research article radar-derived quantitative precipitation...

Research ArticleRadar-Derived Quantitative Precipitation Estimation Based onPrecipitation Classification

Lili Yang12 Yi Yang1 Peng Liu1 and Lina Wang2

1Key Laboratory for Semi-Arid Climate Change of the Ministry of Education Key Laboratory of Arid Climatic Changing andReducing Disaster of Gansu Province College of Atmospheric Sciences Lanzhou University Lanzhou 730000 China2Gansu Province Environmental Monitoring Center Lanzhou 730020 China

Correspondence should be addressed to Yi Yang yangyilzueducn

Received 30 June 2016 Accepted 13 October 2016

Academic Editor Zhe Li

Copyright copy 2016 Lili Yang et alThis is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

A method for improving radar-derived quantitative precipitation estimation is proposed Tropical vertical profiles of reflectivity(VPRs) are first determined from multiple VPRs Upon identifying a tropical VPR the event can be further classified as eithertropical-stratiform or tropical-convective rainfall by a fuzzy logic (FL) algorithm Based on the precipitation-type fields thereflectivity values are converted into rainfall rate using a Z-R relationship In order to evaluate the performance of this rainfallclassification scheme three experiments were conducted using three months of data and two study cases In Experiment I theWeather Surveillance Radar-1988 Doppler (WSR-88D) default Z-R relationship was applied In Experiment II the precipitationregimewas separated into convective and stratiform rainfall using the FL algorithm and correspondingZ-R relationshipswere usedIn Experiment III the precipitation regime was separated into convective stratiform and tropical rainfall and the correspondingZ-R relationships were applied The results show that the rainfall rates obtained from all three experiments match closely with thegauge observations although Experiment II could solve the underestimation when compared to Experiment I Experiment III sig-nificantly reduced this underestimation and generated the most accurate radar estimates of rain rate among the three experiments

1 Introduction

Owing to its significant impacts on human activities precip-itation is one of the most important factors in meteorologicalanalysis accordingly its study has attracted considerableattention Quantitative precipitation estimation (QPE) playsa very important role in generating warnings of meteo-rological hydrological and geological disasters The accu-racy of numerical weather predictions can be significantlyimproved when the precipitation rate is assimilated into theweather prediction models [1 2] Currently precipitationrate data is obtained primarily from rain gauge networkswhich can provide real-time rainfall estimates Howeverthe spatial distribution of rain gauges is sparse in remoteregions particularly in mountainous areas which can resultin insufficient spatial resolution for accurate mapping of therainfall patterns Consequently obtainingQPE products withhigh spatial and temporal resolution is currently one of themain tasks in conducting hydrometeorological studies [3 4]

Weather radars in particular can provide precipitation esti-mates with high spatial and temporal resolution over a largearea [5] Radar reflectivity Z (mm6m3) is related to rainfallrateR (mmh) through a variable power law relationship (Z-R relationship) which can be used to obtain QPE products

Numerous Z-R relationships based on raindrop sizedistribution observations have been proposed [6ndash8] Z-Rrelationships of the formZ =AR119861 can be derived frompairs ofZ and R observations assuming that sufficient correspondingpaired Z-R data samples are available The default WSR-88DZ-R relationship Z = 300R14 [9ndash11] was derived based onsummer convective events in Florida USA However thisdefault Z-R relationship cannot fully address the complicatedcharacteristics of rain in for example China [12ndash14] It isknown that different types of precipitation correspond to dif-ferent raindrop size distributions (DSDs) and precipitationtypes can also be classified based on the formation mech-anism duration and internal structure The precipitationregime can therefore be divided into different rainfall types

Hindawi Publishing CorporationAdvances in MeteorologyVolume 2016 Article ID 2457489 16 pageshttpdxdoiorg10115520162457489

2 Advances in Meteorology

Convective rainfall systems are associatedwith strong verticalvelocity fields small areal coverage and high rainfall inten-sities and an echo that can be extended to a relatively highaltitude andwith a high center ofmass Conversely stratiformrainfall systems have relatively weak vertical velocity fieldsgreat horizontal homogeneity and low rainfall intensityAlthough the intensity of stratiform rainfall is much weakerthan that in adjacent convective cells stratiform rain typicallycovers a larger area and always contributes with a significantportion (40ndash50) of the total rainfall even in convective-dominant systems [15] Tropical rain systems generally occurin lower altitudes and very humid air with condensationoccurring continuously throughout the air descentThemainfeature of tropical rain systems is that their reflectivityincreases monotonically as height decreases yielding a max-imum reflectivity at the lowest levels of the vertical profileof reflectivity (VPR) resulting in considerable error whenconducting precipitation estimation using the default Z-Rrelationship ofWSR-88DMoreover multiple types of precip-itation may coexist within a wide range of rainfall events

