High-resolution mode topographic data construction method

Abstract

The invention discloses a high-resolution mode topographic data construction method, and the principle of the method is as follows: collecting elevation data based on an SRTM-DEM, and carrying out the hole filling of the collected elevation data through employing a moving surface fitting interpolation technology. According to the high-resolution mode terrain data construction method, a small and medium-scale mountain system which is difficult to depict by traditional terrain data is depicted, and implementation / copying / reproduction of a more practical and finer mode underlying surface terrain in the mode is facilitated.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of meteorological and environmental model application, and particularly relates to a high-resolution model terrain data construction method. BACKGROUND

[0002] Due to limitations in model resolution and the like, it is currently difficult to make accurate predictions on the occurrence location and extreme value intensity of extreme disasters. Model resolution includes the resolution (horizontal resolution and vertical resolution) of external input data required for driving the model, and the resolution (horizontal resolution) of terrain data of the model itself, and the application focuses on the scientific problem of how to improve the horizontal resolution of model terrain data.

[0003] Numerical models are advanced tools for simulating and predicting disaster processes, and are the main means and basis for weather forecasting. They include meteorological models (for rainstorm, gale, and heat wave) and environmental models (for haze and dust). The rapid development of models began in the 1990s, when computer technology was on the rise and people began to use various models to simulate and study disaster processes. First, meteorological models developed rapidly, such as MM5 (the fifth-generation NCAR / Penn State Meso-scale Model) and ARPS (the Advanced Regional Prediction System). After 2000, the United States introduced and continuously developed the latest version of the meteorological model WRF (the Weather Research and Forecasting model). Subsequently, scholars at home and abroad began to develop environmental models based on meteorological models. WRF-Chem (WRF coupled to Chemistry) was improved during this period (2002-2009). In 2001, China began to develop the meteorological model GRAPES (Global / Regional Assimilation and Prediction Enhanced System), which is a numerical weather prediction system developed independently by China. It has been successfully put into operation in China's meteorological department. So far, foreign countries have developed WRF, WRF-Chem, CMAQ (the Community Multiscale Air Quality system), and other models that can simulate and predict global or regional meteorological / environmental disasters and provide complete and diverse high-resolution aerosol data products (PM2.5, extinction coefficient, etc.) and meteorological data products (temperature, pressure, wind, humidity, etc.). Chinese scholars have also developed GRAPES, NAQPMS (The Nested Air Quality Prediction Modeling System), GRAPES-CUACE (Global / Regional Assimilation and Prediction System coupled to the CMA Unified Atmospheric Chemistry Environment), and other model systems.

[0004] Since the first generation of MM5 model in the 1970s, numerical models have been continuously developed for half a century, and have become the international advanced and main weather forecasting tool, which can realize more accurate simulation and prediction of global / regional disaster weather. However, when encountering extreme disaster weather, the prediction accuracy of numerical model is not ideal. Taking the extreme rainstorm disaster in 2021 as an example, the model prediction of this rainstorm is outstanding, and the situation field predicted 3-5 days in advance well reflects the characteristics that the low vortex system will be generated in the saddle-shaped low pressure area between the two high pressures in the east and west, and predicts that there will be continuous rainstorm in the west of Huanghuai and the south of North China for two or three days. However, the difficulty lies in that the model is difficult to accurately predict the rainstorm area and the extreme rainfall intensity (such as the extreme intensity of 1 hour rainfall exceeding 200 mm).

[0005] Extreme rainfall belongs to the category of mesoscale meteorology, so the model needs to have super-fine simulation and prediction ability, otherwise the prediction will not be accurate. Therefore, to solve this bottleneck, key technologies of numerical model need to be overcome, including developing new cloud physics scheme of fine model, constructing high-resolution model terrain data, and overcoming other key technologies. SUMMARY

[0006] The purpose of the present application is to provide a high-resolution model terrain data construction method, which can depict the mesoscale mountain range that cannot be depicted by traditional terrain data, and help to realize / reproduce / close to actual and more fine model underlying surface terrain in the model.

[0007] In order to achieve the above purpose, the technical scheme adopted by the present application is: a high-resolution model terrain data construction method, based on SRTM-DEM to collect elevation data, and using mobile curved surface fitting interpolation technology to fill the holes of the collected elevation data.

