CN108647740A - The method for carrying out multi-source precipitation fusion using high-resolution landform and meteorological factor - Google Patents

Abstract

The invention discloses a kind of methods carrying out multi-source precipitation fusion using high-resolution landform and meteorological factor, include the following steps：The DEM raster datas for extracting target basin, obtain the mask files in basin；By target basin DEM raster datas calculate the gradient in target basin, slope aspect, roughness of ground surface, to the distance in coastline；The air speed data in basin is extracted according to mask files；The satellite precipitation data in basin is extracted using mask files；NO emissions reduction is carried out to original satellite precipitation, obtains the satellite precipitation of Rkm：Calculate the deviation of the satellite precipitation and actual measurement website precipitation after NO emissions reduction；Using Geographical Weighted Regression Model, the precipitation deviation of Rkm grids is calculated；Obtain Rkm fusion Precipitation Products.The present invention merges station data and satellite Precipitation Products and carries out NO emissions reduction to satellite precipitation, obtains the Precipitation Products of high-spatial and temporal resolution, can realize the conversion of precipitation data from point to surface, and the input to refine hydrological model provides data supporting.

Description

A Method of Multi-source Precipitation Fusion Using High Resolution Topographic and Meteorological Factors

technical field

The invention belongs to the technical field of hydrology and meteorology, and in particular relates to a multi-source precipitation fusion method using high-resolution terrain and meteorological factors.

Background technique

China is a country with frequent flood disasters, and hydrological model is an important means to solve flood disasters. Precipitation is the most critical input source of hydrological models, and the accuracy and timeliness of precipitation directly affect the accuracy and reliability of simulation results.

At present, the methods of obtaining precipitation data mainly include ground observation, satellite and radar quantitative precipitation estimation, and model quantitative precipitation forecasting. For a long time, the routine observation of precipitation has mainly relied on the observation stations deployed on the surface, and the limited observation results represent the real precipitation within tens or even hundreds of square kilometers around. The size and type of actual precipitation have significant temporal and spatial variability, and there is a problem of substituting points for surface at ground stations, especially in areas where there are few stations, the observed precipitation cannot effectively reflect the spatial variability of spatial precipitation. Difficulties in the research. Radar quantitative precipitation estimation has the advantages of high spatial resolution and strong real-time performance, but its coverage is limited because it is easily affected by cover. With the development of satellite remote sensing technology at home and abroad, remote sensing precipitation observations based on weather radar and satellites have been continuously improved, making up for the lack of spatial distribution of ground stations and providing new means for monitoring precipitation. At present, satellite remote sensing has unique advantages in obtaining global precipitation that changes in time and space, providing unprecedented satellite precipitation products such as TRMM, GPM, COMRPH, PERSIANN, FY-3B, FY-3C, etc. Satellite quantitative precipitation estimation has the advantages of wide coverage and relatively continuous observation time. However, due to the limitations of remote sensing detection instruments and inversion algorithms, satellite precipitation products have low spatial resolution and relatively low accuracy, and the inversion ability for solid precipitation is very poor. limited.

Fusion of station data and satellite radar precipitation has become an effective way to improve precipitation products. At present, there are many methods of precipitation fusion, such as: optimal interpolation, Kalman filter, particle filter, Bayesian estimation, probability density and other methods, none of which are involved Downscaling to satellite precipitation.

Contents of the invention

In order to obtain precipitation products with high spatial and temporal resolution, the present invention provides a multi-source precipitation fusion method using high-resolution terrain and meteorological factors. Utilizing topographic data and meteorological data with high temporal and spatial resolution, combining the actual precipitation at the site and satellite precipitation products, and downscaling the satellite precipitation to obtain precipitation products with high spatial and temporal resolution, which can realize the conversion of precipitation data from point to area, and provide fine The input of the hydrological model provides data support.

