CN105891833A - Method of identifying warm cloud precipitation rate based on Doppler radar information - Google Patents

Abstract

The invention provides a method of identifying a warm cloud precipitation rate based on Doppler radar information. The method is characterized in that a radar echo intensity ref vertical profile VPR identification module is used to identify a warm cloud; a warm cloud identification module is used to establish a polar coordinate in a cache region according to the radar echo intensity, and is used to identify the precipitation rate of the warm cloud of the stratus cloud and the precipitation rate of the warm cloud of the convection cloud according to the ref interpolation and the wet bulb temperature; a processing module is used to calculate the precipitation rate. The method is advantageous in that the precipitation estimation error can be reduced effectively, the scientific concrete calculation method is provided, the high-quality data is provided, and the property loss of the people can be effectively reduced.

Description

Method based on Doppler radar information identification warm cloud precipitation rate

Technical field

The present invention relates to process and the method for estimation of a kind of weather information, particularly relate to a kind of based on how general Strangle radar information identification stratus, convective cloud and the method for warm cloud precipitation.

Background technology

The detection of precipitation rate (I) (rain condition, unit millimeter is per hour) is in meteorological, the hydrology and geological disaster Very important effect is had in early warning；It addition, by energy in precipitation rate Data Assimilation to numerical weather forecast The accuracy rate of significant improvement numerical weather forecast.

The method utilizing radar reflectivity estimation precipitation rate in current operation is to directly utilize radar return reflection The rate factor (Z, unit mm⁶/m³) and precipitation rate (I, unit mm/h) between there is power exponent positive correlation Empirical relation (Z-I relation), i.e. Z-I relation is Z=300I^1.4.But utilize this calculated rainfall of Z-I relation Larger difference is had, mainly by caused by the improper use of Z-I relation with actual ground rainfall.Thus to difference Cloud-type, if the Z-I relation that toboggan method uses U.S. WSR-88D radar acquiescence completely carries out precipitation Calculate, then the precipitation that precipitation and pluviometer record there will be larger difference, if using different Z-I relation carries out calculating will be effectively reduced Precipitation estimation error.

Need badly the most at present and a kind of can rationally detect precipitation rate (I) according to different Z-I relations Method.

Summary of the invention

It is an object of the invention to be to solve the deficiencies in the prior art, it is provided that a kind of based on Doppler radar Data stratus, convective cloud and warm cloud precipitation recognition methods, comprise the steps:

Radar measuring echo strength ref；

Whether VPR identification module exists according to echo strength ref Vertical Profile identification warm cloud；

Warm cloud identification module sets up polar coordinate, warm cloud identification mould according to radar echo intensity ref at buffer area Tuber is according to described polar ref interpolation and wet bulb temperature, it is judged that whether warm cloud exists；If warm cloud is deposited , then warm cloud identification module utilizes 3D-Barnes interpolation algorithm to calculate ref and wet bulb in rectangular coordinate system Temperature, the precipitation rate of the warm cloud in identification stratus, the precipitation rate of the warm cloud in identification convective cloud；

Processing module calculates the step of precipitation rate, this step include processing module according to echo strength ref and Ref=10Lg (Z) calculates radar return reflectivity factor Z, and processing module is according to the convection current of ref, ref Cloud probit P_conv, wet bulb temperature judge cloud type as: convective cloud, stratiform clouds or torrid zone warm cloud, If cloud layer is convective cloud, then use Z=300I^1.4Output precipitation rate I, if cloud layer is stratiform clouds, then uses Z=200I^1.6Output precipitation rate I, if cloud layer is warm cloud, then precipitation uses Z=30.7I^1.66 Output precipitation rate I.

The invention has the beneficial effects as follows: 1. can effectively reduce Precipitation estimation error；2. offer section learning aid The computational methods of body, it is thus possible to identify convective cloud and stratiform clouds definitely；3. for offer height of preventing and reducing natural disasters The data of quality, thus effectively alleviate the property loss of the people.

Accompanying drawing explanation

Fig. 1 is that the precipitation rate of present invention method based on Doppler radar information identification warm cloud precipitation rate divides Class flow chart.

Fig. 2 be the present invention be 12 each process total error curve charts of Precipitation Process.

Fig. 3 is that the present invention adds up the total error curve chart that there is warm cloud precipitation lattice point in 12 Precipitation Process.

Fig. 4 is that the present invention three groups tests the precipitation error curve diagram of radar estimation per hour.

Fig. 5 (a) is the scatterplot comparison diagram of the process 1 of the embodiment of the present invention 1.

Fig. 5 (b) is the scatterplot comparison diagram of the process 3 of the embodiment of the present invention 1.

Fig. 5 (c) is the scatterplot comparison diagram of the process 6 of the embodiment of the present invention 1

Fig. 5 (d) is the scatterplot comparison diagram of the process 7 of the embodiment of the present invention 1.

Fig. 6 (a) is to obtain convective cloud and stratus scattergram by fuzzy logic algorithm.

Fig. 6 (b) is cloud scattergram when convective cloud, stratus and three kinds of type of precipitation of warm cloud.

Fig. 6 (c) is that around radar, in the range of 20-80km, echo strength is more than all layers of 10dBZ The schematic diagram of the VPR of cloud point.

Fig. 6 (d) is the scatterplot of radar quantitative predication precipitation intensity and encryption precipitation station observation precipitation intensity Comparison diagram.

Fig. 7 (a) obtains convective cloud and stratus scattergram by fuzzy logic algorithm；

Cloud scattergram when Fig. 7 (b) convective cloud, stratus and three kinds of type of precipitation of warm cloud；

The echo strength all stratus points more than 10dBz in the range of 20-80km around Fig. 7 (c) radar The schematic diagram of VPR；

Fig. 7 (d) radar quantitative predication precipitation intensity contrasts with the scatterplot of encryption precipitation station observation precipitation intensity Figure；

Fig. 8 is the structural representation of present invention method based on Doppler radar information identification warm cloud precipitation rate Figure.

Detailed description of the invention

In conjunction with the drawings and specific embodiments, claimed technical solution of the invention is made the most in detail Explanation.