Radar research scientists worldwide have undertakenextensive studies to improve the accuracy of radar QPE [16ndash23] Typically precipitation is first classified into varioustypes and corresponding multiple Z-R relationships are thenused [19 24ndash26] Precipitation can generally be classified intoeither convective or stratiform regimes which has been aparticular focus of many studies [27ndash29] Such basic classi-fications typically adopt a fixed threshold or set of boundaryconditions for rainfall recognition However these methodsare known to be sensitive to the used threshold values andbecause there is a considerable overlap between stratiformand convective regimes in many respects using fixed bound-ary conditions or thresholds often leads to false positivesMoreover it is often difficult to directly identify the boundarybetween stratiform and convective regimes [30] Yang et al[30] used the reference values of Steiner et al [29] to derive astatistical relationship between four characteristic parametersand two rain types in order to develop the probability distri-bution of convective and stratiform rainfall based on a fuzzy-logic (FL) algorithm which was able to improve the accuracyof QPE [31] Despite efforts such as these many techniquescan still underestimate rainfall to some extent Based on thestratiform and convective precipitation classification manyresearchers have suggested focusing on tropical rain as anavenue to mitigate the underestimation of QPE and somestudies have been able to reduce QPE underestimation byemploying an automated radar technique to identify tropicalprecipitation in order to divide precipitation into tropicalrain and stratiform and convective types [32 33] Qi et al [34]converted reflectivity into precipitation utilizing the tropicalrain Z-R relationship instead of the default WSR-88D Z-Rrelationship which improved the accuracy to some extentSimilarly Zhang et al [35] improved the accuracy of radar-derived QPE by using an automatic identification methodto identify tropica1 stratiform and convective precipitationtypes based on radar observations and model outputs Theythen applied various Z-R relationships in order to convertreflectivity into rain rates based on precipitation types The

results showed that their method improved the accuracy ofradar-derived QPE

In this study precipitation is first divided into stratiformand convective precipitation types using the FL algorithmsproposed by Yang et al [30] and then tropical rain is identi-fied based on the VPR characteristics [36 37] as discussed byXu et al [33] Tropical rainfall is further divided into tropical-stratiform and tropical-convective rainfall and the respectiveadaptive Z-R relationships are then applied for QPE Athree-month dataset is used to compare the newly proposedtropical rain identification scheme with the results obtainedwith the default WSR-88D Z-R relationship and a convec-tivestratiform rain identification scheme Section 2 describesthe precipitation classificationmethod and the correspondingZ-R relationships and discusses the benefits of precipitationclassification Section 3 illustrates the evaluation results ofusing long-term datasets using descriptive statistics Two casestudies are also provided in Section 4 to evaluate the perfor-mance of the proposed precipitation classification methodFinally Section 5 presents the main results of the study

2 Algorithm Description

Underestimation of rainfall can occur when precipitation isdivided into general stratiform and convective types Thisstudy aims to improve the accuracy of radar QPE by buildingupon the convective stratiform and tropical precipitationtechnique described by Xu et al [33] The proposed methoduses volume scan reflectivity data from a Doppler radarto recognize tropical VPRs by first dividing rain data intostratiform and convective precipitation (using an FL algo-rithm) and then classifying tropical rain events as stratiformconvective or tropical precipitation Figure 1 shows a flowchart illustrating this process of precipitation classification

21 Identification of Tropical VPRs This study uses VPRsto identify the potential presence of tropical rain As afirst step volume scan reflectivity data from Doppler radarare simply classified into assumed stratiform or convectivetypes based on the vertically integrated liquid (VIL) watercontent In particular the data are classified as belonging tothe stratiform type when the VIL value is below 65 kgm2[38] otherwise they are classified as convective Stratiformreflectivity observations larger than 10 dBZ from all tilts inan annular region between two predefined ranges 1199031 and1199032 (with the empirical values of 20 and 80 km resp) areused to obtain an averaged VPR [39] The number of datasamples within a 20-km radius of the radar is limited owingto a terrain clearance rule incorporated into a quality control(QC) scheme [40] whereby any radar bin whose bottomis within 50m of the ground is removed Furthermore theregion of interest has to be far enough from the radar receiverto avoid the cone of silence in the receiverrsquos immediatevicinity The accuracy of precipitation classification is alsolimited by the decrease in vertical resolution beyond an 80-km radius