In order to obtain high-resolution precipitation fusion products, the present invention specifically adopts the following technical solutions:

A method for multi-source precipitation fusion using high-resolution terrain and meteorological factors, characterized in that it comprises the following steps:

Step 1, extract the DEM raster data of the target watershed, and obtain the mask file of the watershed;

Step 2, calculate the slope, aspect, surface roughness, and distance to the coastline of the target watershed from the DEM raster data of the target watershed;

Step 3, extract the wind speed data of the watershed according to the mask file;

Step 4, using the mask file to extract the satellite precipitation data of the watershed;

Step 5, downscale the original satellite precipitation to obtain the satellite precipitation of Rkm:

Step 6, calculating the deviation between the satellite precipitation after downscaling and the precipitation at the measured station;

Step 7, using the geographically weighted regression model to calculate the precipitation deviation of the Rkm grid; where the geographically weighted regression model is:

In the formula: y _i is the precipitation deviation; a ₀ is a constant item; x _ik is the matrix of i*k, i is the number of grids, and k represents the type of variable, namely DEM, slope, aspect, surface roughness, and coastline distance and wind speed; a _ik is the corresponding coefficient term;

a _ik ＝(x _ik ^T w(i)x _ik ) ^-1 x _ik ^T w(i)y _i (5)

In the formula: x _ik ^T is matrix transposition; w(i) is weight;

Step 8, the precipitation deviation of the Rkm grid is added to the satellite precipitation of Rkm to obtain the Rkm fusion precipitation product;

In the formula: To incorporate precipitation products, is the satellite precipitation of the original Rkm, is the deviation of the Rkm grid.

The coefficient term a _ik of the geographically weighted regression model is:

a _ik ＝(x _ik ^T w(i)x _ik ) ^-1 x _ik ^T w(i)y _i (5)

In the formula: x _ik is the matrix of DEM, slope, aspect, surface roughness, distance to the coastline, and wind speed; X ^T is the matrix transpose; w(i) is the weight.

In step 5, the satellite precipitation downscaled to Rkm from the original satellite precipitation is:

In the formula, f(x, y) is the satellite precipitation at the coordinate point (x, y); Q ₁₁ = (x ₁ , y ₁ ), Q ₁₂ = (x ₁ , y ₂ ), Q ₂₁ = (x ₂ , y ₁ ), Q ₂₂ =(x ₂ , y ₂ ) are the four acquired coordinate points of the original satellite.

In the step 6, the precipitation deviation is calculated by combining the site measured data and the downscaled satellite precipitation data, including:

Step 61, according to the latitude and longitude of the actual measurement site, determine the position of the actual measurement site in the grid, that is, the row and column of the site in the watershed;

In the formula: row is the row of the station data, col is the column of the station data, delta is the spatial resolution of the watershed data, lat _u is the maximum latitude of the watershed, lon _l is the minimum longitude of the watershed, lat _s is the latitude of the station ; lon _s is the longitude of the site;

Step 62, according to the number of rows and columns of the station, read the grid precipitation data corresponding to the station;

In step 63, the satellite precipitation of the grid corresponding to the station precipitation is subtracted from the actual measured precipitation of the station to obtain the precipitation deviation of the grid corresponding to the station.

In the step 8, the precipitation deviation of the Rkm grid is added to the satellite precipitation of Rkm to obtain the Rkm fusion precipitation product, including:

In the formula: To incorporate precipitation products, is the satellite precipitation of the original Rkm, is the deviation of the Rkm grid.

Extracting the DEM raster data of the target watershed in the step 1 specifically includes the following steps:

Step 11, filling the pit;

Step 12, calculate the flow direction;

Step 13, calculating the confluence flow;

Step 14, determine the outlet site of the watershed;

Step 15, extracting the target watershed.

In the step 2, the watershed slope, aspect, surface roughness, and distance to the coastline are calculated using the watershed DEM raster data, including:

Step 21, calculate the watershed slope of Rkm by the DEM data of the target watershed;

Step 22, calculating the watershed aspect of Rkm from the DEM grid data of the target watershed;

Step 23, calculating the watershed surface roughness of Rkm from the DEM grid data of the target watershed;

In step 24, the distance to the coastline in Rkm is calculated from the DEM grid data of the target watershed and the coastline of China.

In the step 3, according to the watershed mask file, the wind speed data of the watershed is obtained, including:

Step 31, download global wind speed data at the European Union Meteorological Center;

Step 32, according to the mask file in step 1 as the watershed boundary, extract the wind speed data of the target watershed.

In the step 4, extract the satellite precipitation data of the basin according to the mask file, including:

Step 41, download the global CMORPH satellite precipitation data, and obtain the coverage, spatial resolution, and time resolution of CMORPH;

Step 42, according to the mask in step 1 as the watershed boundary, extract the satellite precipitation data of the target watershed.

