阅读说明：本技术 一种基于回波强度和回波顶高外推的短时定量降水预报方法 (Short-time quantitative rainfall forecasting method based on echo intensity and echo top height extrapolation ) 是由 邹海波 吴珊珊 单九生 吴文心 易雪婷 于 2021-09-01 设计创作，主要内容包括：本发明涉及大气科学领域,具体是一种基于回波强度和回波顶高外推的短时定量降水预报方法,用于基于雷达的短时定量降水预报。为了达到上述目的,本发明所采用的技术方案是：雷达基数据处理,剔除雷达回波观测中的非气象回波,并将数据从极坐标格式转换至笛卡尔直角坐标格式；雷达回波外推,对雷达回波强度和雷达回波顶高同时进行外推,获得两者的预报值；雷达降水反演,基于实时雷达回波强度、雷达回波顶高和降水观测资料,拟合出不同回波顶高的Z-R关系表达式；雷达降水预报,基于雷达回波强度和回波顶高的预报值,利用实时拟合出的不同回波顶高的Z-R关系表达式,对外推预报的回波强度进行降水反演,获得短时定量降水预报。(The invention relates to the field of atmospheric science, in particular to a short-time quantitative rainfall forecasting method based on extrapolation of echo intensity and echo top height, which is used for short-time quantitative rainfall forecasting based on radar. In order to achieve the purpose, the invention adopts the technical scheme that: radar base data processing, namely eliminating non-meteorological echoes in radar echo observation, and converting data from a polar coordinate format to a Cartesian rectangular coordinate format; extrapolating the radar echo, namely extrapolating the intensity of the radar echo and the height of the radar echo simultaneously to obtain predicted values of the intensity and the height of the radar echo; radar rainfall inversion, namely fitting Z-R relational expressions with different echo top heights based on real-time radar echo intensity, radar echo top height and rainfall observation data; and radar rainfall forecast, namely carrying out rainfall inversion on the echo intensity of the extrapolation forecast by utilizing Z-R relational expressions with different echo top heights, which are fit in real time, based on the forecast values of the radar echo intensity and the echo top heights, so as to obtain the short-time quantitative rainfall forecast.)

1. A short-time quantitative precipitation forecast method based on echo intensity and echo top height extrapolation is characterized by comprising the following steps:

s1, radar base data processing, namely decoding, quality control and coordinate conversion are carried out on the radar base data to obtain radar echo data of Cartesian coordinates;

s2, performing radar echo extrapolation, obtaining a radar echo motion vector (TREC vector) by a cross correlation Tracking (TREC) method based on gridded radar combined reflectivity data, filtering noise and incoordination in a heavier vector by using Barnes space to obtain a BTREC vector, and finally extrapolating the radar echo combined reflectivity and the radar echo top height by using a backward extrapolation scheme to obtain a prediction value of the radar echo intensity and the echo top height in 2 hours in the future;

and S3, radar precipitation inversion (based on the Z-R relation of echo top height grading), and interpolating radar combined reflectivity and radar echo top height data subjected to quality control processing and gridding processing to a precipitation observation station by using a bilinear interpolation method. Based on the echo peak height, the radar combined reflectivity and the observed rainfall are divided into different classes according to the interval of 1km, and the optimal Z-R relation of the different classes is fitted by using an optimal method.

Step S4: and (3) radar rainfall forecast, based on extrapolated radar combined reflectivity and echo top height data of different times within 2 hours in the future, calculating quantitative rainfall forecast values of different places at different moments within 2 hours in the future by using a Z-R relational expression of different echo top heights fitted (dynamically) at the current time.