[0008] The technical scheme of the present application also has the following characteristics: The mobile curved surface fitting interpolation technology is used to fill the holes of the collected elevation data, specifically: first, select appropriate error parameters to evaluate the data hole filling effect; then, through the construction of mobile curved surface, the fitting of polynomial, the construction of error equation, and the solving of equation, the new elevation data value of the interpolation point is obtained, and the obtained new elevation data constitutes the model terrain 3s data set / data source.

[0009] By selecting appropriate error parameters, the data hole filling effect can be evaluated; by constructing a mobile curved surface, the coordinate origin can be matched; by fitting a polynomial, the curved surface data can be fitted; by constructing an error equation, the weight coefficient can be obtained; by solving the equation, the curved surface coefficient can be obtained.

[0010] The specific error parameters are selected as follows: Through the mobile surface fitting interpolation technology, the filling experiment is carried out, and the elevation difference is compared with the original data elevation value, the average elevation error u and the mean error RMSE are selected, and the numerical size of the two types of errors is calculated, and the u and RMSE are obtained to evaluate the effect of hole filling, and the specific formula is as follows:

[0011] Among them, u is the average value of the elevation difference between the real point and the interpolation point, n is the number of data points;

[0012] Among them, RMSE is the mean error of the real point and the interpolation point, that is, the square root of the square mean value of the elevation difference between the real point and the interpolation point, the mean error represents the data accuracy of the interpolation point, the smaller the mean error, the higher the data accuracy of the interpolation point, and the more reliable the interpolation algorithm, and vice versa. The algorithm is less reliable; Among the above two types of error indicators, di is the elevation difference between the real point and the interpolation point, and the calculation formula is:

[0013] Among them, h r and h f are the elevation values of the real point and the interpolation point, respectively. If the elevation difference is positive, the elevation value of the interpolation point is less than that of the real point, and if it is negative, the elevation value of the interpolation point is greater than that of the real point.

[0014] The construction of the moving surface is as follows: Take the point to be solved P(xp, yp) as the coordinate origin, and make a circular area with R as the radius. Select n known elevation data points in the circular area to fit and construct the surface, and complete the interpolation by solving the surface coefficients; According to the following formula, match the original DEM elevation coordinate origin, and move the coordinate origin to the interpolation point; .

[0015] The fitting of the polynomial is as follows: Based on the n DEM points in the circular area, a quadratic polynomial is selected to fit the data elevation P(xp, yp) of the interpolation point Z , and the formula is as follows:

[0016] wherein, A, B, C, D, E, F are undetermined coefficients, respectively, x , y is the coordinate of the P point.

[0017] The construction of the error equation is specifically as follows: The weight coefficients of the known data points i are calculated one by one, and the formula is as follows:

[0018] wherein, di is the distance of the interpolation point; is obtained from Z - Zi The error equation of the point pi is as follows:

[0019] wherein, Zi is the true elevation value, and satisfies the following relationship:

[0020] Therefore, n the error equation of the n data points can be constructed as follows:

[0021] wherein,

[0022] .

[0023] The equation is solved by homogenization, and the quadratic surface coefficients are solved by simultaneously solving the equation set according to the adjustment theory: The above error equation is solved, and the quadratic surface coefficients are solved by simultaneously solving the equation set according to the adjustment theory:

[0024] wherein, Z is the weight matrix, and satisfies the following relationship:

[0025]

[0026] Therefore, the undetermined coefficients A, B, C, D, E, F can be solved. Since the interpolation point is to be interpolated, the value solved by the equation is the final elevation value of the interpolation point F , and F is the undetermined coefficient solved by the above formula. P(xp, yp) Zp ​​

[0027] The beneficial effects of the present application are: (1) The high-resolution mode terrain data construction method of the present application, through the key technology research and development of data cavity filling by MCSFIT, introduces SRTM-DEM elevation data into the numerical mode for the first time, and realizes the resolution of the existing mode self-provided terrain data to be improved by 10 times (from 30 s to 3 s). (2) The high-resolution mode terrain data construction method of the present application, based on the new terrain data driven mode, can depict the fine / complex underlying surface terrain which cannot be depicted by the original mode, and lay the foundation for high-resolution simulation of disaster weather. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a principle diagram of a high-resolution mode terrain data construction method of the present application.