Beneficial effects of the present invention: the present invention provides a multi-source precipitation fusion method using high-resolution topographical and meteorological factors. According to the DEM of the watershed, the slope, aspect, surface roughness, and distance to the coastline of the watershed are proposed ; Extract wind speed and satellite precipitation data according to the mask file of the river basin, combine satellite precipitation and station precipitation, realize the downscaling of satellite precipitation while integrating multi-source precipitation. This method uses topographic and meteorological factors with high temporal and spatial resolution as the basic data. The data source is stable and reliable. The functional relationship between variables in the method is clear. The deviation between the station and satellite precipitation is used to correct the satellite precipitation, ensuring The objective and rationality of the results; make up for the respective shortcomings of satellite precipitation and station precipitation, realize point-to-surface conversion, obtain precipitation products with high temporal and spatial resolution, and provide data support for the input of refined hydrological models.

Description of drawings

Fig. 1 is a schematic diagram of the calculation flow of the present invention.

Fig. 2 is a schematic diagram of a watershed DEM extracted by the present invention.

Fig. 3 is a schematic diagram of the watershed mask extracted by the present invention.

Fig. 4 is a schematic diagram of the watershed slope calculated in the present invention.

Fig. 5 is a schematic diagram of the slope aspect of the watershed calculated in the present invention.

Fig. 6 is a schematic diagram of surface roughness extracted in the present invention.

Fig. 7 is a schematic diagram of the distance to the coastline extracted by the present invention.

Fig. 8 is a schematic diagram of the multi-year average wind speed extracted by the present invention.

Fig. 9 is a schematic diagram of satellite original precipitation extracted in the present invention.

Fig. 10 is a schematic diagram of satellite precipitation after bilinear interpolation in the present invention.

Fig. 11 is a schematic diagram of the fusion precipitation product obtained in the present invention.

Fig. 12 is a schematic diagram of comparison between original satellite data, site data and fusion results in the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

As shown in Fig. 1, a kind of utilization high-resolution terrain and meteorological factor provided by the present invention carries out multi-source precipitation fusion method, comprises the following steps:

Step 1, extract the DEM raster data and mask file of the target watershed, specifically:

Step 11, filling the pit;

Step 12, calculate the flow direction;

Step 13, calculating the confluence flow;

Step 14, determine the outlet site of the watershed;

Step 15, extracting the target watershed;

Step 2, use the watershed DEM raster data to calculate the watershed slope, aspect, surface roughness, and distance to the coastline, specifically:

Step 21, calculate the watershed slope of Rkm by the DEM data of the target watershed;

Step 22, calculating the watershed aspect of Rkm from the DEM grid data of the target watershed;

Step 23, calculating the watershed surface roughness of Rkm from the DEM grid data of the target watershed;

Step 24, calculate the distance to the coastline of Rkm from the DEM grid data of the target watershed and the coastline of China;

Step 3, according to the watershed mask file, obtain the wind speed data of the watershed, specifically:

Step 31, download global wind speed data at European Union Meteorological Center (ECWMF);

Step 32, according to the mask file in step 1 as the watershed boundary, extract the wind speed data of the target watershed;

Step 4, according to the mask file, extract the satellite precipitation data of the basin, specifically:

Step 41, downloading global CMORPH satellite precipitation data, specifying CMORPH coverage, spatial resolution, time resolution, etc.;

Step 42, according to the mask in 1 as the watershed boundary, extract the satellite precipitation data of the target watershed;

Step 5, the obtained original satellite precipitation, adopts the method of bilinear interpolation, obtains the satellite precipitation data of Rkm, specifically:

Bilinear interpolation is a linear interpolation extension of the interpolation function with two variables, and its core idea is to perform linear interpolation in two directions respectively. The original satellite precipitation spatial resolution is 8km, and the satellite precipitation data with a spatial resolution of Rkm is obtained through bilinear interpolation.