2. The method of claim 1, wherein the short-term quantitative precipitation forecast is based on extrapolation of echo intensity and echo top height, and comprises: the specific content of the S1 includes:

s11, decoding radar base data, namely decoding the radar base data according to the type of the radar and the storage format of the base data to obtain radar echo data in polar coordinate formats at different elevation angles;

s12, radar echo quality control, namely utilizing an echo (reflectivity) vertical gradient RGDZ (W) (Z-Z)_up) Eliminating abnormal clutter of ground objects in radar echo observation at a depth of more than or equal to 20dBZ, wherein the weight coefficient W is equal to 1 when the radar slant range is 0-40km, the weight coefficient W is linearly decreased to 0 between 40-200km, Z is the radar echo (dBZ) of the current elevation angle, and Z is the radar echo of the current elevation angle_upThe radar echo (dBZ) of the upper elevation is obtained by using a fuzzy logic algorithm P_xRejecting isolated non-meteorological echoes, wherein N is the number of effective observations of the radar in a 5 multiplied by 5 window around an x point on an azimuth radial plane (namely a scanning plane of the radar on a certain elevation angle), wherein N is not more than 25 and not more than 0.75;

s13, calculating the combined reflectivity and the echo height, based on the radar reflectivity data in polar coordinate formats of different elevation angles after quality control processing, calculating the combined reflectivity (the maximum value of different elevation angle layers) and the radar echo height (namely, the highest height of a reflectivity factor of more than or equal to 18 dBZ), wherein the calculation formula of the radar echo height (relative to a radar station) is h-R sin theta + R²cos²θ/(2R_e) Where R is the radar detected slope distance, R_e8500km is the equivalent earth radius under the standard atmosphere, and theta is the highest elevation angle of the radar detection reflectivity factor of more than or equal to 18 dBZ.

S14, converting coordinates of radar data, namely, firstly, taking a radar survey station as a center, constructing a Cartesian rectangular coordinate grid with 600 multiplied by 600 grid points and 1 multiplied by 1km horizontal resolution, and then, according to a radar beam horizontal distance calculation formulaWherein h is_rCalculating the projection of the radar slant distance on a horizontal plane for the altitude of the radar station, and finally, transferring the radar data of a polar coordinate system (L, beta) to a Cartesian rectangular coordinate system through x ═ L · sin β and y ═ L · cos β, wherein β is the azimuth angle observed by the radar.

3. The method of claim 1, wherein the short-term quantitative precipitation forecast is based on extrapolation of echo intensity and echo top height, and comprises: the specific content of the S2 includes:

s21: calculating a TREC vector, dividing radar combined reflectivity data gridded at the time t (the number of lattice points is 600 multiplied by 600, and the horizontal resolution is 1km multiplied by 1km) into 40 multiplied by 40 small regions without overlapping, wherein the number of lattice points of each small region is 15 multiplied by 15, and converting the radar reflectivity data of each small region into a one-dimensional sequence Z according to the mode of firstly arranging the small regions in the x direction and then arranging the small regions in the y direction₁Is a reaction of Z₁Z consisting of any 15 multiplied by 15 grid points in radar combined reflectivity data of sequence and t + delta t time₂The sequence is subjected to spatial correlation analysis, and the correlation coefficient is calculated according to the formulaAnd the correlation number is the largest Z₂The small area corresponding to the sequence is used as a target area of the initial small area, and a vector from the center of the initial small area to the center of the target area is a TREC vector and reflects the movement condition of the radar echo;

step S22: in the BTREC vector calculation, due to the small scale change of the echo and the tracking failure, some noise or uncoordinated vectors exist in the TREC vector, the accuracy of the extrapolated echo is obviously influenced, and the components of the TREC vector in the x direction and the y direction are respectively used as variables F (x, y) to be input into a formulaAndobtaining component F of the BTREC vector₁(x, y), wherein: x and y are the positions of the radar returns in a cartesian orthogonal coordinate system,andis a Gaussian parameter, M is the number of valid lattice points of the TREC vector in the range of 100km from the (x, y) point, r_kIs point (x, y) and point (x)_k,y_k) The filter parameter G is 0.35 and C is 300.