Introduce a kind of radar based on Assimilate Doppler Radar Data observation of the present invention below in conjunction with accompanying drawing 1～8 to return Intensity of wave ref (unit dBZ) identifies the technical scheme bag of the method for stratus, convective cloud and warm cloud precipitation rate Include following steps:

Radar measuring echo strength ref；

Whether VPR identification module exists according to echo strength ref Vertical Profile identification warm cloud；

Warm cloud identification module sets up polar coordinate, warm cloud identification mould according to radar echo intensity ref at buffer area Tuber is according to described polar ref interpolation and wet bulb temperature, it is judged that whether warm cloud exists；If warm cloud is deposited , then warm cloud identification module utilizes 3D-Barnes interpolation algorithm to calculate the ref in rectangular coordinate system, and profit With wet bulb temperature, the precipitation rate of the warm cloud in identification stratus, the precipitation rate of the warm cloud in identification convective cloud； Wherein, 3D-Barnes interpolation algorithm is:

The fundamental formular of 3-dimensional Barnes interpolation is:

$f^a\left(r_k\right)=\frac{\underset{i}{\Σ}{\mathrm{\ω}}_i{\mathrm{ref}}^o\left(r_i\right)}{\underset{i}{\Σ}{\mathrm{\ω}}_i}$

Here variable ref is radar echo intensity, and subscript a and o represents amount of analysis (lattice point after interpolation respectively On value) and observed quantity, r_kIt is the coordinate of mesh point k, r_iIt is the observation in observation station i, ω_iIt it is weight Coefficient, is defined by the formula:

${\mathrm{\ω}}_i=\exp \[-({r_h}^2/2{R_h}^2+{r_v}^2/2{R_v}^2)\]$

Wherein r_hIt is the horizontal range between observation station i and mesh point k, r_vIt is vertical dimension, R_h、R_vRespectively Be the prespecified horizontal and vertical radius of influence (Rh take 5 HORIZONTAL PLAIDs away from, Rv take 3 vertical lattice away from). 3D-Barnes scheme is a kind of Weighted Average Algorithm, has extension function, can filter part measurement noise, And calculate simplicity.But the selection of interpolation result and the uniformity of surrounding observation station and the radius of influence has relatively Important Relations.

Processing module calculates the step of precipitation rate, this step include processing module according to echo strength ref and Ref=10Lg (Z) calculates radar return reflectivity factor Z, and processing module is according to the convection current of ref, ref Cloud probit P_conv, wet bulb temperature judge cloud type as: convective cloud, stratiform clouds or torrid zone warm cloud, If cloud layer is convective cloud, then use Z=300I^1.4Output precipitation rate I, if cloud layer is stratiform clouds, then uses Z=200I^1.6Output precipitation rate I, if cloud layer is warm cloud, then precipitation uses Z=30.7I^1.66 Output precipitation rate I.

The invention has the beneficial effects as follows: 1. can effectively reduce Precipitation estimation error；2. offer section learning aid The computational methods of body, it is thus possible to identify convective cloud and stratiform clouds definitely；3. for offer height of preventing and reducing natural disasters The data of quality, thus effectively alleviate the property loss of the people.

The concrete steps of the present invention see Fig. 1,8:

S111, according to VIL, (Vertical cumulative liquid water content, vertically adds up liquid Water content) simply the observation station that radar body is swept in coordinate is divided into stratus and non-stratus, if VIL < 6.5 kg/m²Then it is considered stratus；

Wherein, the computing formula of VIL is:

VIL=Σ 3.44 × 10^-6×[(Z_i+Z_i+1)/2]^4/7×Δh_i

Wherein, VIL unit is kg/m², hanging down in the height layer of effective reflectivity factor value can be detected for radar Straight accumulation Liquid water content, is to be all to be caused by aqueous water by supposing the return of all reflectivity factor, The liquid aqueous content of vertical accumulation beyond this height layer is ignored, reflectivity factor Z_i,Z_i+1Unit be mm⁶/m³,Δh_i=500m.

S112, set up polar coordinate, VPR according to radar echo intensity at buffer area by warm cloud identification module The echo strength all stratus points more than 10dBZ in the range of 20-80km around identification module Discrimination Radar Average VPR；

Here around selection radar, in the range of 20-80km, the reason of echo strength is: when using VPR In order to ensure the reliable of measurement data and avoid radar shadown to limit the scope of observation, if scope Expanding, radar resolution in vertical direction then can decrease and then affect the accuracy of identification again.

S121, processing module judge whether the step of bright band:

S1211, obtained 0 DEG C of isothermal line height h_0 DEG C, and from h_0 DEG C of overhead 500 by WRF simulation Maximum echo strength Z in VPR is found downwards at m_peakAnd height h_peak.Here parameter 500m It is to obtain according to model predictions error estimation；

S1212, from h_peakZ is upwards searched at highly place_peakEcho strength Z when reducing 10%_topAnd it is high Degree h_top；From h_peakZ is searched for downwards at highly place_peakEcho strength Z when reducing 10%_bottomAnd it is high Degree h_bottom；

If S1213 meets following condition and thinks and there is bright band simultaneously:

h_top-h_bottom≤D₀ (4.1)

h_top-h_peak≤D₁ (4.2)

h_peak-h_bottom≤D₁ (4.3)

Wherein, D₀And D₁It is to become with radar scanning mode and the vertical resolution of reflectivity factor Change, respectively analyze an example, D in literary composition in conjunction with this chapter₀And D₁Take 1.5km and 1km respectively.

S1214 is if there is bright band, bright band heights of roofs BB_tWith height BB at the bottom of bright band_bBe calculated as follows:

${\mathrm{BB}}_t=\left\{\begin{array}{cc}{h}_{top},& h_{top}-h_{peak}\&le;D_t\\ h_{peak}+D_{top},& h_{top}-h_{peak}>D_t\end{array}\right.---\left(4.4\right)$

${\mathrm{BB}}_b=\left\{\begin{array}{cc}{h}_{bottom},& h_{peak}-h_{bottom}\&le;D_b\\ h_{peak}-D_{bottom},& h_{peak}-h_{bottom}>D_b\end{array}\right.---\left(4.5\right)$

Wherein, D_tAnd D_bFor adjusting at the bottom of bright band top and bright band, value is respectively and h_peakWith h_bottomAnd h_peak With h_topBetween slope relevant, take default value D_t=0.5km, D_b=0.7km.

If it addition, the echo strength value at height at the bottom of bright band is less than 28dBZ, then again from h_peakHigh Search for downwards minimum altitude during echo strength value >=28dBZ at degree, and its place height is made At the bottom of bright band, this purpose is to avoid the too much correction of bright band.But this parameter value is led sometimes Cause bright band and correct inadequate or excessive.

