducive to the WRF model to carry out accurate numerical weather forecasting. The present invention analyzes the historical forecast errors of the potential height of the global climate model on different isobaric surfaces, takes the minimum forecast error as the optimization goal, and optimizes the boundary position of the WRF model calculation area in a quantitative optimization manner, thereby reducing the forecast error transmitted to WRF by the global climate model, thereby improving the weather forecast effect of the WRF model. Finally, a WRF model calculation area boundary position optimization method that minimizes the boundary field forecast error is proposed. This method optimizes the WRF model calculation area boundary position by evaluating the forecast error of the global climate model boundary field, which can be used to determine the optimal position of the WRF model calculation area, thereby improving the weather forecast accuracy of the WRF model. The method includes the following four parts:

[0032] I. Determining the forecast period and global climate model

[0033] 1. Set the forecast period length (number of days) of the numerical precipitation forecast according to the forecast requirements, between 1 and 7 days, including 1 day and 7 days.

[0034] 2. Select the global climate model that drives the WRF model based on forecast requirements and data availability.

[0035] 2. Setting the optimal spatial range for the boundary position of the WRF model calculation area

[0036] Based on the calculation area of ​​the WRF model, the minimum calculation area boundary of the WRF model is determined; the two sides of the minimum calculation area boundary in the longitude direction and the two sides in the latitude direction are expanded respectively, and the expanded calculation area is used as the optimal spatial range for the calculation area boundary position of the WRF model.

[0037] Specifically, the WRF model calculation region is a rectangle bounded by two longitude lines and two latitude lines, encompassing the entire forecast area. This rectangle determines the minimum WRF model calculation region boundary. Based on this boundary, the minimum calculation region is expanded by at least one longitudinal length on either side of the minimum calculation region boundary, and by at least one latitudinal length on either side of the minimum calculation region boundary. The expanded calculation region serves as the optimal spatial range for the WRF model calculation region boundary location.

[0038] 3. Analysis of the forecast errors of the global climate model on geopotential heights at different isobaric surfaces

[0039] Based on the month of the forecast start date, within the optimization space, obtain the daily forecast geopotential height data (Geopotential height) of the global climate model for different isobaric surfaces within the forecast period and the daily geopotential height data of the fifth generation of the European Centre for Medium-Range Weather Forecasts (ERA5) for the corresponding isobaric surfaces in the same period. After uniformly interpolating the daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5 to the same spatial grid resolution, calculate the root mean square error of the geopotential height of each grid point within the optimization space. The specific steps include the following:

[0040] 1. For the month M where the forecast start date falls, based on the optimal spatial range of the WRF calculation area boundary position, download the forecast results of the global climate model for the Mth month of multiple years in history for each start date of the month, extract the daily forecast geopotential height data of the global climate model for different isobaric surfaces within the forecast period; download the daily geopotential height data of the ERA5 for the corresponding isobaric surface during the same period in history, and use it as the true value of the geopotential height to calculate the forecast error of the global climate model;

[0041] 2. Using the bilinear interpolation method, the daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5 are uniformly interpolated to the same spatial grid resolution;

[0042] 3. Based on the interpolated daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5, calculate the root mean square error of the geopotential height of each grid point within the optimization space;

[0043]

[0044] Among them, RMSE (i,j) is the root mean square error of the potential height of each grid point; A is the optimization space range; (i, j) is each grid point in the optimization space range; is the geopotential height forecast result of the i, j grid point and the k isobaric surface on the tth reporting date and the tth day; is the geopotential height of the i-th and j-th grid points and the k-th isobaric surface provided by ERA5 on the same date; t0 represents each day within the forecast period; T is the length of the forecast period; t is each reporting start date within the M-th month; N is the number of reporting start dates; k is each isobaric surface; S is the number of isobaric surfaces.