Step S23: and (2) extrapolation of radar echo, namely interpolating a BTREC vector with a horizontal resolution of 15km multiplied by 15km to a grid (1km multiplied by 1km) with radar combined reflectivity and radar echo top height by utilizing a bilinear interpolation method, and extrapolating the radar combined reflectivity and the echo top height by utilizing a radar echo backward extrapolation method to obtain an extrapolation prediction value of the radar combined reflectivity and the radar echo top height at any time within 2 hours in the future, wherein the radar echo backward extrapolation method is Z_i,j(t+Δt)＝Z_i-Δi,j-Δj(t), wherein: z is the radar combined reflectivity or the echo top height, i and j are the serial numbers of a certain echo point in the x and y directions respectively, Δ t is the extrapolation prediction time, Δ i equals int (u · Δ t +0.5) is the moving distance of the echo point in the x direction in the prediction time, Δ j equals int (v · Δ t +0.5) is the moving distance of the echo point in the y direction in the prediction time, and u and v are the components of the BTREC vector in the x and y directions respectively.

Technical Field

The invention relates to the field of atmospheric science, in particular to a short-time quantitative rainfall forecasting method based on echo intensity and echo top height extrapolation.

Background

The Doppler weather radar data has high space-time resolution and has incomparable advantages with other observation data in the mesoscale meteorological service and research. Due to the spin-up problem of the numerical mode (i.e. the mode needs to be adjusted to reach the equilibrium state under the condition of non-equilibrium initial value or disturbance), the forecasting effect of the short-time (within 2 hours) precipitation is not good. Currently, business forecasting of short-term precipitation relies primarily on the extrapolation of radar returns.

At present, the short-term rainfall forecasting method based on radar echo extrapolation mainly adopts different technologies (such as cross-correlation tracking method TREC, COTREC based on continuous equation constraint, BTREC processed by Barnes spatial filtering, optical flow method and the like) to obtain the motion vector of radar echo. And under the condition that the echo intensity is not obviously changed in a short time, extrapolating the radar echo to obtain echo forecast values at different moments in 2 hours in the future. And finally, an empirical formula Z (aR) of the rainfall intensity R and the radar echo intensity Z is utilized^bAnd (namely Z-R relation) performing inverse calculation (inversion) to obtain radar precipitation at different moments in the future 2 hours, namely the forecasted precipitation.

Therefore, the short-time rainfall forecast precision based on the radar echo extrapolation is not only influenced by the radar echo extrapolation precision, but also influenced by the extrapolation echo rainfall inversion precision. Most of the previous radar precipitation forecasting methods focus on improving the extrapolation accuracy of radar echoes, such as TREC, COTREC, BTREC, optical flow method and the like, and most of the precipitation forecasting methods adopt a fixed Z-R relationship (that is, Z is 300R) for the precipitation inversion of extrapolated echoes^1.4) And (4) an algorithm. However, the coefficients a and b in the Z-R relationship are mainly determined by the characteristics of precipitation raindrops and vary with the weather system, the type of precipitation, the geographical location, the season, and other factors. Therefore, even in the radar returnUnder the condition of high wave extrapolation prediction precision, radar precipitation prediction values in different seasons in different areas are difficult to accurately obtain by using a fixed Z-R relation. In recent years, researchers also develop some Z-R relation algorithms with higher precision, such as dynamic Z-R relation algorithms and dynamic Z-R relation algorithms based on echo-peak high-level classification, based on the characteristic that coefficients a and b in the Z-R relation change with seasons, precipitation types, geographical positions, and the like, but these algorithms not only need to be fitted in real time by an optimization method, but also need additional forecast parameters (such as echo peaks and the like), and therefore, the algorithms are rarely used for radar-based short-time precipitation forecast.

Disclosure of Invention

Technical problem to be solved

The invention aims to overcome the defects of the prior art, and provides a short-time quantitative rainfall forecasting method based on extrapolation of echo intensity and echo top height.