S131, then correcting bright band if there is bright band, the method correcting bright band is as follows:

If there is effective BB_tAnd BB_bValue and the echo strength Z of correspondence_BtAnd Z_Bb, and BB_b<h_peak, BB_t>h_peak, then BB is calculated_tAnd BB_bEcho strength slope between height:

S=(Z_Bt-Z_Bb)/(BB_t-BB_b) (4.6)

It is pointed to BB_tAnd BB_bBetween certain height h_iAll echo strengths assignment again, again after assignment Echo strength is:

Z_i=Z_Bt-S×(BB_t-h_i) (4.7)

If there is effective BB_bValue, and there is not effective BB_tValue, and BB_b<h_peak, then BB_bAll echo strength values of level above are more than Z_BbValue be all assigned to Z_Bb。

S141, recognized whether warm cloud VPR by VPR identification module, thus judge warm cloud precipitation There is a possibility that.Recalculate around radar echo strength in the range of 20-80km after correcting bright band to be more than The average VPR of all stratus points of 10dBZ.If bright band exist, then to bottom bright band to VPR Bottom part least square fitting VPR, if bright band does not exists, from h_0 DEG C height to Part bottom VPR least square fitting VPR.The body calculating this moment and front 4 moment is swept VPR slope, if meeting slope≤0 of wherein 3, it is believed that there is warm cloud VPR, i.e. there is warm cloud , otherwise it is assumed that there is not warm cloud precipitation in the probability of precipitation.It is noted that when echo strength is more than Think during 50dBZ and there is hail, do not carry out Precipitation estimation.

S200, set up polar coordinate according to radar return echo strength at buffer area by warm cloud identification module, And according to above-mentioned polar ref interpolation and temperature, it is judged that whether warm cloud exists.If there is further Utilize the ref in rectangular coordinate system and temperature, identify the precipitation rate of the warm cloud in stratus, identify convective cloud In the step of precipitation rate of warm cloud；

S210, by warm cloud identification module according to radar return reflectivity factor buffer area set up pole sit Mark, sweeps radar body observational data and is interpolated into etc. on the normal grid lattice point of longitude and latitude.

Analyzed area center of the present invention is radar site, 301 × 301 lattice points of horizontal direction, lattice away from for 0.01 °, vertical direction takes 73 layers, and lattice are away from for 0.25km.The fuzzy logic algorithm proposed due to Yang etc. is examined Consider the three-dimensional configuration feature of reflectivity factor distribution, can reasonably identify major part convective cloud and stratiform clouds, The present invention utilizes this algorithm to carry out convective cloud and stratiform clouds identification.Simple introduction about fuzzy logic algorithm As follows.

S211, first warm cloud identification module detect each identification parameter.Here identification parameter has four: The echo ref that 2km highly locates_wrkAnd standard deviation std, echo high and 2km highly locate the product of echo p_ztop, vertical accumulative Liquid water content VIL.

Wherein, echo high be radar echo intensity be the maximum height of 18.3dBZ.VIL is by supposing The return of all reflectivity factor is all to be caused by aqueous water, specifically calculates reference formula (2.10).

The membership function u of S212, warm cloud identification module structure cloud identification_k,e(x), its function expression is as follows:

${\mathrm{\&mu;}}_{k,C}(x,a,b)=\left\{\begin{array}{cc}0,& x\&le;a\\ (x-a)/(b-a),& a<x<b\\ 1,& x\&GreaterEqual;b\end{array}\right.,(k=1,2,...,N)---\left(4.8\right)$

${\mathrm{\&mu;}}_{k,S}(x,a,b)=\{\begin{array}{cc}1,& x\&le;a\\ (b-x)/(b-a),& a<x<b\\ 0,& x\&GreaterEqual;b\end{array},(k=1,2,...,N)---\left(4.9\right)$

Wherein, k represents identification parameter, totally 4 identification parameters, so taking N=4；E is variable element, e Taking C and represent convective cloud, e takes S and represents stratiform clouds；X is each identification parameter numerical value；Different identification parameters pair Reference thresholds a answered, b is different.For ref_wrk, a=20dBZ, b=45dBZ；For std, a=1 DBZ, b=14dBZ；For p_ztop, a=100km dBZ, b=500km dBZ；For VIL, a=0.5 kg/m², b=5.0kg/m²；Using above-mentioned identification parameter as input variable, i.e. obfuscation is input to be become Amount is converted into Fuzzy dimension in the way of membership function, and its value excursion is [0,1], and its expression formula is as follows

μ_k,S(x, a, b)=1-μ_k,C(x,a,b) (4.10)

S213, warm cloud identification module utilizes following expression to seek the conditional probability of stratiform clouds and convective cloud:

$P_e=\frac{\underset{k=1}{\overset{N}{\&Sigma;}}{w}_k\&CenterDot;P_{k,e}}{\underset{k=1}{\overset{N}{\&Sigma;}}{w}_k}---\left(4.11\right)$

Wherein, P_k,e=μ_k,e(X_k), e represents stratiform clouds or convective cloud；w_kWeight system for each identification parameter Number, weight coefficient is usually and is given by policymaker, but, policymaker is often difficult to or at all cannot be true The exact value of each target weight fixed；On the other hand, though policymaker can not provide a weight determined, But a scope substantially can be given.The method of conventional determination weight is a lot, but all with denseer Subjective colo(u)r, in some cases, subjectivity determines that weight still has objective one side, reacts to a certain extent Practical situation.Yang etc. utilize historical data to add up each characteristic parameter and the relation of precipitation classification, and Find out their general contribution degree so that identification parameter weight w_kAll take same value；N is identification parameter Number.It can be seen that Pc Yu Ps sum is equal to 1, it is possible to represent it is the probability of convective cloud with Pc, I.e. Pc >=0.5 is considered convective cloud, otherwise it is assumed that be stratiform clouds.

S214, probability if there is warm cloud precipitation, and lattice point meets the echo that 1km highly locates Intensity level is more than 2 DEG C more than 25dBZ earth's surface wet bulb temperature simultaneously, it is believed that be warm cloud precipitation lattice point, no Then think it is non-warm cloud precipitation lattice point.Earth's surface wet bulb temperature W_BcBe calculated as follows:

W_Bc=(0.00066 × P × T_c+4098×E/T_dc×(237.7+T_dc)²)

/(0.00066×P+4098×E/(237.7+T_dc)²) (4.12)

Wherein,T_dcIt is dew point wet bulb temperature, unit DEG C；P is air pressure, Unit mb；T_cIt is temperature, unit DEG C.

Accordingly, then stratus precipitation being divided into Stratiform Cloud Precipitation and warm cloud precipitation, it is right to be divided into by Convective Cloud Precipitation Stream cloud precipitation and warm cloud precipitation.

S300, processing module calculate the step of precipitation rate detection, and this step includes that processing module is according to radar Echo reflection rate factor Z (unit: mm⁶.m^-3), and relation ref=10log (Z) of echo strength ref) right Stream cloud precipitation uses Z=300I^1.4And exporting precipitation rate I (unit mm/h), Stratiform Cloud Precipitation uses Z=200I^1.6And exporting precipitation rate I, torrid zone warm cloud precipitation uses Z=30.7I^1.66 And export precipitation rate I.It specifically comprises the following steps that

The present invention devises three groups of tests.