[0045] 4. Determining the optimal boundary location of the WRF model

[0046] By expanding the grid points in the longitude and latitude directions of the minimum calculation area of ​​the WRF model, all possible boundary positions of the calculation area within the optimization space are traversed. The root mean square error of the potential height of each grid point within the optimization space is calculated based on the average root mean square error of all grid points at each boundary position, and the optimal boundary position of the WRF model calculation area is determined based on the calculation results.

[0047] Specifically, within the optimization space of the WRF model's calculation area boundary position, starting with the minimum calculation area, the grid points are expanded in the longitude and latitude directions to traverse all possible calculation area boundary positions. The root mean square error of the potential height of each grid point within the optimization space is used to calculate the average root mean square error of all grid points at each boundary position until the maximum optimization space range is reached. When the average root mean square error takes the minimum value, the calculation area boundary position is the optimal boundary position of the WRF model calculation area.

[0048]

[0049] Among them, RMSE m is the average value of the root mean square error of each grid point; B is the set of grid points at the boundary of the current calculation area; ng is the number of grid points.

[0050] Example 2

[0051] like Figure 2 As shown, in this embodiment, the upper reaches of the Jinsha River, including the Shigu River, is used as the forecast area, March 1, 2023 is used as the forecast start date, and the forecast period is 3 days. Daily weather forecasts are conducted for this area. Based on this, the method of the present invention is described.

[0052] I. Determining the forecast period and global climate model

[0053] 1. According to the forecast requirements, the forecast period of numerical precipitation forecast is set to 3 days.

[0054] 2. Based on forecast requirements and data availability, the Global Weather Service (GFS) global climate model, published by the National Centers for Environmental Prediction (NCEP), was selected to drive the WRF model. The GFS model provides 6-hourly weather forecast boundaries for a global 1°×1° grid for the next 16 days and is a commonly used global climate model for driving the WRF model for numerical weather forecasts.

[0055] 2. Setting the optimal spatial range for the boundary position of the WRF model calculation area

[0056] The WRF model's calculation region is a rectangle bounded by two lines of longitude and two lines of latitude, encompassing the entire forecast area. The longitude and latitude range of the upper Jinsha River basin is 91°–104°E and 22°–36°N, representing the model's minimum calculation region boundary. Based on this boundary, the model's minimum calculation region is expanded by a length of 13 degrees along each longitudinal direction of the minimum calculation region boundary, extending the longitude range to 78°–117°. The model's minimum calculation region is expanded by a length of 14 degrees along each latitudinal direction of the minimum calculation region boundary, extending the latitude range to 8°–50°. This expanded region serves as the optimal spatial range for the WRF calculation region boundary location.

[0057] 3. Analysis of the forecast errors of the global climate model on geopotential heights at different isobaric surfaces

[0058] (1) The forecast start date is March 1st, and the month it falls on is March. Based on this, the forecast results of the GFS model for March each year from 2015 to 2022 are downloaded. The spatial range of the download is 78° to 117° east longitude and 8° to 50° north latitude. The download address is https: / / www.ncei.noaa.gov / products / weather-climate-models / global-forecast.

[0059] The GFS model publishes four forecasts daily. For geopotential height forecast error analysis, the forecast results released daily at 00:00 Universal Time (UTC) were selected. Three-day forecast results were downloaded for 31 forecast start dates, from March 1 to March 31, 2015 to 2022. Forecast geopotential height data for 200hPa, 500hPa, and 800hPa isobaric surfaces for the next one, two, and three days were extracted from the forecast result files for each of the 31 March forecast start dates, 2015 to 2022. In addition, daily geopotential height data for 200hPa, 500hPa, and 800hPa isobaric surfaces from the ERA5 reanalysis data from March 1 to March 31, 2015 to 2022, were downloaded from https: / / cds.climate.copernicus.eu / cdsapp#! / dataset / reanalysis-era5-pressure-levels.

[0060] (2) The bilinear interpolation method is used to interpolate the daily forecast geopotential height data of the GFS model with a grid resolution of 1° to the 0.25° grid resolution of the ERA reanalysis data.