The second step combines the output of the first step withthe method of bright band identification proposed by Zhang

Advances in Meteorology 3

Start

Stratiform cloud No stratiform cloud

Bright band (BB) present

Correct BB

Volume scan reflectivity data wereinterpolated to a Cartesian coordinatesystem and for any grid calculate theprobability of convectivefuzzy-logic algorithms

Polar VPR Hail

Convective rain Stratiform rain

Convective rainTropical rain Tropical rain Stratiform rain

Yes

Yes

Yes

Yes

Yes

YesYes

No

No

No

No

NoNo

Do not correct BB

VIL lt 65 kgm2

Use WRF to get 0∘C height

Ref lt 50

1205724 Get the slope (1205720)middot middot middot

3 of 5 120572(1205724 1205720) lt 0

dBZ

For any grid of 1 kmaltitude

(1) ref gt 25

(2) wet bulb temperaturegt 2∘C

dBZ


(1) ref gt 25


dBZ

Pconv gt 05

(Pconv) with

Figure 1 Flow chart of the precipitation classification method devised in this study

et al [39] to determine whether a bright band is presentThespecific method adopted for this step can be summarized asfollows

(1) The local maximum reflectivity (119885peak) and its height(ℎpeak) in the VPR are found The search for the

local maximum in the VPR starts at 500m abovethe 0∘C isotherm altitude (h 0∘C) obtained from theWeather Research and Forecasting (WRF) model andcontinues downward A 500-m cushion is used toaccount for uncertainties in the modelrsquos 0∘C height


(2) Once the local maximum is found the height ℎtop(ℎbottom) above (below) the maximum level at whichreflectivity decreases monotonically by a given per-centage (default = 10) [39] of the maximum reflec-tivity (in dBZ) is determined along with its corre-sponding reflectivity 119885top (119885bottom)

(3) If the following three criteria are met a bright band isconsidered to exist

ℎtop minus ℎbottom le 1198630ℎtop minus ℎpeak le 1198631

ℎpeak minus ℎbottom le 1198631(1)

The parameters 1198630 and 1198631 are assumed to be depen-dent on both the vertical resolution of the reflectivityobservations and the radar scan strategyHere1198630 and1198631 are set to 15 and 1 km respectively

(4) If a bright band exists its top (BB119905) and bottom (BB119887)heights are defined as follows

BB119905 = ℎtop ℎtop minus ℎpeak le 119863119905ℎpeak + 119863119905 ℎtop minus ℎpeak gt 119863119905

BB119887 = ℎbottom ℎpeak minus ℎbottom le 119863119887ℎpeak minus 119863119887 ℎpeak minus ℎbottom gt 119863119887

(2)

Here 119863119905 and 119863119887 serve as top and bottom caps of the brightband (BB) with default values of 500 and 700m respectively[39]

To avoid excessive corrections of the bright band theheight of the local maximum reflectivity is searched againif the minimum reflectivity is lower than a preset threshold(default = 28 dBZ) This height is defined as the lowest heightat which the reflectivity is larger than or equal to the threshold[38] This parameter can sometimes lead to insufficient orexcessive bright band correction an issue that will be solvedin the future using observations from a dual-polarizationradar [41]

The third step involves the correction of any existingbright bands [42] If the values of BB119905BB119887 and their corre-sponding reflectivities (119885B119905 and 119885B119887 resp) are valid BB119887 ltℎpeak and BB119905 gt ℎpeak and then the slope (119878) of reflectivitybetween BB119905 and BB119887 is calculated as follows

119878 = (119885B119905 minus 119885B119887)(BB119905 minus BB119887) (3)

If the reflectivity height ℎ119894 is located betweenBB119905 andBB119887the reflectivity is reassigned as

119885119894 = 119885B119905 minus 119878 times (BB119905 minus ℎ119894) (4)

If BB119905 does not exist the value of BB119887 is valid and BB119887 ltℎpeak and the reflectivities located above BB119887 (with valuelarger than 119885B119887) are reassigned to 119885B119887

The fourth step which involves reflectivity observationsfrom all tilts in an annular region between 20 and 80 kminvolves recalculation of the average stratiform VPR afterbright band correction