[0029] Figure 2 is the simulation result of the new terrain (3 s terrain data) obtained by the present application; Figure 3 is the actual terrain (the picture is from the network); Figure 4 is the simulation result of the existing mode self-provided 30 s terrain data; Figure 5 is a grid point configuration schematic diagram of the moving curved surface fitting interpolation algorithm of the present application. DETAILED DESCRIPTION

[0030] The technical solutions of the present application will be further described in detail below in combination with the description of the drawings and specific embodiments.

[0031] Embodiment 1 As shown in Figure 1 , the high-resolution mode terrain data construction method of the present application is based on SRTM-DEM (Shuttle Radar Topography Mission-Digital Elevation Model) to collect elevation data, and uses moving curved surface fitting interpolation technology (Moving Curved Surface Fitting Interpolation Technique, abbreviated as MCSFIT) to fill the cavity of the collected elevation data.

[0032] The mobile curved surface fitting interpolation technology is used for filling the holes of the collected height data, specifically: firstly, appropriate error parameters are selected to judge the data hole filling effect; then, the mobile curved surface is constructed, the polynomial is fitted, the error equation is constructed, and the equation is solved by normalization to obtain the new height data value of the interpolation point, and the obtained new height data forms the 3s data set / source of the model terrain.

[0033] By selecting appropriate error parameters, the data hole filling effect is judged; by constructing a mobile curved surface, the coordinate origin is matched; by fitting a polynomial, the curved surface data is fitted; by constructing an error equation, the weight coefficient is obtained; by solving the equation by normalization, the curved surface coefficient is obtained.

[0034] At present, the highest horizontal resolution of the model self-provided terrain data is 900 m, which is the GTOPO 30s height data, which is a terrain data publicly released in the 1990s. This resolution is sufficient for simulating global or large-area disaster processes, but it is rough for accurately simulating small-scale local extreme disaster processes, that is, it is difficult to depict small-scale terrain. At this time, it is particularly necessary to produce higher resolution and more actual model terrain data.

[0035] The present application introduces SRTM-DEM height terrain data and fills the data holes, and drives the numerical model with the new terrain data after filling to realize high-resolution replication and simulation of the actual terrain.

[0036] SRTM-DEM is the current international advanced terrain model data, which is obtained by remote sensing measurement of global terrain by space shuttle carrying an interferometric aperture radar system. The publicly shared height data has a resolution of up to 90 m. However, due to the influence of radar shadow, water body mirror reflection and other factors during data acquisition, data remote sensing errors may occur, resulting in blank missing data, useless or abnormal data, which is called data hole. Therefore, when using SRTM-DEM terrain data, the data holes in the target latitude and longitude range must be repaired and filled in combination with the actual local terrain. The core of the present application is the data hole filling method of the mobile curved surface fitting interpolation technology (MSCFIT).

[0037] The high-resolution model terrain data construction method of the present application realizes the goal of improving the resolution of model terrain data by 10 times (i.e. constructing 3s terrain data). The terrain data under the new model will depict the small-scale mountain ranges that the traditional terrain data cannot depict, which will help to realize / replicate / reproduce more actual and more detailed model underlying surface terrain in the model.

[0038] Experiments prove that the application can successfully achieve the goal of driving mode with SRTM-DEM high-resolution terrain data by accurate and effective hole filling, that is, successfully achieving the goal of improving the resolution of mode terrain data from 30s to 3s, and breaking the bottleneck of small-scale terrain simulation.

[0039] In order to demonstrate the advancement of the application, the internationally advanced environmental model WRF-Chem is selected for experiment, and the typical terrain underlay surface of China, "the climate watershed of north and south of China-Qinling terrain", is selected as the experimental content, and the east section of Qinling Mountain (Qinling terrain in the west of Henan Province) is focused on.

[0040] 1. Experimental platform In order to ensure the smoothness of high-resolution mode operation, the mode platform is built on a Dell T7920 tower workstation: equipped with a Xeon Gold 6428R / 24 core / 48 thread processor with a main frequency of 3.0 GHz, in addition, the workstation memory is 128G, containing 1T solid state disk+4T hard disk, and the graphics card is configured as RTX 3060-12G.