(II) technical scheme

The technical scheme of the invention is as follows: a short-time quantitative precipitation forecast method based on echo intensity and echo top height extrapolation is used for radar-based short-time quantitative precipitation forecast. In order to achieve the purpose, the invention adopts the technical scheme that: radar base data processing, namely eliminating non-meteorological echoes in radar echo observation, and converting data from a polar coordinate format to a Cartesian rectangular coordinate format; extrapolating the radar echo, namely extrapolating the intensity of the radar echo and the height of the radar echo simultaneously to obtain predicted values of the intensity and the height of the radar echo; radar rainfall inversion, namely fitting Z-R relational expressions with different echo top heights based on real-time radar echo intensity, radar echo top height and rainfall observation data; and radar rainfall forecast, namely carrying out rainfall inversion on the echo intensity of the extrapolation forecast by utilizing Z-R relational expressions with different echo top heights, which are fit in real time, based on the forecast values of the radar echo intensity and the echo top heights, so as to obtain the short-time quantitative rainfall forecast.

The method specifically comprises the following steps:

s1: radar base data processing: and decoding, quality control and coordinate conversion are carried out on the radar base data to obtain radar echo data of Cartesian coordinates.

Step S2: and (3) radar echo extrapolation: based on gridded radar combined reflectivity data, a radar echo motion vector (namely a TREC vector) is obtained through a cross correlation Tracking (TREC) method, then Barnes space is used for filtering noise and incoordination in a heavier vector to obtain a BTREC vector, and finally a backward extrapolation scheme is used for extrapolating the radar echo combined reflectivity and the radar echo height to obtain the forecast values of the radar echo intensity and the echo height in 2 hours in the future.

Step S3: radar precipitation inversion (Z-R relationship based on echo-top height grading): and (4) interpolating the radar combined reflectivity and the radar echo height data subjected to the quality control processing and the gridding processing to a precipitation observation station by using a bilinear interpolation method. And classifying the radar combined reflectivity and the observed rainfall into different classes according to the echo peak height and the interval of 1km, wherein the echo peak height is in the range of 0-1km, 1-2km, 2-3km, 3-4km, 4-5km, 5-6km, 6-7km, 7-8km, 8-9km, 9-10km, 10-11km, 11-12km, 12-13km, 13-14km, 14-15km and more than 15km, and fitting the optimal Z-R relationship of the different classes by using an optimal method. The optimal method has the formulaWherein: n is the number of samples for observing precipitation, G_iFor the observed precipitation at the ith observation station,for radar inversion at the ith observation station, Z_iCombined reflectivity of radar for the ith observation station, a_j＝1,2，……，1200，b_k100,101, … …, 300, continuously adjusting a_jAnd b_kThe CTF value is minimized, at which time a_jAnd b_kThe corresponding Z-R relationship is the optimal Z-R relationship.

Step S4: radar precipitation forecast. Based on the extrapolated radar combined reflectivity and echo top height data of different times in the future 2 hours, the quantitative rainfall forecast values of different positions at different moments in the future 2 hours are calculated by using the Z-R relational expression of different echo top heights fitted (dynamically) at the current time.

In the technical solution of the present invention, the specific content of S1 includes:

s11: decoding radar base data: and decoding the radar base data according to the type of the radar (such as CIRADSA/SB/SC/CB/CC/CD) and the storage format of the base data to obtain radar echo data in polar coordinate formats at different elevation angles.

S12: controlling the quality of radar echo: firstly, using echo (reflectivity) vertical gradient RGDZ ═ W (Z-Z)_up) Eliminating abnormal clutter of ground objects in radar echo observation by more than or equal to 20dBZ, wherein: the weight coefficient W is equal to 1 when the radar slant range is 0-40km, and linearly decreases to 0 between 40-200km, Z is the radar echo (dBZ) of the current elevation angle, Z_upIs the radar echo (dBZ) at the previous elevation. Reuse of fuzzy logic algorithm P_xAnd eliminating isolated non-meteorological echoes, wherein N is equal to or less than 25 and less than or equal to 0.75, wherein N is the number of effective observations of the radar in a 5 multiplied by 5 window around x points on an azimuth radial plane (namely a scanning plane of the radar at a certain elevation angle).

S13: combined reflectance and echo top height calculation: based on the radar reflectivity data of polar coordinate format of different elevation angles after quality control processing, the combined reflectivity (the maximum value of different elevation angle layers) and the radar echo height (namely, the highest height of reflectivity factor of more than or equal to 18 dBZ) are calculated, and the calculation formula of the radar echo height (relative to a radar station) is h-Rsin theta + R²cos²θ/(2R_e) Where R is the radar detected slope distance, R_e8500km is the equivalent earth radius under the standard atmosphere, and theta is the highest elevation angle of the radar detection reflectivity factor of more than or equal to 18 dBZ.