[0061] (3) Based on the geopotential height provided by ERA5 and the predicted geopotential height of the GFS model, the root mean square error (RMSE) of the predicted geopotential height at each 0.25 degree grid point within the optimization space is calculated:

[0062]

[0063] Among them, A is the optimization space range of 78 degrees to 117 degrees east longitude and 8 degrees to 50 degrees north latitude; (i, j) are the grid points in the optimization space range; is the geopotential height forecast result of the i, j grid point and the k isobaric surface on the tth reporting date and the tth day; is the geopotential height of the i-th and j-th grid points and the k-th isobaric surface provided by ERA5 on the same date; t0 is the day of the forecast period; T = 3 is the length of the forecast period (number of days); t represents the reporting start date in March; N = 248 (= 8 years × 31 days) represents the total number of reporting start dates from 2015 to March 2022; k represents the isobaric surface (200, 500, 850 hPa); S = 3 represents the number of isobaric surfaces.

[0064] 4. Traverse the optimization space to determine the optimal boundary position of the WRF model

[0065] In the optimization space of 78° to 117° east longitude and 8° to 50° north latitude, the minimum calculation area (91° to 104° east longitude and 22° to 36° north latitude) determined in step 3 is used as the starting area. Each time, a 0.25-degree grid point is expanded in the longitude and latitude directions, and all possible boundary positions of the calculation area are traversed. The average root mean square error (RMSE) of all grid points at each boundary position is calculated. m , until the maximum optimization space range is reached. Among them, RMSE m The calculation formula is:

[0066]

[0067] Where b is the grid point set at the boundary of the current calculation area; ng = 26208 is the number of grid points.

[0068] Based on the above traversal method, the average root mean square error RMSE of the potential height at different boundary positions is calculated. m At 132~1403m 2 / s 2 This indicates that the average prediction error of geopotential height at different boundary locations still has large spatial variability.

[0069] When RMSE m Reached a minimum of 132m 2 / s 2 At this time, the boundaries of the calculation area are located at 76 degrees east longitude, 107.5 degrees east longitude, 19.25 degrees north latitude, and 39.75 degrees north latitude, respectively. This calculation area boundary is the optimal boundary position of the WRF model calculation area. The WRF model is constructed based on this calculation area boundary. The forecast error of the GFS model boundary field is minimized, which supports accurate weather forecasts.

[0070] By adopting the above technical solution disclosed in the present invention, the following beneficial effects are obtained:

[0071] The present invention provides a WRF model calculation area boundary location optimization method that minimizes the boundary field forecast error. This method overcomes the problem of high subjectivity in the traditional WRF model calculation area boundary location determination method. By analyzing the forecast error of the global climate model, a quantitative optimization method is used to find the calculation area boundary location that minimizes the global climate model boundary field forecast error. This method has the characteristics of being quantitative, objective, and simple. It can provide a standardized process for the construction of the WRF model and facilitate the later maintenance and promotion of the WRF model. By quantitatively optimizing the boundary location of the WRF model calculation area, the present invention can minimize the forecast error of the boundary field provided by the global climate model, thereby reducing the forecast error transmitted by the global climate model to the WRF model, and ultimately improving the numerical meteorological forecast accuracy of the WRF model.