The single Z-I relation of the Doppler radar acquiescence of test one (default), only applied code is estimated Precipitation, i.e. Z=300I^1.4；

Test two (2cls), are divided into Convective Cloud Precipitation and stratus precipitation with fuzzy logic algorithm by precipitation, then Precipitation is estimated according to each self-corresponding Z-I relation；

Test three (3cls), the method identification warm cloud precipitation proposed by Xu etc., it is divided into warm cloud to drop precipitation Water, Convective Cloud Precipitation and stratus precipitation three types altogether, then estimates fall according to each self-corresponding Z-I relation Water.Wherein Convective Cloud Precipitation uses Z=300I^1.4, Stratiform Cloud Precipitation uses Z=200I^1.6, torrid zone warm cloud precipitation Use Z=30.7I^1.66.It is emphasized that: in order to get rid of the impact of non-precipitation echo and hail, thunder as far as possible Reach data when carrying out Precipitation estimation through quality control.

It addition, Z-I relation estimates that another subject matter in precipitation is radar reflectivity factor and ground The space discordance of precipitation intensity.Owing to air is to the refraction of radar wave (ultrashort wave) and earth curvature Impact, if estimating precipitation with PPI scanning information, actually the data on differing heights is pressed Same Z-I relation estimates precipitation.When detecting remote precipitation target, even if the elevation angle is the lowest, its The height of sampling volume still has a few km, and its intensity in dropping process of the precipitation on this height just has can Can change, as evaporation cause the reduction of precipitation intensity, water become the condensation of thing sublimate growth and cohesion and Touch and increase and all Precipitation estimation can be caused error, thus cause between radar estimated value and Land Surface Temperatures Significant difference.This chapter combines local geography and height of radar antenna, the Z highly located with 1km Estimate precipitation, the error that space discordance is brought can be reduced, and can reduce owing to bright band is corrected not exclusively The impact brought.

Embodiment 1:

See Fig. 2,3, in order to obtain conclusion more accurately, the present invention has added up Hefei in June, 2010 To the radar quantitative predication precipitation in whole summer in August, and by error analysis, each test is carried out effect Assessment.

Radar Data used by the present invention takes from June, 2010 to August Hefei Doppler radar every 6 minutes one Secondary swept-volume data, observation precipitation data takes from Anhui Province's encryption precipitation station.0 DEG C of isothermal line height H_0 DEG C is obtained by the simulation of pattern WRF3.5.1 with earth's surface wet bulb temperature, its just border data employing 2013 The FNL data in June to August in year, its horizontal resolution is 1 ° × 1 °.

Radar quantitative predication precipitation 115.758 ° of region E～118.758 ° of E and 30.367 ° of N～33.367 ° of N, Horizontal direction is 301 × 301 lattice points, and lattice are away from for waiting longitude and latitude 0.01 °, and vertical direction takes 73 layers, lattice Away from for 0.25km, survey region center is radar position, Hefei.

For the radar quantitative precipitation estimation effect of qualitative assessment different tests, have selected here average deviation, Mean absolute error rate and root-mean-square error are as statistical indicator.Wherein, average deviation (bias) definition As follows:

$bias=\frac{1}{n}\underset{i=1}{\overset{n}{\&Sigma;}}\left(R_a\right(i)-R_g(i\left)\right)---\left(4.13\right)$

In order to reflect the overall precision of radar pinch-reflex ion diode field, define the average of absolute relative error, letter Title mean absolute error rate (rae):

$rae=\frac{\underset{i=1}{\overset{n}{\&Sigma;}}|R_a\left(i\right)-R_g\left(i\right)|}{\underset{i=1}{\overset{n}{\&Sigma;}}{R}_g\left(i\right)}\&times;100\%---\left(4.14\right)$

In order to reflect the radar pinch-reflex ion diode field estimation result to heavy rain, use root-mean-square error (rmse):

$rmse=\sqrt{\frac{1}{n}\underset{i=1}{\overset{n}{\&Sigma;}}{\left(R_a\right(i)-R_g(i\left)\right)}^2}---\left(4.15\right)$

The most various middle R_aI () is the radar quantitative predication precipitation on the i-th website, R_gI () is on the i-th website Pluviometer observation, n be precipitation website sum.

It can be seen that bias value is closer to 0 explanation radar estimates that precipitation more connects with precipitation station pinch-reflex ion diode Closely；The overall precision of rae value the least explanation radar pinch-reflex ion diode field is the highest；Rmse value the least explanation radar Estimation to heavy rain part is the most accurate.

In order to see that dewatering effect estimated by different tests radar concisely, table 4.1 gives 2010 Test on June to August three groups Precipitation Process total error is analyzed, it can be seen that bias, rae, rmse respectively miss Difference statistical indicator all shows, test three is not only the highest to the estimation precision of overall precipitation field, and to heavy rain The estimation of part is the most accurate, tests three radars and estimates that dewatering effect is better than testing one and testing two.Work as observation When precipitation intensity is less than 2mm/h, radar estimates that precipitation is relatively big, so that affecting whole with observation precipitation error The recruitment evaluation of precipitation field, during so radar precipitation estimation difference is analyzed herein, eliminates observation precipitation strong The degree website less than 2mm/h.

Table in June, 4.1 2010 to August Precipitation Process total error is analyzed

The error analysis of table 4.1 comprises the error analysis that all type of precipitation are total, in order to further appreciate that examination Testing three and estimate that the effect of precipitation is distinguished with test one and test two radar, table 4.2 gives again 2010 6 Month to all total error analyses that there is warm cloud precipitation lattice point in August, if the most a certain lattice point does not exist warm cloud Error statistics is no longer carried out during precipitation.By table it can be seen that bias, rae, rmse each error statistics index All show that testing three effects is substantially better than test one and test two, especially tests three bias values significantly lower than examination Test one and test two, illustrating that test three radars estimate that precipitation are with precipitation station pinch-reflex ion diode closely.It addition, The overall precision of rae value explanation test three radar pinch-reflex ion diode fields is the highest；The explanation test three of rmse value is to greatly The estimation of rain part is also the most accurately.Contrast table 4.1 and table 4.2, illustrate that the classification of warm cloud precipitation makes thunder Reach quantitative predication dewatering effect and have significant raising.