[0041] 2. Experimental results and effect comparison The east section of Qinling Mountain is selected for illustration. Figure 2 The simulation results of the new terrain (3s terrain data) are Figure 3 The actual terrain (picture from the network), Figure 4 The simulation results of the mode built-in 30s terrain data.

[0042] The new terrain data has more advantages than the mode built-in 30s terrain data in small-scale terrain simulation: the new terrain data simulates five small-scale mountains in the east section of Qinling Mountain, including XiaoShan, XiongErShan, WaiFangShan, FuNiuShan and SongShan; the new terrain data even simulates five γ medium-scale mountain ranges in the five mountains, including the southeast mountain range of SongShan marked with 1-2 and the south mountain range of FuNiuShan marked with 3-5. In contrast Figure 4 It can be seen that such fine terrain features cannot be simulated by the mode built-in terrain data.

[0043] Example 2 As Figure 1 shown, the same as example 1, in the high-resolution mode terrain data construction method of the application, the appropriate error parameter is selected as follows: The moving surface fitting interpolation technology is used to carry out the hole filling experiment, and the elevation difference is compared with the original data elevation value, and the average elevation error uand the mean error RMSE , and the numerical size of the two types of errors is obtained, and u and RMSE evaluate the effect of cavity filling, and the specific formula is as follows:

[0044] wherein, u is the average value of the height difference between the true point and the interpolation point, n is the number of data points;

[0045] wherein, RMSE is the mean error between the true point and the interpolation point, that is, the square root of the square mean of the height difference between the true point and the interpolation point, the mean error represents the data accuracy of the interpolation point, the smaller the mean error, the higher the data accuracy of the interpolation point, and the more reliable the interpolation algorithm, and vice versa. The algorithm is less reliable; In the above two types of error indicators, di is the height difference between the true point and the interpolation point, and the calculation formula is:

[0046] wherein, h r and h f are the true point elevation value and the interpolation point elevation value, respectively, if the elevation difference value is positive, the interpolation point elevation value is less than the true point elevation value, and if it is negative, the interpolation point elevation value is greater than the true point elevation value.

[0047] Example 3 As shown in Figure 1 , different from example 2, in the high-resolution mode terrain data construction method of the present application in example 3, the construction of the moving surface is specifically: As shown in Figure 5 , taking the point to be solved P(xp, yp) (red point) as the coordinate origin, and making a circular region with R as the radius. Select n known elevation data points (black points) in the circular region to fit and construct the surface, and complete the interpolation by solving the surface coefficients; According to the following formula, match the original DEM elevation coordinate origin, and move the coordinate origin to the interpolation point; .

[0048] Example 4 As shown in Figure 1 , different from example 3, in the high-resolution mode terrain data construction method of the present application in example 4, the fitting of the polynomial is specifically: Based on the DEM points in the circular region n , the data elevation of the interpolation points P(xp, yp) is fitted by a quadratic polynomial, and the formula is as follows: Z

[0049] wherein, A, B, C, D, E, F are undetermined coefficients, x , y are the coordinates of the P point.

[0050] Example 5 As shown in the table, different from example 4, in the high-resolution mode terrain data construction method of the present application in example 5, the construction of the error equation is as follows: Figure 1 The weight coefficient of each known data point is calculated, and the formula is as follows: i

[0051] wherein, di is the distance of the point to be interpolated; The error equation of the point Z - Zi is as follows: pi wherein,

[0052] is the true elevation value, and satisfies the following relationship: Zi

[0053] Therefore, n the error equation of the data point can be constructed as follows:

[0054] wherein,

[0055] .

[0056] Example 6 As shown in the table, different from example 5, in the high-resolution mode terrain data construction method of the present application in example 6, the equation is solved by normalization, and the specific process is as follows: Figure 1 The above error equation is solved, and according to the adjustment theory, the quadratic surface coefficients can be solved by simultaneously solving the equation group:

[0057] wherein, Z is the weight matrix, and satisfies the following relationship:​​​​

[0058]

[0059] Therefore, the undetermined coefficients can be obtained. A, B, C, D, E, F .