If you want to get the value of the unknown function f at P=f(x, y), suppose the known function f is at Q ₁₁ =(x ₁ ,y ₁ ), Q ₁₂ =(x ₁ ,y ₂ ), Q ₂₁ = (x ₂ , y ₁ ), Q ₂₂ = (x ₂ , y ₂ ) the values of the four points. The interpolation announcement is as follows:

Step 6. Combining the measured data at the station with the downscaled satellite precipitation data, calculate the precipitation deviation, specifically:

Step 61, according to the latitude and longitude of the actual measurement site, determine the position of the actual measurement site in the grid, that is, the row and column of the site in the watershed;

In the formula: row is the row where the station data is located, delta is the spatial resolution of the watershed data, lat _u is the maximum latitude of the watershed, and lat _s is the latitude of the station.

In the formula: col is the column where the station data is located, delta is the spatial resolution of the watershed data, lon _l is the minimum longitude of the watershed, and lon _s is the longitude of the station.

Step 62, according to the number of rows and columns of the station, read the grid precipitation data corresponding to the station;

In step 63, the satellite precipitation of the grid corresponding to the station precipitation is subtracted from the actual precipitation at the station to obtain the precipitation deviation of the grid corresponding to the station; in step 7, the geographic weighted regression model is constructed to calculate the precipitation deviation step of each grid in Rkm, specifically :

Step 71, constructing a geographically weighted regression model according to deviation, topographic and meteorological factors with high temporal and spatial resolution;

In the formula: y _i is the precipitation deviation, x _ik is the DEM, slope, aspect, surface roughness, distance to the coastline, wind speed, and a _ik is the corresponding coefficient item.

Step 72, estimating the coefficients of the geographically weighted regression model;

a _ik ＝(x _ik ^T w(i)x _ik ) ^-1 x _ik ^T w(i)y _i (5)

In the formula: a _ik is the coefficient of the i-th grid, x _ik is the matrix of DEM, slope, aspect, surface roughness, distance to the coastline, and wind speed, and w(i) is the weight.

Step 73, bring publicity 5 into formula 4, and calculate the deviation of each grid;

Step 8, the precipitation deviation of the Rkm grid is added to the Rkm satellite precipitation to obtain the Rkm fusion precipitation product, specifically:

In the formula: To incorporate precipitation products, is the satellite precipitation of the original Rkm, is the deviation of the Rkm grid.

Taking the Ziwu River Basin in Shaanxi Province as an example, the original DEM data of the study area are digital elevation data jointly provided by the United States Geological Survey (USUG) and the National Basic Geographic Information Center, specifically:

Step 1, extract the DEM raster data of the target watershed and the mask file of the watershed, specifically:

Step 11, filling the pit;

Step 12, calculate the flow direction;

Step 13, setting the flow threshold and calculating the confluence flow;

Step 14, determine the outlet site of the watershed;

Step 15, extract the DEM of the target watershed, as shown in Figures 2 and 3.

Step 2, use the watershed DEM raster data to calculate the watershed slope, aspect, surface roughness, and distance to the coastline, specifically:

Step 21, calculate the watershed slope of Rkm by the DEM data of the target watershed, as shown in Figure 4;

Step 22, calculate the watershed slope aspect of Rkm from the DEM raster data of the target watershed, as shown in Figure 5;

Step 23, calculate the watershed surface roughness of Rkm from the DEM raster data of the target watershed, as shown in Figure 6;

Step 24, calculate the distance to the coastline in Rkm from the DEM grid data of the target watershed and the coastline of China, as shown in Figure 7;

Step 3, according to the watershed mask file, obtain the wind speed data of the watershed, specifically:

Step 31, download global wind speed data at European Union Meteorological Center (ECWMF);

Step 32, according to the mask file in step 1 as the watershed boundary, extract the wind speed data of the target watershed, as shown in Figure 8;

Step 4, according to the mask file, extract the satellite precipitation data of the basin, specifically:

Step 41, downloading global CMORPH satellite precipitation data, specifying CMORPH coverage, spatial resolution, time resolution, etc.;

Step 42, according to the mask in 1 as the watershed boundary, extract the satellite precipitation data of the target watershed, as shown in Figure 9;

Step 5, the original satellite precipitation is taken, and the bilinear interpolation method is used to obtain the satellite precipitation of Rkm, specifically:

Bilinear interpolation is a linear interpolation extension of the interpolation function with two variables, and its core idea is to perform linear interpolation in two directions respectively. The original satellite precipitation spatial resolution is 8 km, and the satellite precipitation data with a spatial resolution of R km is obtained through bilinear interpolation, as shown in Figure 10.