Table in June, 4.2 2010 to August Precipitation Process has the website total error analysis of warm cloud precipitation

In order to be better understood upon each Precipitation Process of in June, 2010 to August, and each group of test is described Dewatering effect estimated by radar, and according to rainy persistent time, 3 months precipitation is divided into 12 Precipitation Process, as Shown in table 4.3, its cumulus mediocris cloud mixed type precipitation refers to convective cloud and Stratiform Cloud Precipitation in a Precipitation Process Exist simultaneously.Fig. 2 is 12 each process total error curve charts of Precipitation Process.From figure 2 it can be seen that Except process 3, remaining process is all that test three radars estimate that dewatering effects are better than testing one and test two, especially Its process 1 and process 6 and process 7, test three effects and be substantially better than test one and test two.

4.3 12 Precipitation Process introductions of table

The same with Fig. 2, Fig. 3 is the total error song that there is warm cloud precipitation lattice point in 12 Precipitation Process of statistics Line chart, compares Fig. 2, eliminates the lattice point that there is not warm cloud precipitation during statistical error.From Fig. 3 permissible Finding out, for bias value, except process 3, the bias value of remaining process, all near 0, illustrates when fall When water process exists warm cloud precipitation, testing three radars and estimate precipitation with observation precipitation closely, effect is excellent In test one and test two.For rae and rmse value, except process 3 and process 4, the value of remaining process Test three closer to 0, shows to test three and is better than the estimation effect of overall Precipitation estimation and heavy rain part Test one and test two.

Choose radar and estimate that the preferable process of dewatering effect 1 and process 6 and process 7, alternative take estimation As a example by bad process 3, carry out labor.

Fig. 4 be three groups test per hour radar estimate precipitation error curve diagram (the most often row represent: Process 1, process 3, process 6, process 7；Each column represents from left to right: bias, rae, rmse；Horizontal Coordinate: rainy persistent time (unit: h)；Vertical coordinate: each error amount；Solid line: test one；Long void Line: test two；Short dash line: test three)；It can be seen that Fig. 4 with Fig. 2 conclusion is consistent, for process 1, process 6, process 7, test three not only total error is minimum, and Precipitation estimation error is the most per hour Little；For process 3, indivedual moment test three each errors on the contrary more than test one and test two, so that its Test three total errors maximum.Analyze its reason, find that this Precipitation Process is mainly by scattered convection cell Formed, either precipitation scope or rainy persistent time all ratios are relatively decentralized, precipitation off and on, every time Rainy persistent time is less than 6h, and the precipitation station detecting precipitation per hour is the most few.Another one is former Cause, may be relevant with the Z-I relation chosen, because Z-I relation estimates that precipitation is obtained by statistics, Relevant with the geographical position of selected data and seasonal climate, for this problem, the later stage continues to relevant real Time be suitable for local Z-I relation research dynamically.

Fig. 5 is each process encryption precipitation station observation precipitation intensity and different tests radar quantitative predication precipitation Intensity scatterplot comparison diagram.Owing to whole Precipitation Process website is a lot, scatterplot cannot be seen clearly divide if being all given Cloth, so in addition to the precipitation intensity scatterplot comparison diagram that the 3rd process is 4h, other 3 processes are all It it is 6h precipitation intensity scatterplot comparison diagram.Fig. 5 (a) is the scatterplot comparison diagram of process 1, and this process is the strongest Precipitation intensity is 14mm/h, and three groups of tests estimate that precipitation is corresponding with observation precipitation very well, especially test three Estimate that precipitation is generally within best-fitting straight line both sides.When precipitation intensity is less, especially less than 2mm/h Time, radar estimation difference is the biggest.Fig. 5 (b) is the scatterplot comparison diagram of process 3, this time Precipitation Process without Opinion be Annual distribution or rainfall distribution the most relatively decentralized, almost without continuous 6h precipitation event, so choosing Its 4h precipitation intensity carries out scatterplot contrast.For this process, although test three-phase is than test one and test two Improve the problem of underestimating of radar quantitative predication precipitation to a certain extent, but some point but occurs in that bright Aobvious over-evaluates.It addition, test two and test one estimation precipitation result are similar, illustrate that this process is divided into two It is mainly Convective Cloud Precipitation during class type of precipitation, is consistent with table 4.3 result.Fig. 5 (c) is process 6 Scatterplot comparison diagram, this Precipitation areal extent is relatively wide, and a lot of websites all this time precipitation, precipitation is strong Degree reaches 70mm/h, and three groups of tests estimate that precipitation is corresponding with observation precipitation very well, though test three still has certain Underestimate, but compare that front two groups of tests significantly improve radar quantitative predication precipitation underestimate problem.Fig. 5 D () is the scatterplot comparison diagram of process 7, this process precipitation intensity reaches 60mm/h, same and observation Precipitation is compared, and precipitation all can be well estimated in three groups of tests, and especially test three estimation dewatering effect is the most excellent In front two groups of tests.

Embodiment 2

The method that precipitation is divided into warm cloud, stratus and Convective Cloud Precipitation three class that this chapter proposes, for when fall When aqua region and precipitation time change, radar quantitative predication dewatering effect how？To this end, this section Additionally choose twice precipitation example, choose Anhui Province's Heavy Precipitation inspection 21 to 23 July in 2009 Test the same radar quantitative predication dewatering effect at different periods；Choose on June 11st, 2008 to 13 The different radar of day Liuzhou Heavy Precipitation inspection is at the quantitative predication dewatering effect of different periods.

Liuzhou Radar Data takes from every 6 minutes swept-volume data once of Liuzhou Doppler radar, radar type Number CINRAD/SB, position (109.456 ° of E, 24.357 ° of N, 346.8m), 360 ° of comprehensive scannings, Spacing is 1 °, and the scanning elevation angle is 14 elevations angle between 0.5 °～19.5 °, and the body total number is according to from the low elevation angle Starting to the high elevation angle to terminate, 5～6min complete a swept-volume.The a length of 1km of reflectance range bin, Big range bin number is 460, and speed and a length of 0.25km of spectrum width range bin, ultimate range storehouse number is 920.

21 to 23 July of embodiment 2.1:2009, Yangze river and Huai river east, northeast, the south of the River and the Huaibei Most area falls over a large area moderate rain or heavy rain, isolated storm, torrential rain；Chuzhou and Chaohu big containing mountain and county Portion's point rainfall reaches 100 millimeter, the biggest point rainfall: 152 millimeters of Fengyang Lu Tang reservoir, it is long 112 millimeters, Pusan, 116 millimeters of east, Dingyuan storehouse reservoir, containing mountain city close 120 millimeters and county's army bridge reservoir 133 millimeters.