The fifth step involves determining whether a tropicalVPR has been captured Following Xu et al [33] the iden-tification (or diagnosis) of tropical VPR is made with a least-squares fit from the bottom of the bright band to the bottomof the VPR in cases where a bright band exists otherwisethe fit is made from h 0∘C to the bottom of the VPR Theslope (120572) of the VPR is then calculated The resulting VPRis considered as tropical if three out of five values of 120572 (forthe present and the four previous moments) are lower thanor equal to 0 otherwise it is regarded as a nontropical VPR

22 Convective and Stratiform Rainfall Classification Toimprove the accuracy of radar QPE this study adopted the FLalgorithm developed by Yang et al [30] to classify convectiveand stratiform rainfall Brief details of the FL algorithm aregiven below

First the volume scan reflectivity data from the Dopplerradar are interpolated to a Cartesian coordinate system withthe same latitude and longitude range (001∘) of the originaldata Four features denoted as F1ndashF4 are calculated basedon the reflectivity data These features are respectively (a)the reflectivity at 2 km (b) the reflectivity standard deviationin the horizontal direction (c) the product of the radar topheight and the reflectivity value at 2 km and (d) the verticallyintegrated liquid water content [30]

As a second step membership functions are used todetermine the degree to which each feature belongs to eachrain type using a fuzzification process Linear functions areused as membership functions for convection

120583119896119862 (119909 119886 119887) =

0 119909 le 119886(119909 minus 119886)(119887 minus 119886) 119886 lt 119909 lt 1198871 119909 ge 119887

(119896 = 1 2 119873)(5)

and stratification

120583119896119878 (119909 119886 119887) =


(119896 = 1 2 119873) (6)

where119873 is the input feature parameter subscript 119896 is set to 12 3 or 4119862 and 119878 identify the convective and stratiform casesrespectively 119909 is the feature value 119886 is the left breaking pointand 119887 is the right breaking point Different feature parametersare assigned different breaking points For F1 the parametersare set to 119886 = 20 dBZ and 119887 = 45 dBZ for F2 they are a =1 dBZ and 119887 = 14 dBZ for 1198653 they are 119886 = 100 kmsdotdBZ and


b = 500 kmsdotdBZ and for 1198654 they are a = 05 kgkm2 and 119887 =50 kgkm2 [30] From (5) and (6) it is clear that

120583119896119878 (119909 119886 119887) = 1 minus 120583119896119862 (119909 119886 119887) (7)

Third the weighted average values of the measurementbelonging to each specific class are obtained as follows

119875119890 = (sum119873119896=1119882119896 sdot 119875119896119890)(sum119873119896=1119882119896) (8)

where119875119896119890 = 120583119896120576(119883119896) 119890 identifies the convective or stratiformcase 119882119896 is the weighting factor (with equal weightingassigned in this study for simplicity) 119873 is the numberof input feature parameters and 119875119888 and 119875119904 indicate theweighted average values of the convective and stratiformclasses respectively As the sum of 119875119888 and 119875119904 is 1 the valueof 119875119888 can be used to express classifications for convectionin a probabilistic manner Grid points for which 119875119888 ge 05are classified as convective rainfall all other grid points areclassified as stratiform rainfall

23 Identification of Tropical Rainfall Regions To improve theaccuracy of the radarQPE tropical rainfall is considered to bepresent if tropical VPR is present and the grid at 1 km altitudemeets the following two criteria [33] (1) the reflectivity isgreater than 25 dBZ and (2) the surface wet bulb temperatureis greater than 2∘C The surface wet bulb temperature iscalculated as follows

119864 = 611 times 1075times119879119889119888(2377+119879119889119888)119882119861119888= (000066 times 119875 times 119879119888 + 4098 times 119864119879119889119888 times (2377 + 119879119889119888)2)(000066 times 119875 + 4089 times 119864 (2377 + 119879119889119888)2)

(9)

where 119879119889119888 is the dew point temperature (in ∘C) E is the vaporpressure from dew point (in mb)119882119861119888 is the surface wet bulbtemperature (in ∘C) 119875 is the pressure (in mb) and 119879119888 is theair temperature (in ∘C) The above parameters are simulatedusing the WRF model or from other data

In this way stratiform rainfall is further divided intostratiform and tropical-stratiform rainfall and convectiverainfall is further divided into convective and tropical-convective rainfall