If you want to get the value of the unknown function f at P=f(x, y), suppose the known function f is at Q ₁₁ =(x ₁ ,y ₁ ), Q ₁₂ =(x ₁ ,y ₂ ), Q ₂₁ = (x ₂ , y ₁ ), Q ₂₂ = (x ₂ , y ₂ ) the values of the four points. The interpolation announcement is as follows:

Step 6. Combining the measured data of the station with the downscaled satellite precipitation data, the precipitation deviation is calculated as follows:

Step 61, according to the latitude and longitude of the actual measurement site, determine the position of the actual measurement site in the grid, that is, the row and column of the site in the watershed;

In the formula: row is the row where the station data is located, delta is the spatial resolution of the watershed data, lat _u is the maximum latitude of the watershed, and lat _s is the latitude of the station.

In the formula: col is the column where the station data is located, delta is the spatial resolution of the watershed data, lon _l is the minimum longitude of the watershed, and lon _s is the longitude of the station.

Step 62, according to the number of rows and columns of the station, read the grid precipitation data corresponding to the station;

In step 63, the satellite precipitation of the grid corresponding to the precipitation at the site is subtracted from the actual precipitation at the site to obtain the precipitation deviation of the grid corresponding to the site; in step 7, the precipitation deviation of each grid in Rkm is calculated using the geographically weighted regression model, specifically:

Step 71, constructing a geographically weighted regression model according to deviation, topographic and meteorological factors with high temporal and spatial resolution;

In the formula: y _i is the precipitation deviation, x _ik is the DEM, slope, aspect, surface roughness, distance to the coastline, wind speed, and a _ik is the corresponding coefficient item.

Step 72, estimating the coefficients of the geographically weighted regression model;

a _ik ＝(x _ik ^T w(i)x _ik ) ^-1 X ^xikT w(i)y _i (5)

In the formula: a _ik is the coefficient of the i-th grid, x _ik is the matrix of DEM, slope, aspect, surface roughness, distance to the coastline, and wind speed, and w(i) is the weight.

Step 73, bring publicity 5 into formula 4, and calculate the deviation of each grid;

Step 8, the precipitation deviation of the Rkm grid is added to the Rkm satellite precipitation to obtain the Rkm fusion precipitation product, specifically:

In the formula: To incorporate precipitation products, is the satellite precipitation of the original Rkm, is the deviation of the Rkm grid.

The product of fused precipitation is shown in Figure 11.

In order to show the results of the method, the summer in the Meridian River Basin is selected for verification, as shown in Figure 12, which are satellite original precipitation, station precipitation, and fusion precipitation.

Claims (9)