[0072] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for optimizing the boundary position of a WRF model calculation region to minimize boundary field forecast errors, characterized by: The following steps are included: S1. Set the optimization space range of the boundary position of the calculation area of ​​the WRF mode: Based on the calculation area of ​​the WRF model, the minimum calculation area boundary of the WRF model is determined; the two sides of the minimum calculation area boundary in the longitude direction and the two sides in the latitude direction are expanded respectively, and the expanded calculation area is used as the optimization spatial range of the calculation area boundary position of the WRF model; S2. Analyze the forecast errors of geopotential height on different isobaric surfaces by global climate models: Based on the month of the forecast start date, within the optimization space, obtain the daily forecast geopotential height data of the global climate model for different isobaric surfaces within the forecast period and the daily geopotential height data of ERA5 for the corresponding isobaric surfaces in the same period. After uniformly interpolating the daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5 to the same spatial grid resolution, calculate the root mean square error of the potential height of each grid point within the optimization space. Step S2 specifically includes the following contents: S21. For the month M in which the forecast start date falls, based on the optimal spatial range of the WRF calculation region boundary position, download the forecast results of the global climate model for the month M of multiple historical years for each start date of the month, extract the daily forecast geopotential height data of the global climate model for different isobaric surfaces within the forecast period; download the daily geopotential height data of ERA5 for the corresponding isobaric surfaces during the same historical period, and use it as the true value of the geopotential height to calculate the forecast error of the global climate model; S22. Use bilinear interpolation to uniformly interpolate the daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5 to the same spatial grid resolution; S23, calculating the root mean square error of the geopotential height of each grid point within the optimization space based on the interpolated daily forecast geopotential height data of the global climate model and the daily geopotential height data of ERA5; Among them, RMSE (i,j) is the root mean square error of the potential height of each grid point; A is the optimization space range; (i, j) is each grid point in the optimization space range; is the geopotential height forecast result of the i, j grid point and the k isobaric surface on the tth reporting date and the tth day; is the geopotential height of the i-th and j-th grid points and the k-th isobaric surface provided by ERA5 on the same date; t0 represents each day within the forecast period; T is the length of the forecast period; t is each reporting start date within the M-th month; N is the number of reporting start dates; k is each isobaric surface; S is the number of isobaric surfaces; S3. Traverse the optimization space to determine the optimal boundary position of the WRF model: By expanding the grid points in the longitude and latitude directions of the minimum calculation area of ​​the WRF model, all possible boundary positions of the calculation area within the optimization space are traversed. The root mean square error of the potential height of each grid point within the optimization space is calculated based on the average root mean square error of all grid points at each boundary position, and the optimal boundary position of the WRF model calculation area is determined based on the calculation results.

2. The WRF model calculation region boundary position optimization method for minimizing boundary field forecast error according to claim 1 is characterized by: Specifically, step S1 is as follows: the calculation area of ​​the WRF model is a rectangle surrounded by two longitude lines and two latitude lines, which includes the entire forecast area. According to the calculation area of ​​the WRF model, the minimum calculation area boundary of the WRF model can be determined, and the longitudinal length of at least one minimum calculation area is expanded on both sides of the longitude direction of the minimum calculation area boundary, and the latitudinal length of at least one minimum calculation area is expanded on both sides of the latitude direction of the minimum calculation area boundary, and the expanded calculation area is used as the optimal spatial range for the calculation area boundary position of the WRF model.

3. The WRF model calculation region boundary position optimization method for minimizing boundary field forecast error according to claim 1 is characterized by: Specifically, step S3 is as follows: within the optimization space range of the WRF model calculation area boundary position, with the minimum calculation area as the starting area, all possible calculation area boundary positions are traversed by expanding the grid points in the longitude and latitude directions, and the root mean square error of all grid points at each boundary position is calculated based on the root mean square error of the potential height of each grid point in the optimization space range until the maximum optimization space range is reached. The calculation area boundary position with the smallest root mean square error average is the optimal boundary position of the WRF model calculation area; Among them, RMSE m is the average value of the root mean square error of each grid point; B is the set of grid points at the boundary of the current calculation area; ng is the number of grid points.

4. The WRF model calculation region boundary position optimization method for minimizing boundary field forecast error according to any one of claims 1 to 3, characterized in that: Before step S1, the method further includes: S0, determining the forecast period and the global climate model: setting the forecast period length of the numerical precipitation forecast according to the forecast requirements; and selecting the global climate model driving the WRF model according to the forecast requirements and data availability.

5. The WRF model calculation region boundary position optimization method for minimizing boundary field forecast error according to claim 4 is characterized in that: The length of the forecast period is between 1 and 7 days, inclusive.

Images