S14: and (3) coordinate conversion of radar data: firstly, a Cartesian rectangular coordinate grid (shown as a dotted grid in figure 2) with 600 x 600 grid points and 1km x 1km horizontal resolution is constructed by taking a radar survey station as a center, and then a formula for calculating the horizontal distance of radar beams is calculatedWherein h is_rThe projection of the radar slope on the horizontal plane, L in fig. 2, is calculated for the altitude of the radar station. Finally, the radar data of the polar coordinate system (L, beta) is transferred to a Cartesian rectangular coordinate system through x ═ L · sin β and y ═ L · cos β, wherein β is the azimuth angle observed by the radar.

In the technical solution of the present invention, the specific content of S2 includes:

s21: and (5) calculating a TREC vector. Dividing the radar combined reflectivity data gridded at the time t (the number of grid points is 600 multiplied by 600, and the horizontal resolution is 1km multiplied by 1km) into 40 multiplied by 40 small regions without overlapping, wherein the number of grid points of each small region is 15 multiplied by 15, and converting the radar reflectivity data of each small region (initial small region) into a one-dimensional sequence Z in a mode of firstly arranging the X direction and then arranging the Y direction₁. Will Z₁Z consisting of any 15 x 15 grid points in radar combined reflectivity data of sequence and t + delta t (default 12 minutes)₂The sequence is subjected to spatial correlation analysis, and the correlation coefficient is calculated according to the formulaAnd the correlation number is the largest Z₂The small region corresponding to the sequence is used as a target region of the initial small region, and a vector from the center of the initial small region to the center of the target region is a TREC vector, which reflects the movement of the radar echo, as shown in fig. 3.

S22: and calculating the BTREC vector. Due to the small scale change of the echo and the tracking failure, some noise or uncoordinated vectors exist in the TREC vector, and the accuracy of the extrapolated echo is obviously influenced. The components of the TREC vector in the x direction and the y direction are respectively used as variables F (x, y) to be input into a formulaAndobtaining BTREC vectors (with noise removed and incoordination corrected)Component F₁(x, y). Wherein x and y are the positions of radar echoes in a Cartesian rectangular coordinate system and Gaussian parametersAndm is the number of lattice points of TREC vectors in a range of 100km from the (x, y) point, r_kIs point (x, y) and point (x)_k,y_k) The filter parameter G is 0.35 and C is 300.

S23: and (4) extrapolation of the radar echo. Firstly, a bilinear interpolation method is utilized to interpolate a BTREC vector with a horizontal resolution of 15km multiplied by 15km to a grid (1km multiplied by 1km) with radar combined reflectivity and radar echo top height, then a radar echo backward extrapolation method is utilized to extrapolate the radar combined reflectivity and the echo top height, and an extrapolation prediction value of the radar combined reflectivity and the radar echo top height at any time (such as 6 minutes) in the future 2 hours is obtained. The backward extrapolation of the radar echo is Z_i,j(t+Δt)＝Z_i-Δi,j-Δj(t), where Z is the radar combined reflectivity or the echo top height, i and j are the serial numbers of a certain grid point in the x and y directions respectively, Δ t is the extrapolation prediction time (e.g. 6 minutes, half an hour or 2 hours, etc.), Δ i ═ int (u · Δ t +0.5) is the moving distance of the grid point in the x direction in the prediction time, Δ j ═ int (v · Δ t +0.5) is the moving distance of the grid point in the y direction in the prediction time, and u and v are the components of the BTREC vector in the x and y directions respectively.

(III) advantageous effects

The invention has the advantages that: based on the common extrapolation of the radar echo intensity and the echo peak height, the echo peak height capable of reflecting the rising motion inside the cloud cluster is introduced into the radar rainfall forecast, and the accuracy of the short-time rainfall quantitative forecast is favorably improved.