1. A method utilizing high-resolution terrain and meteorological factors to carry out multi-source precipitation fusion, is characterized in that, comprises the following steps: Step 1, extract the DEM raster data of the target watershed, and obtain the mask file of the watershed; Step 2, calculate the slope, aspect, surface roughness, and distance to the coastline of the target watershed from the DEM raster data of the target watershed; Step 3, extract the wind speed data of the watershed according to the mask file; Step 4, using the mask file to extract the satellite precipitation data of the watershed; Step 5, downscale the original satellite precipitation to obtain the satellite precipitation of Rkm: Step 6, calculating the deviation between the satellite precipitation after downscaling and the precipitation at the measured station; Step 7, using the geographically weighted regression model to calculate the precipitation deviation of the Rkm grid; where the geographically weighted regression model is: In the formula: y _i is the precipitation deviation; a ₀ is a constant item; x _ik is the matrix of i*k, i is the number of grids, and k represents the type of variable, namely DEM, slope, aspect, surface roughness, and coastline distance and wind speed; a _ik is the corresponding coefficient item; θ _i is the calculation deviation; a _ik ＝(x _ik ^T w(i)x _ik ) ^-1 x _ik ^T w(i)y _i (5) In the formula: x _ik ^T is matrix transposition; w(i) is weight; Step 8, the precipitation deviation of the Rkm grid is added to the satellite precipitation of Rkm to obtain the Rkm fusion precipitation product; In the formula: To incorporate precipitation products, is the satellite precipitation of the original Rkm, is the deviation of the Rkm grid. 2. the method for utilizing high-resolution topography and meteorological factors according to claim 1 to carry out multi-source precipitation fusion, is characterized in that, the coefficient term a _ik of geographic weighted regression model is: a _ik ＝(x _ik ^T w(i)x _ik ) ^-1 x _ik ^T w(i)y _i (5) In the formula: x _ik is the matrix of DEM, slope, aspect, surface roughness, distance to the coastline, and wind speed; X ^T is the matrix transpose; w(i) is the weight. 3. the method for utilizing high-resolution topography and meteorological factors according to claim 1 to carry out multi-source precipitation fusion, it is characterized in that, step 5 carries out downscaling to the satellite precipitation of Rkm to original satellite precipitation: In the formula, f(x, y) is the satellite precipitation at the coordinate point (x, y); Q ₁₁ = (x ₁ , y ₁ ), Q ₁₂ = (x ₁ , y ₂ ), Q ₂₁ = (x ₂ , y ₁ ), Q ₂₂ =(x ₂ , y ₂ ) are the four acquired coordinate points of the original satellite. 4. according to the method described in claim 3 utilizing high-resolution topography and meteorological factors to carry out multi-source precipitation fusion, it is characterized in that: in the described step 6, in conjunction with the satellite precipitation data after the site measured data and downscale, calculate Precipitation deviations, including: Step 61, according to the latitude and longitude of the actual measurement site, determine the position of the actual measurement site in the grid, that is, the row and column of the site in the watershed; In the formula: row is the row of the station data, col is the column of the station data, delta is the spatial resolution of the watershed data, lat _u is the maximum latitude of the watershed, lon _l is the minimum longitude of the watershed, lat _s is the latitude of the station ; lon _s is the longitude of the site; Step 62, according to the number of rows and columns of the station, read the grid precipitation data corresponding to the station; In step 63, the satellite precipitation of the grid corresponding to the station precipitation is subtracted from the actual measured precipitation of the station to obtain the precipitation deviation of the grid corresponding to the station. 5. the method for utilizing high-resolution terrain and meteorological factors according to claim 4 to carry out multi-source precipitation fusion, is characterized in that, in described step 8, the precipitation deviation of Rkm grid adds the satellite precipitation of Rkm, obtains Rkm fusion precipitation products, including: In the formula: To incorporate precipitation products, is the satellite precipitation of the original Rkm, is the deviation of the Rkm grid. 6. the method for utilizing high-resolution terrain and meteorological factors according to claim 1 to carry out multi-source precipitation fusion, is characterized in that, extracting the DEM raster data of target watershed in the described step 1, specifically comprises the following steps: Step 11, filling the pit; Step 12, calculate the flow direction; Step 13, calculating the confluence flow; Step 14, determine the outlet site of the watershed; Step 15, extracting the target watershed. 7. The method of utilizing high-resolution terrain and meteorological factors according to claim 1 to carry out multi-source precipitation fusion, wherein, in the step 2, the watershed slope, aspect, surface Roughness, distance to shoreline, including: Step 21, calculate the watershed slope of Rkm by the DEM data of the target watershed; Step 22, calculating the watershed aspect of Rkm from the DEM grid data of the target watershed; Step 23, calculating the watershed surface roughness of Rkm from the DEM grid data of the target watershed; In step 24, the distance to the coastline in Rkm is calculated from the DEM grid data of the target watershed and the coastline of China. 8. The method of utilizing high-resolution terrain and meteorological factors according to claim 1 to carry out multi-source precipitation fusion, wherein, in the step 3, according to the watershed mask file, the wind speed data of the watershed is obtained, including: Step 31, download global wind speed data at the European Union Meteorological Center; Step 32, according to the mask file in step 1 as the watershed boundary, extract the wind speed data of the target watershed. 9. according to the method for utilizing high-resolution terrain and meteorological factors described in claim 1 to carry out multi-source precipitation fusion, it is characterized in that, in the described step 4, extract the satellite precipitation data of basin according to mask file, comprise: Step 41, download the global CMORPH satellite precipitation data, and obtain the coverage, spatial resolution, and time resolution of CMORPH; Step 42, according to the mask in step 1 as the watershed boundary, extract the satellite precipitation data of the target watershed.