Fig. 6 is the analysis result of Hefei radar 07:00 to 08:00 on the 22nd July in 2009.Fig. 6 (a) It is the stratus and convective cloud scattergram obtained by fuzzy logic algorithm, it can be seen that this process is to drop with stratus Water is main cumulus mixed cloud precipitation.Stratiform clouds region, cloud cover area is big, in whole Precipitation Process The stratiform clouds persistent period is long, with low uncertainty, more stable, and subregion is identified as convective cloud, and it is embedded in greatly In the stratiform clouds of scope, area coverage is little, and cloud change is frequently.Thus, estimate that precipitation scope is bigger, And Hefei radar is western it is possible that local extra torrential rain.Precipitation is divided into Convective Cloud Precipitation and stratus Although precipitation improves the problem of underestimating of radar quantitative predication precipitation to a certain extent, but drops with reality Water still has larger difference, in order to improve the precision of radar quantitative predication precipitation, is divided into by precipitation further Warm cloud precipitation, Convective Cloud Precipitation and stratus precipitation.Fig. 6 (b) is three kinds of cloud-type recognition results, permissible Find out, fuzzy logic algorithm the semiconvection cloud obtained and stratus are re-identified as warm cloud, this time mistake Journey is redefined the mixed type precipitation of stratus precipitation and warm cloud precipitation.Around radar shown in Fig. 6 (c) VPR, the WRF simulation of the reflectivity factor all stratus points more than 10dBZ in the range of 20-80km This process h_0 DEG C is 5462.2m, it can be seen that this process exists bright band.In order to compare three The difference of group test, if Fig. 6 (d) is radar quantitative predication precipitation intensity and encryption precipitation station observation precipitation The scatterplot comparison diagram of intensity.Though it can be seen that the precipitation of three groups of test estimations is on the low side in precipitation station observation fall Water, but have preferable correspondence, and test three and improve radar quantitative predication precipitation to a great extent and estimate Meter underestimate problem, the precipitation of test three estimation are substantially better than test two and test one, closer to rainfall The part observing precipitation, especially raininess bigger of standing becomes apparent from.

During 11 days 08 June of embodiment 2.2:2008, Liuzhou meets with heavy showers weather, 12 days 04 up to 12 days 16 time 12h precipitation reach 233mm, meanwhile, Liujiang County occur 307 Mm extra torrential rain, during to 12 days 19 June, Liujiang water level has risen above the warning line 2.86 meters.To 6 The moon 13, relaxing occurs in Liuzhou rainfall.

This Precipitation Process is a difference in that with previous Precipitation Process, divides two classes falls with fuzzy logic algorithm It is identified as based on Convective Cloud Precipitation during water, and there is not bright band, i.e. without correcting bright band.Fig. 7 It it is the analysis result of Liuzhou radar 12:00 to 13:00 on the 12nd June in 2008.By fuzzy logic algorithm The convective cloud identified and stratus distribution such as Fig. 7 (a), find that in whole Precipitation Process, convective cloud is predominantly Position, convective cloud distribution is relatively decentralized, focuses primarily upon near radar, in southwest-northeast trend, observation drop Water figure (figure is slightly) is known, near lucky radar, southwest-northeast trend has more than the extra torrential rain of 25mm/h. From the point of view of three kinds of cloud-type recognition results that Fig. 7 (b) is given, this time process is stratus precipitation, convective cloud Precipitation and the mixed type precipitation of warm cloud precipitation.The semiconvection cloud obtained by fuzzy logic algorithm and part layer Cloud is re-identified as warm cloud, is positioned near (108.4 ° of E, 24.0 ° of N) this point and western part near radar Subregion is still convective cloud.Fig. 7 (c) is that around radar, in the range of 20-80km, reflectivity factor is big In the VPR of all stratus points of 10dBz, there is not bright band, this process of WRF simulation in this process H_0 DEG C is 5042.6m.Equally in order to compare the difference of three groups of tests, Fig. 7 (d) shows that radar is fixed Amount estimates the scatterplot contrast of precipitation intensity and encryption precipitation station observation precipitation intensity.It can be seen that test two The precipitation estimated slightly is better than testing the precipitation of an estimation, the precipitation of test three estimation be substantially better than test two with Test one, when especially raininess is bigger.

The problem underestimated for radar quantitative predication precipitation, this chapter is by the identification bright band of the propositions such as Zhang Method judges whether bright band, then corrects bright band if there is bright band, then uses the propositions such as Xu The probability of warm cloud precipitation is judged whether according to VPR；Then fuzzy logic algorithm is utilized to be divided by precipitation For Convective Cloud Precipitation and stratus precipitation；The reflectivity factor value highly located further according to VPR slope, 1km with And earth's surface wet bulb temperature, precipitation is divided into warm cloud precipitation, Convective Cloud Precipitation and stratus precipitation three class, finally The Z highly located with 1km uses different Z-I relations to estimate precipitation for different precipitation type.Design altogether Three groups of tests, test one, and the single Z-I relation only using the Doppler radar of standard to give tacit consent to estimates Meter precipitation, i.e. Z=300I^1.4；Test two, is divided into Convective Cloud Precipitation and stratus precipitation, then root by precipitation Precipitation is estimated according to each self-corresponding Z-I relation；Test three, is divided into warm cloud precipitation and convective cloud fall by precipitation Water and stratus precipitation, then estimate precipitation according to each self-corresponding Z-I relation.Wherein Convective Cloud Precipitation Use Z=300I^1.4, Stratiform Cloud Precipitation uses Z=200I^1.6, torrid zone warm cloud precipitation uses Z=30.7I^1.66.Logical Cross the radar quantitative predication precipitation error analysis in Hefei whole summer in June, 2010 to August, and choose 21 to 23 July in 2009 Anhui Province's Heavy Precipitation and 11 to 13 June in 2008 wide Western Liuzhou Heavy Precipitation, when precipitation is divided into warm cloud, stratus and Convective Cloud Precipitation three class by inspection, radar is fixed Amount estimates dewatering effect, tentatively obtains as drawn a conclusion:

(1) one time Precipitation Process comprises multiple different type of precipitation, if only applied code Doppler sky The single Z-I relation of gas radar acquiescence estimates precipitation, though precipitation can be estimated to a certain extent, but corresponding There is error in precipitation station observation precipitation；Though fuzzy logic algorithm can well identify cumulus mixed cloud precipitation and Convective Cloud Precipitation and stratus precipitation, and improve underestimating of radar quantitative predication precipitation to a certain extent Problem, but yet suffer from certain error；Precipitation is divided into warm cloud precipitation, Convective Cloud Precipitation and stratus During precipitation three types, the precipitation field of radar quantitative predication not only overall precision is the highest, and to heavy rain portion The estimated accuracy divided is the highest.