24 Radar QPE Calculation Three sets of experiments wereconducted to test the proposed method In Experiment Ionly the WSR-88D default Z-R relationship (119885 = 300R14)was used In Experiment II as the rain type is identified bya FL algorithm the corresponding Z-R relationships wereapplied to the rain rate calculation at each grid point Foreach grid point the relationship 119885 = 300R14 was appliedfor the convective rainfall calculation if 119875119888 ge 05 Oth-erwise the relationship 119885 = 200R16 [6] was applied forstratiform rainfall calculation In Experiment III stratiformand convective rainfall in tropical rainfall events were further

divided into stratiform and tropical-stratiform rainfall andconvective and tropical-convective rainfall respectively Therelationships were the same as those used in Experiment IIexcept that the tropical rainfall relationship 119885 = 307R166was adopted (Yadong Wang personal communication) Inan attempt to exclude the impact of nonprecipitation echoesand hail quality control was conducted prior to precipitationclassification Additionally reflectivity at 1 km was used toreduce the errors caused by spatial inconsistencies and theeffects of bright bandmeasurements not yet properly revised

25 Performance Evaluation The mean deviation error(BIAS) relative absolute error (RAE) rate and root meansquare error (RMSE) are selected as statistical indicatorsfor quantitatively assessing the quality of radar QPE inthe different experiments [43] The mean deviation error isdefined as follows

BIAS = 1119899119899sum119894=1

(119877119886 (119894) minus 119877119892 (119894)) (10)

where 119877119886(119894) is the radar QPE of the 119894th station 119877119892(119894) is theobserved precipitation of the 119894th site and 119899 is the total numberof samples involved in the performance assessment

The radar QPE error may be positive or negative there-fore its mean will not reflect the radar QPE performance Toobjectively reflect performance the mean absolute error rateis used and is defined as follows

RAE = sum119899119894=1 10038161003816100381610038161003816119877119886 (119894) minus 119877119892 (119894)10038161003816100381610038161003816sum119899119894=1 119877119892 (119894) times 100 (11)

In addition downpour and drizzle may coexist duringa precipitation process as the RAE of downpour is largerthan that of drizzle the root mean square error primarilyrepresents the performance of the downpour component ofthe rainfall

RMSE = (sum119899119894=1 (119877119886 (119894) minus 119877119892 (119894))2119899 )12

(12)

As the above three equations imply a BIAS closer to0 corresponds to more similar (in the average) QPE andobserved rain types whereas a smaller RAE correspondsto a higher radar QPE accuracy for the entire precipitationprocess Similarly a smaller RMSE indicates a more accurateradar QPE for the heavy precipitation portion of the process

3 Long-Term Data Statistics

Many previous studies have shown that extreme precipitationis likely to take place in the Yangtze RiverndashHuaihe Rivervalley the Yangtze River valley and on the southeast coastof China with persistent heavy rainfall events being moreabundant in June and July [44ndash46] Tropical rain hasbeen found in these regions although underestimationoccurs when conventional QPE is utilized We evaluatedthree months of data obtained from the Yangtze RiverndashHuaihe River valley in order to assess the degree to which


Table 1 Hefei CINRAD radar parameters

Wavelength Beam width Scan mode Elevation angle Bin spacing Obtained data Timeresolution

10 cm 1∘ half-powerbeam width

360∘ azimuthalvolume 05∘ndash195∘

250m for velocity andspectral width 1000m for

reflectivity

Reflectivity radial velocityspectral width

Approximately6min

the proposed identification process improves estimationaccuracy when compared with analyses conducted usingthe default WSR-88D Z-R relationship and convective-stratiform precipitation segmentation rain identificationschemeDoppler radar data for this studywere taken from theHefei Doppler radar (117258∘E 31867∘N 1655m) a ChineseNew Generation Radar S-band radar instrument (CINRADWSR-98DSA) The WSR-98DSA is a 10-cm wavelengthDoppler radar with a 1∘ half-power beam width The dataobtained consisted of volume scans of radar reflectivity radialvelocity and spectrum width collected in a polar coordinatesystem at increasing elevation angles During periods ofprecipitation the radar operates in a 360∘ azimuthal volumescan mode with the elevation angle increasing from 05∘ to195∘ and the temporal resolution of the data depending onthe operational mode of the radar Bin spacing is 250m in theradial direction with reflectivity values averaged over fourbins to increase the number of independent measurementscollected for each recorded value Accordingly reflectivityvalues are recorded at 1-km intervals along the radar beamwhereas velocity parameters are recorded at 250m intervalsEach volume scan takes approximately 6min The radarparameters are shown in Table 1