[0060] Due to the interpolation point Therefore, the equation requires F The value is the interpolation point. P(xp, yp) The final elevation value ( Zp Here, F is the undetermined coefficient obtained from the above formula.

[0061] The foregoing description illustrates and describes several preferred embodiments of the invention. However, as previously stated, it should be understood that the invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the inventive concept described herein through the foregoing teachings or techniques or knowledge in related fields. Any modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the invention should be within the protection scope of the appended claims.

Claims

1. A method of constructing high resolution mode terrain data, characterized by, Based on SRTM-DEM collection elevation data, and using mobile surface fitting interpolation technology to fill the collected elevation data cavity.

2. The high-resolution mode terrain data construction method according to claim 1, characterized by, The mobile surface fitting interpolation technology is used to fill the collected elevation data cavity, specifically: first select the appropriate error parameter, to evaluate the data cavity filling effect; then through the construction of mobile surface, polynomial fitting, error equation construction, equation of law solving, get the new elevation data value of interpolation point, the new elevation data obtained constitutes the pattern of terrain 3s dataset / source.

3. The high-resolution mode terrain data building method according to claim 2, characterized by, By selecting the appropriate error parameter method, the data cavity filling effect is realized; by constructing the mobile surface method, the matching coordinate origin is realized; by fitting the polynomial method, the surface data fitting is realized; by constructing the error equation method, the weight coefficient is realized; by solving the equation of law method, the surface coefficient is realized.

4. The high-resolution mode terrain data building method according to claim 3, characterized by, Selecting the appropriate error parameter is specifically: Through the mobile surface fitting interpolation technology to carry out the filling hole experiment, with the original data elevation value elevation difference comparison, select the average elevation error u and mean error RMSE , and the numerical value of the two kinds of error, according to the obtained u and RMSE evaluation of the hole filling effect, the specific formula is as follows: wherein, u is the average of the true point and interpolated point elevation difference, n is the number of data points; wherein, RMSE is the mean error of the real points and the interpolation points, i.e., the square root of the square mean of the elevation difference between the real points and the interpolation points. The mean error represents the data accuracy of the interpolation points. The smaller the mean error, the higher the data accuracy of the interpolation points, and the more reliable the interpolation algorithm. In the above two kinds of error indicators, di For the height difference between the true point and the interpolation point, the formula is: wherein h r and h f are the true point elevation value and the interpolated point elevation value, respectively, the interpolated point elevation value being less than the true point elevation value if the elevation difference value is positive, and the interpolated point elevation value being greater than the true point elevation value if the elevation difference value is negative.

5. The high-resolution mode terrain data building method of claim 4, wherein, The construction of mobile surface is specifically: The point to be sought P(xp, yp) Take the coordinate origin, make a circular area with R As the radius; select n The known elevation data points in the circular area to fit the surface, and complete the interpolation by solving the surface coefficients; According to the following formula, match the original DEM elevation coordinate origin, and move the coordinate origin to the interpolation point; 。 6. The high-resolution mode terrain data building method of claim 5, wherein, Polynomial fitting is specifically: Based on the circular region n For each DEM point, a quadratic polynomial is used to interpolate the points. P(xp, yp) Data elevation Z The formula for surface fitting is as follows: wherein, A, B, C, D, E, F are the undetermined coefficients, respectively, x , y are the coordinates of point P.

7. The high-resolution mode terrain data building method of claim 6, wherein, The construction of error equation is specifically: The weight coefficients of the known data points are sought one by one, as follows: i w = (y - y0) / (x - x0) wherein di is the distance of the point to be interpolated from; obtained by Z-Zi the point pi The error equation for the point is as follows: wherein Zi is a true elevation value and satisfies the following relationship: Thus, n The error equation can be constructed from the data points as follows: Among them, 。 8. The high-resolution mode terrain data building method of claim 7, wherein, The equation of law solving is specifically: Solve the above error equation, according to the adjustment theory, the simultaneous equations can be solved quadratic surface coefficient: wherein Z is a weight matrix and satisfies the following relation: Therefore, the undetermined coefficients can be obtained. A, B, C, D, E, F ; Since the point to be interpolated , the equation is solved F for the value of the interpolated point P(xp, yp) , which is the final elevation value of the point Zp , where F is the coefficient to be determined from the above equation.

Images