(2) precipitation all can be preferably estimated in three groups of tests of each process, though estimating that precipitation is on the low side in rainfall Stand and observe precipitation, but corresponding consistent with precipitation station observation precipitation, and the precipitation of test two estimation are slightly better than test One precipitation estimated, the precipitation of test three estimation is substantially better than test two and test one, and especially raininess is bigger Time, test three estimation precipitation is compared front two groups of tests and is observed precipitation closer to precipitation station.

(3) the radar quantitative predication precipitation method that this chapter proposes, estimates the radar in different time and place Precipitation is equally applicable.

The embodiment of the above is only presently preferred embodiments of the present invention, and the present invention not does any form On restriction.Any those of ordinary skill in the art, without departing from technical solution of the present invention ambit Under, all may utilize the technology contents of the disclosure above technical solution of the present invention is made more possible variation and Retouching, or it is revised as the Equivalent embodiments of equivalent variations.Therefore all contents without departing from technical solution of the present invention, The equivalent equivalence change made according to the thinking of the present invention, all should be covered by protection scope of the present invention.

Claims (9)

1. based on Assimilate Doppler Radar Data identification stratus, convective cloud and a method for warm cloud precipitation rate, its It is characterised by comprising the steps:

Radar measuring echo strength ref；

Whether VPR identification module exists according to echo strength ref Vertical Profile identification warm cloud；

Warm cloud identification module sets up polar coordinate, warm cloud identification mould according to radar echo intensity ref at buffer area Tuber is according to place polar ref interpolation and wet bulb temperature, it is judged that whether warm cloud exists；If warm cloud is deposited , then warm cloud identification module utilizes ref that 3D-Barnes interpolation algorithm calculates in rectangular coordinate system and wet Bulb temperature, the precipitation rate of the warm cloud in identification stratus, the precipitation rate of the warm cloud in identification convective cloud；

Processing module calculates the step of precipitation rate, this step include processing module according to echo strength ref and Ref=10Lg (Z) calculates radar return reflectivity factor Z, and processing module is according to the convection current of ref, ref Cloud probit P_conv, wet bulb temperature judge cloud type as: convective cloud, stratiform clouds or torrid zone warm cloud, If cloud layer is convective cloud, then use Z=300I^1.4Output precipitation rate I, if cloud layer is stratiform clouds, then uses Z=200I^1.6Output precipitation rate I, if cloud layer is warm cloud, then precipitation uses Z=30.7I^1.66 Output precipitation rate I.

The most according to claim 1 based on Assimilate Doppler Radar Data stratus, convective cloud and warm cloud precipitation Rate recognition methods, it is characterised in that processing module according to convective cloud probit P_conv of ref, ref, Wet bulb temperature judges in the cloud type step as: convective cloud, stratiform clouds or torrid zone warm cloud, including as follows Step:

If meeting in the lattice point that the 1km of the rectangular coordinate coordinate of described stratus highly locates: ref > 25dBz And wet bulb temperature > 2 DEG C, then it is considered as the warm cloud with stratus.

The most according to claim 1 based on Assimilate Doppler Radar Data stratus, convective cloud and warm cloud precipitation Recognition methods, it is characterised in that processing module according to convective cloud probit P_conv of ref, ref, Wet bulb temperature judges in the cloud type step as: convective cloud, stratiform clouds or torrid zone warm cloud, including as follows Step:

If meeting in the lattice point that polar 1km of described convective cloud highly locates: ref > 25dBz and wet Bulb temperature > 2 DEG C, then it is considered as the warm cloud with convective cloud.

The most according to claim 1 based on Assimilate Doppler Radar Data stratus, convective cloud and warm cloud precipitation Rate recognition methods, it is characterised in that the step of described identification warm cloud comprises the following steps:

The first step, processing module calculates VIL according to equation below:

VIL=∑ 3_.44×10^-6×[(Z_i+Z_i+1)/2]^4/7×Δh_i

Wherein,

VIL is that the vertical accumulation aqueous water that radar can detect in the height layer of effective reflectivity factor value contains Amount；

Z_i,Z_i+1It it is reflectivity factor；

Δh_i=500m；

Observation station in described polar coordinate is divided into stratus and non-stratus according to described VIL by processing module, so The radar echo intensity all stratus points more than 10dBZ in the range of 20-80km around rear calculating radar Average VPR；

Second step, processing module is according to stratus VPR calculated in above-mentioned steps and combines identification bright band Method judges whether bright band, and concrete grammar is as follows:

(1) 0 DEG C of isothermal line height h_0 DEG C is obtained by WRF simulation, it is contemplated that h_0 DEG C is by pattern mould Intend the height obtained, so finding downwards maximum reflectivity factor Z in VPR at h_0 DEG C of overhead 500m_peak And height h_peak；Described Δ h_i=500m is to obtain according to model predictions Error Calculation；

(2) from h_peakZ is upwards searched at highly place_peakReflectivity factor Z when reducing 10%_topAnd height h_top；From h_peakZ is searched for downwards at highly place_peakReflectivity factor Z when reducing 10%_bottomAnd height h_bottom；

(3) if meet following condition and think and there is bright band simultaneously:

h_top-h_bottom≤D₀

h_top-h_peak≤D₁

h_peak-h_bottom≤D₁

Wherein, D₀And D₁It is to change with radar scanning mode and the vertical resolution of radar echo intensity Coefficient, wherein D₀And D₁Take 1.5km and 1km respectively；

(4) if there is bright band, bright band heights of roofs BB_tWith height BB at the bottom of bright band_bBe calculated as follows:

${\mathrm{BB}}_t=\left\{\begin{array}{cc}{h}_{top},& h_{top}-h_{peak}\&le;D_t\\ h_{peak}+D_{top},& h_{top}-h_{peak}>D_t\end{array}\right.$

${\mathrm{BB}}_b=\left\{\begin{array}{cc}{h}_{bottom},& h_{peak}-h_{bottom}\&le;D_b\\ h_{peak}-D_{bottom},& h_{peak}-h_{bottom}>D_b\end{array}\right.$

Wherein, D_tAnd D_bFor adjusting at the bottom of bright band top and bright band, wherein, D_t=0.5km, D_b=0.7km； If the radar echo intensity value at height at the bottom of bright band is less than 28dBZ, then again from h_peakHighly place to Lower search radar echo strength value is more than or equal to minimum altitude during 28dBZ, and is made by its place height At the bottom of bright band, this purpose is to avoid the too much correction of bright band；

3rd step, when there is bright band, then processing module need to correct bright band, and processing module corrects the side of bright band Method is as follows:

Assume to there is effective BB_tAnd BB_bValue and the radar echo intensity Z of correspondence_BtAnd Z_Bb, and BB_b<h_peak, BB_t>h_peak, then BB is calculated_tAnd BB_bThe slope of the radar echo intensity between height:

S=(Z_Bt-Z_Bb)/(BB_t-BB_b)