31 Data and Domain The data from the Hefei Dopplerradar and gauge of Anhui Province were collected fromJune to August in 2010 The horizontal domain (115758∘Endash118758∘E 30367∘Nndash33367∘N) in this study was a 300 times300 km grid centered at the radar with a 1 km horizontalresolution (a 301 times 301 grid)The vertical domain consisted of73 layers and had an altitude of 18 km (025-km resolution)The center of the horizontal domain is the site of Hefei radarThe domain is shown in Figure 2

32 Statistical Results for Three Months of Data To com-pare the estimates obtained with different schemes somestatistical indicators were obtained for the JunendashAugust 2010period and are presented in Table 2 As shown ExperimentII performed consistently better than Experiment I butExperiment III produced the results closest to the observedprecipitation and with the greatest precision It should bementioned that there is a significant difference between theobserved precipitation and the radar estimated rainfall whenthe intensity of the observed precipitation is less than 2mmhtherefore rainfall with intensity below 2mmh was excluded

The statistical indicators in Table 2 were obtained fromall precipitation types to further understand the particulareffects of tropical rain the statistical indicators specific oftropical rain events that occurred in the JunendashAugust 2010period are presented in Table 3 In other words in contrast

33∘N

32∘30998400N

32∘N

31∘30998400N

31∘N

30∘30998400N

116∘E 116∘30998400E 117∘E 117∘30998400E 118∘E 118∘30998400E

Figure 2 Study domain (the ldquo+rdquo represents the Hefei radar loca-tion)

Table 2 Error analysis of the JunendashAugust 2010 data

Error Exp I Exp II Exp IIIBIAS (mm) minus1299 minus1124 minus0219RAE () 310 283 271RMSE (mm) 2191 2100 1877

Table 3 Error analysis of stations with tropical rain in the JunendashAugust 2010 period

Error Exp I Exp II Exp IIIBIAS (mm) minus1912 minus1791 0041RAE () 323 310 227RMSE (mm) 3354 3314 2327

with Table 2 the statistical indicators in Table 3 do not con-sider data from grid points not experiencing tropical rainfallThe results in Table 3 match those of Table 2 in that radarQPE is noticeably improved using the FL method proposedhereThis improvement becomes evenmore significant whenin the presence of tropical rainfall events

To further understand the various precipitation eventsthat took place in JunendashAugust 2010 and better describe howQPE is affected under three rainfall conditions (stratiformconvective and a hybrid precipitation case in which convec-tive and stratiformprecipitation are present at the same time)the three-month period was divided into 12 precipitation


Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

05

00

minus05

minus10

minus15

BIA

S

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

030

020

010

000

RAE

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

30

25

20

15

10

05

00

RMSE

(c)Figure 3 Error statistics considering all stations for each precipitation process Label ldquo2clsrdquo represents classification in two precipitationclasses (Experiment II) and ldquo3clsrdquo represents classification in three precipitation classes (Experiment III) The units of BIAS and RMSE aremm RAE is dimensionless

cases (Table 4) The analysis of the several events is shownin Figure 5 which presents the error curves for each case Asshown Experiment III consistently produced the lowest errorvalues (with the exception of case 3) particularly in cases 1 6and 7

Figure 4 shows the same type of information in Fig-ure 3 but this time considering only tropical rainfall eventsThus as shown separating precipitation into convective and

stratiform using the FL algorithm produces better resultsthan simply using the default Z-R relationship althoughseparating the precipitation into convective stratiform andtropical rain produces results even closer to the observedprecipitation not only for the overall precipitation processbut also for the heavy precipitation portion only

In cases 1 6 and 7 Experiment III produced the bestresults however its results for case 3 are not very good


Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

BIA

S20

10

00

minus10

minus20

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

DefaultRA

E

040

030

020

010

000

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

RMSE

60

40

20

00

(c)

Figure 4 Error statistics considering only tropical rain events Label ldquo2clsrdquo represents classification in two precipitation classes (ExperimentII) and ldquo3clsrdquo represents classification in three precipitation classes (Experiment III) The units of BIAS and RMSE are mm RAE isdimensionless