Drawings

FIG. 1 is a flow chart of the method of the present invention;

FIG. 2 is a schematic diagram of a gridding of data of a radar polar coordinate;

FIG. 3 is a schematic diagram of TREC vector calculation.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Example 1

Referring to fig. 1-3, a short-term quantitative precipitation forecast method based on extrapolation of echo intensity and echo top height, as shown in fig. 1, includes the following steps:

s1: and (5) radar base data processing. Decoding, quality control and coordinate conversion are carried out on the radar base data to obtain radar echo data of Cartesian coordinates, and the specific content comprises the following steps:

s11: and decoding the radar base data. And decoding the radar base data according to the type of the radar (such as CIRAD SA/SB/SC/CB/CC/CD) and the storage format of the base data to obtain radar echo data in polar coordinate formats at different elevation angles.

S12: and controlling the quality of the radar echo. Firstly, using echo (reflectivity) vertical gradient RGDZ ═ W (Z-Z)_up) Eliminating abnormal clutter of ground objects in radar echo observation at a depth of more than or equal to 20dBZ, wherein the weight coefficient W is equal to 1 when the radar slant range is 0-40km, the weight coefficient W is linearly decreased to 0 between 40-200km, Z is the radar echo (dBZ) of the current elevation angle, and Z is the radar echo of the current elevation angle_upIs the radar echo (dBZ) at the previous elevation. Reuse of fuzzy logic algorithm P_xAnd eliminating isolated non-meteorological echoes, wherein N is equal to or less than 25 and less than or equal to 0.75, wherein N is the number of effective observations of the radar in a 5 multiplied by 5 window around x points on an azimuth radial plane (namely a scanning plane of the radar at a certain elevation angle).

S13: the calculation of the combined reflectivity and echo top height. Based on the radar reflectivity data of polar coordinate format of different elevation angles after quality control processing, the combined reflectivity (the maximum value of different elevation angle layers) and the radar echo height (namely, the highest height of reflectivity factor of more than or equal to 18 dBZ) are calculated, and the calculation formula of the radar echo height (relative to a radar station) is h-Rsin theta + R²cos²θ/(2R_e) Where R is the radar detected slope distance, R_e8500km is the equivalent earth radius under the standard atmosphere, and theta is the highest elevation angle of the radar detection reflectivity factor of more than or equal to 18 dBZ.

S14: and (5) converting the coordinates of the radar data. Firstly, a Cartesian rectangular coordinate grid with 600 x 600 grid points and a horizontal resolution of 1km x 1km (shown as a dotted grid in figure 2, wherein the dotted grid is a Cartesian coordinate system, a thick elliptic solid line is radar polar coordinates, x and y are projections of radar polar coordinate observation on the Cartesian coordinates, A is a radar observation point, L is a projection of radar slant distance on a plane, and beta is an azimuth angle observed by the radar) is constructed by taking a radar survey station as a center, and then a radar beam horizontal distance calculation formula is used for calculating the horizontal distance of the radar beamWherein h is_rThe projection of the radar slope on the horizontal plane, L in fig. 2, is calculated for the altitude of the radar station. Finally, the radar data of the polar coordinate system (L, beta) is transferred to a Cartesian rectangular coordinate system through x ═ L · sin β and y ═ L · cos β, wherein β is the azimuth angle observed by the radar.

S2: and (4) extrapolation of the radar echo. Based on gridded radar combined reflectivity data, a radar echo motion vector (namely a TREC vector) is obtained through a cross correlation Tracking (TREC) method, then noise and incoordination in a heavier vector are filtered through Barnes space, a BTREC vector is obtained, finally, a backward extrapolation scheme is used for extrapolating the radar echo combined reflectivity and the radar echo top height, and forecast values of the radar echo intensity and the echo top height in 2 hours in the future are obtained, and the specific content comprises the following steps:

s21: and (5) calculating a TREC vector. Dividing the radar combined reflectivity data gridded at the time t (the number of grid points is 600 multiplied by 600, and the horizontal resolution is 1km multiplied by 1km) into 40 multiplied by 40 small regions without overlapping, wherein the number of grid points of each small region is 15 multiplied by 15, and converting the radar reflectivity data of each small region (initial small region) into a one-dimensional sequence Z in a mode of firstly arranging the X direction and then arranging the Y direction₁. Will Z₁Arbitrary 15 × 15 in the radar combined reflectance data of the sequence and time t + Δ t (default 12 minutes)Z consisting of lattice points₂The sequence is subjected to spatial correlation analysis, and the correlation coefficient is calculated according to the formulaAnd the correlation number is the largest Z₂The small region corresponding to the sequence is used as a target region of the initial small region, and a vector from the center of the initial small region to the center of the target region is a TREC vector, which reflects the movement of the radar echo, as shown in fig. 3.

S22: and calculating the BTREC vector. Due to the small scale change of the echo and the tracking failure, some noise or uncoordinated vectors exist in the TREC vector, and the accuracy of the extrapolated echo is obviously influenced. The components of the TREC vector in the x direction and the y direction are respectively used as variables F (x, y) to be input into a formulaAndcomponent F of the BTREC vector (noise eliminated and mismatch corrected) is obtained₁(x, y). Wherein x and y are the positions of radar echoes in a Cartesian rectangular coordinate system and Gaussian parametersAndm is the lattice point number of TREC vectors in a range of 100km from the (x, y) point, r_kIs point (x, y) and point (x)_k,y_k) The filter parameter G is 0.35 and C is 300.

S23: and (4) extrapolation of the radar echo. Firstly, a bilinear interpolation method is utilized to interpolate a BTREC vector with a horizontal resolution of 15km multiplied by 15km to a grid (1km multiplied by 1km) with radar combined reflectivity and radar echo top height, then a radar echo backward extrapolation method is utilized to extrapolate the radar combined reflectivity and the echo top height, and an extrapolation prediction value of the radar combined reflectivity and the radar echo top height at any time (such as 6 minutes) in the future 2 hours is obtained. Radar echoBackward extrapolation of Z_i,j(t+Δt)＝Z_i-Δi,j-Δj(t), where Z is the radar combined reflectivity or the echo top height, i and j are the serial numbers of a certain grid point in the x and y directions respectively, Δ t is the extrapolation prediction time (e.g. 6 minutes, half an hour or 2 hours, etc.), Δ i ═ int (u · Δ t +0.5) is the moving distance of the grid point in the x direction in the prediction time, Δ j ═ int (v · Δ t +0.5) is the moving distance of the grid point in the y direction in the prediction time, and u and v are the components of the BTREC vector in the x and y directions respectively.

S3: radar precipitation inversion (Z-R relationship based on echo-top height grading). And (4) interpolating the radar combined reflectivity and the radar echo height data subjected to the quality control processing and the gridding processing to a precipitation observation station by using a bilinear interpolation method. And classifying the radar combined reflectivity and the observed rainfall into different classes according to the echo peak height and the interval of 1km, wherein the echo peak height is in the range of 0-1km, 1-2km, 2-3km, 3-4km, 4-5km, 5-6km, 6-7km, 7-8km, 8-9km, 9-10km, 10-11km, 11-12km, 12-13km, 13-14km, 14-15km and more than 15km, and fitting the optimal Z-R relationship of the different classes by using an optimal method. The optimal method has the formulaWherein n is the number of samples for observing precipitation, G_iFor the observed precipitation at the ith observation station,for radar inversion at the ith observation station, Z_iCombined reflectivity of radar for the ith observation station, a_j＝1,2，……，1200，b_k100,101, … …, 300, continuously adjusting a_jAnd b_kThe CTF value is minimized, at which time a_jAnd b_kThe corresponding Z-R relationship is the optimal Z-R relationship.

S4: radar precipitation forecast. Based on extrapolated radar combined reflectivity and echo top height data of different times in the future 2 hours, quantitative rainfall forecast values of different positions at different moments in the future 2 hours are calculated by using a Z-R relational expression of different echo top heights fitted (dynamically) at the previous time.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.