It is pointed to BB_tAnd BB_bBetween certain height h_iAll radar echo intensities assignment again, assignment again After radar echo intensity be:

Z_i=Z_Bt-S×(BB_t-h_i)

Wherein, h_iIt is the vertical height needing to correct lattice point, if there is effective BB_bValue, and do not exist Effective BB_tValue, and BB_b<h_peak, then BB_bAll radar echo intensity values of level above are big In Z_BbValue be all assigned to Z_Bb；

4th step, processing module recalculates around radar in the range of 20-80km after correcting bright band The average VPR of the radar echo intensity all stratus points more than 10dBZ,

5th step, processing module judges whether warm cloud:

If bright band exists, then processing module to bottom bright band to the part least square bottom VPR Method matching VPR；

If bright band does not exists, then processing module at h_0 DEG C of height to the part minimum bottom VPR Square law matching VPR, and calculate the slope of VPR, if calculating moment and the body in front 4 moment thereof Sweep and have 3 α≤0 in VPR slope, then processing module is thought and be there is the probability of warm cloud precipitation, otherwise recognizes For there is not warm cloud precipitation.

The most according to claim 2 based on Assimilate Doppler Radar Data stratus, convective cloud and warm cloud precipitation Rate recognition methods, it is characterised in that also include the preliminary step identifying stratus,

When radar vertical integrated liquid water content meets VIL < 6.5kg/m²Time, then processing module is judged to have Stratus.

The most according to claim 1 based on Assimilate Doppler Radar Data stratus, convective cloud and warm cloud precipitation Rate recognition methods, it is characterised in that the step of described identification stratus precipitation and Convective Cloud Precipitation rate include with Lower step:

The first step, processing module obtains each identification parameter, and described parameter has four: 2km highly to locate Echo strength ref_wrkAnd standard deviation std, echo high and 2km highly locate the product p of echo strength_ztop、 Vertical accumulative Liquid water content VIL；

Wherein, echo high be radar echo intensity be the maximum height of 18.3dBZ, VIL be by suppose The return of all reflectivity factor is all to be caused by aqueous water；

Second step, the membership function u of processing module structure cloud identification_k,e(x), its function expression is as follows:

${\mathrm{\&mu;}}_{k,C}(x,a,b)=\left\{\begin{array}{cc}0,& x\&le;a\\ (x-a)/(b-a),& a<x<b\\ 1,& x\&GreaterEqual;b\end{array}\right.,(k=1,2,...,N)$

${\mathrm{\&mu;}}_{k,S}(x,a,b)=\left\{\begin{array}{cc}1,& x\&le;a\\ (b-x)/(b-a),& a<x<b\\ 0,& x\&GreaterEqual;b\end{array}\right.,(k=1,2,...,N)$

Wherein, k represent identification parameter (k=1,2 ..., N), totally 4 identification parameters, so taking N=4；e For variable code name, taking C and represent convective cloud, e takes S and represents stratiform clouds；X is each identification parameter numerical value；No With reference thresholds a that identification parameter is corresponding, b is different；

For ref_wrk, a=20dBZ, b=45dBZ；

For std, a=1dBZ, b=14dBZ；

For p_ztop, a=100km dBZ, b=500km dBZ；

For VIL, a=0.5kg/m², b=5.0kg/m²；

It is with the side of membership function using above-mentioned identification parameter by input variable as input variable, i.e. obfuscation Formula is converted into Fuzzy dimension, and its value excursion is [0,1], and its expression formula is as follows:

μ_k,S(x, a, b)=1-μ_k,C(x,a,b)

3rd step, utilizes following expression to seek the conditional probability of stratiform clouds and convective cloud:

$P_e=\frac{\underset{k=1}{\overset{N}{\&Sigma;}}{w}_k\&CenterDot;P_{k,e}}{\underset{k=1}{\overset{N}{\&Sigma;}}{w}_k}$

Wherein, P_k,e=μ_k,e(X_k), e is the code name of change, represents stratiform clouds (s) or convective cloud (c), w_k For the weight coefficient of each identification parameter, N is identification parameter number, so taking N=4, Pc and Ps generation respectively Table is identified as the probit of convective cloud and stratiform clouds；Pc Yu Ps sum is equal to 1,

When Pc >=0.5, then processing module judges is convective cloud, otherwise judges it is stratiform clouds.

The most according to claim 1 based on Assimilate Doppler Radar Data stratus, convective cloud and warm cloud precipitation Rate recognition methods, it is characterised in that the recognition methods of described warm cloud precipitation comprises the following steps:

When there is warm cloud, and lattice point meets radar echo intensity value that 1km highly locates more than 25dBZ, And when earth's surface wet bulb temperature is more than 2 DEG C, described lattice point is warm cloud precipitation lattice point, otherwise it is assumed that right and wrong Warm cloud precipitation lattice point, earth's surface wet bulb temperature W_BcBe calculated as follows:

W_Bc=(0.00066 × P × T_c+4098×E/T_dc×(237.7+T_dc)²)/(0.00066×P+4098×E/(237.7+T_dc)²)

Wherein, E is obtained by WRF simulation,T_dcObtained by WRF simulation, T_dc Being dew point wet bulb temperature, P is air pressure, T_cIt is temperature.

The most according to claim 7 based on Assimilate Doppler Radar Data stratus, convective cloud and warm cloud fall Water rate recognition methods, it is characterised in that the Z-I relation described in the method for described precipitation assessment estimates fall Water, uses the echo reflection rate factor Z detection precipitation rate that 1km highly locates.

The most according to claim 1 based on Assimilate Doppler Radar Data stratus, convective cloud and warm cloud fall Water rate recognition methods, it is characterised in that the formula of described 3D-Barnes interpolation algorithm is:

$f^a\left(r_k\right)=\frac{\underset{i}{\&Sigma;}{\mathrm{\&omega;}}_i{\mathrm{ref}}^o\left(r_i\right)}{\underset{i}{\&Sigma;}{\mathrm{\&omega;}}_i}$

Here variable ref is radar echo intensity, and subscript a and o represents amount of analysis and observed quantity respectively, Described amount of analysis is the value after interpolation on lattice point, r_kIt is the coordinate of mesh point k, r_iIt is in observation station i Observation, ω_iIt is weight coefficient, is defined by the formula:

ω_i=exp [-(r_h ²/2R_h ²+r_v ²/2R_v ²)]

Wherein r_hIt is the horizontal range between observation station i and mesh point k, r_vIt is vertical dimension, R_h、R_vPoint The prespecified horizontal and vertical radius of influence, described Rh take 5 HORIZONTAL PLAIDs away from, described Rv takes 3 vertical lattice away from.