These four cases were therefore selected to further studythe different results of the three sets of experiments Thecorresponding errors for each station are shown in Figure 5as a function of time cases 1 3 6 and 7 are represented fromthe top to the bottom row respectively and the BIAS RAEand RMSE values are represented in columns from left toright respectively It should be noted that these results are

the same ones presented in Figure 3 as shown in cases 1 6and 7 Experiment III not only produced the smallest totalerrors but also consistently exhibited the lowest errors onan hourly basis However the results of Experiment III arenot better than those of II or I in case 3 Further analysisrevealed that case 3 represents an intermittent rainfall eventthat was primarily formed by scattered convective cells in


3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60

Figure 5 Hourly errors for all stations Cases 1 3 6 and 7 The abscissas represent rainfall duration (unit h)


Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)

Figure 6 Scatter plot of gauge versus radar-derived 6-h accumulated precipitation estimates (case 3 uses 4-h accumulated precipitationvalues) (a) Case 1 (b) Case 3 (c) Case 6 (d) Case 7 (The diagonal line represents the ideal line corresponding to an equality between theradar and gauge estimates)

which either the precipitation ranges or the rainfall durationswere scattered with each rainfall lasting less than 6 h andwith limited rainfall per hour The shortcomings in case 3may also relate to the selected reflectivity-rainfall rate (Z-R) relationships as these were obtained by statistics relatedto seasonal climate and geography Therefore further workwill have to be performed to better capture local real-timedynamic Z-R relationships in the future

To better understand the differences between gaugemeasurements and QPE in terms of the amount of pre-cipitation scatter plots of gauge versus radar-derived 6-haccumulated precipitation estimates except for case 3 forwhich 4-h accumulated precipitation values are used becausein this case the rainfall lasted less than 6 h are shown inFigure 6 Figure 6(a) shows the results of case 1 in which

the strongest precipitation process intensity is 14mmh It isseen that the precipitation produced by all three experimentscorresponds well with observations in particular the rainfallestimate from Experiment III and that most of the resultsapproximate the ideal line In addition smaller rainfallintensities correspond to bigger errors particularly at rainfallintensities less than 2mmh Figure 6(b) shows the resultsfor case 3 in which either the precipitation ranges or rainfalldurations were scattered It is clear that the radar-retrievedaccumulation precipitation estimations produced in Experi-ment III mostly improved the problem of underestimationalthough overestimation is demonstrated in some stationsMoreover the results from Experiment II and Experiment Iare similar in that convective rainfall precipitation processesare mainly obtained from the FL method classification a


Table 4 Twelve precipitation events selected for performance evaluation

Event Precipitation duration Binary classification Tropical rain present1 June 07ndash09 Stratiform Yes2 June 13-14 Stratiform Yes3 June 17ndash20 Convective Yes4 June 23-24 Stratiform Yes5 June 27ndash30 Stratiform Yes6 July 02ndash04 Convective and stratiform Yes7 July 06ndash19 Convective and stratiform Yes8 July 21-22 Convective and stratiform Yes9 July 24ndashAugust 02 Convective and stratiform Yes10 August 04ndash11 Convective and stratiform Yes11 August 14ndash20 Convective and stratiform Yes12 August 21ndash31 Convective and stratiform Yes

result consistent with Table 4 Figure 6(c) shows the resultsfor case 6 in which the precipitation area is broader and thestrongest intensity is 70mmh The correspondence betweengauge and QPE improves from Experiment I to ExperimentIII Figure 6(d) shows the results for case 7 in which thestrongest intensity is 60mmh It is clear that the precipitationfor all three experiments corresponds well with observationsespecially in the case of Experiment III which shows asignificantly reduced underestimation when compared withExperiments I and II

4 Case Studies

Thecases of heavy rainfall events inHefei (valid at 0800UTCJuly 22 2009) and Liuzhou (valid at 1300UTC June 12 2008)were analyzed to test how easily the proposed method couldbe applied in different places and timesThe LiuzhouDopplerradar (109456∘E 24357∘N 3468m) is of the same type as theHefei Doppler radar

41 Heavy Rainfall in Hefei on July 21ndash23 2009 Most parts ofthe Yangtze RiverndashHuaihe River valley and the southern partof the Yangtze River experienced heavy rainfall on July 21ndash232009 [25] with local downpours of approximately 150mm ofrain During this period there was a wide range of precipi-tation including heavy rain that was distributed primarily inthe north of Anhui Province with the precipitation intensityreaching 40mmh

The distributions of convective and stratiform rainfallbased on the classification of the FL algorithm are shown inFigure 7(a) which shows a rainfall process overall dominatedby widespread stratiform rainfall embedded with convectiverainfall A stratiform cloud covering a large area appearedthroughout the entire precipitation process and was relativelystable Only a few parts of Hefei could be identified asconvective rainfall regions embedded in a wide region ofstratiform rainfall The convective rainfall regions changedfrequently and their coverage area was very limitedThus therange of estimated precipitation is relatively large

Although the problem of underestimation is mitigated tosome extent by classifying the precipitation into convective

and stratiform precipitation significant differences betweenobservation and estimation remain the primary objectiveof classifying precipitation into convective stratiform andtropical precipitation is the improvement of QPE accuracyFigure 7(b) shows the results obtained for the three types ofcloud classification It is clear that some convective and strati-form precipitation obtained from the FL algorithms are againidentified as tropical precipitation with the correspondingcases redefined as a stratiform convective or tropical mixedprecipitation

As can be seen from Figure 7(c) the characteristics ofthe precipitation event in Hefei on July 21ndash23 2009 includethe presence of a bright band it was identified as convectiveand stratiform mixed precipitation using the FL algorithmand the VPR slope after bright band correction was found tobe negative in other words tropical rainfall may have beenpresent in this case

To demonstrate the differences between the three exper-imentsrsquo results a scatter plot of gauge versus radar-derived1-h accumulated precipitation is shown in Figure 7(d) Eventhough all three experiments produce some degree of under-estimation the correspondence between gauge and radar-derived 1-h accumulated precipitation is improved fromExperiment I (greatest underestimation) to Experiment III(least underestimation) Thus Experiment III produces asignificantly more accurate estimation of rainfall comparedto the other two experiments particularly when the rainfallintensity is high

42 Liuzhou Rainfall Event during June 11-12 2008 Theprimary difference between the Liuzhou precipitation eventand theHefei event is that the Liuzhou precipitation case tookplace at a different location

Liuzhou in the Guangxi Province received heavy rain at0800UTC on June 11 2008 The 12-h rainfall accumulationfrom 0400 to 1600UTC on June 12 reached 233mm withheavy rainfall (307mm) also occurring in Liujiang CountyThe water level in Liujiang exceeded the warning level of286m until 1900UTC on June 12 The heavy rainfall causedwaterlogging in some areas and a number of road trafficinterruptions until June 13 when the rain began to ease


333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)

Figure 7 Hefei heavy rain (valid at 0800UTC July 22 2009) (a) Results of the FL algorithm (the abscissas represent longitude the verticalaxis represents latitude the ldquo+rdquo sign represents the radar location and the colors of ldquoConvrdquo and ldquoStrardquo represent convective and stratiformrainfall regions resp) (b)Distribution of the convective stratiform and tropical rainfall (c) Averaged stratiformVPR (the abscissas representthe value of reflectivity in dBZ the vertical axis represents the altitude in km the long dashed line is the 0∘C isotherm altitude the red solidline is the VPR before bright band correction and the blue short dashed line is the VPR after bright band correction) (d) Scatter plot of gaugeversus radar-derived 1-h accumulated precipitation (the abscissas represent gauge results in mm the vertical axis represents radar-derived1-h accumulated precipitation in mm and the diagonal line represents the ideal line)

Figure 8 shows an analysis of the heavy rainfall in Liuzhouduring June 11-12 2008 valid at 1300UTC on June 12Comparing the 1-h rainfall accumulation from the gaugeswith the convective and stratiformdistribution obtained fromthe FL algorithm (Figure 8(a)) it is clear that the entireprocess was based primarily on convective rainfall that didnot aggregate butwas found in the vicinity of the radarHeavyrainfall reached 25mmh just next to the radar

The distributions of the three types of rainfall are illus-trated in Figure 8(b) which shows that the Liuzhou eventwas amixed stratiform convective and tropical precipitationcase Most of the convective and stratiform precipitationobtained from the FL algorithm is again distinguishable astropical rain although there is some convective precipitationnear the radar which recognized the area as being dominatedby convective precipitation


ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls

(d)Figure 8 Heavy rainfall in Liuzhou during June 11-12 2008 as recorded at (or analyzed for) 1300UTC on June 12 (a) results of the FLalgorithm (the abscissas represent longitude the vertical axis represents latitude the ldquo+rdquo sign represents the radar location and colors ofldquoConvrdquo and ldquoStrardquo represent convective and stratiform rainfall resp) (b) Distribution of convective stratiform and tropical rainfall (c)Averaged stratiform VPR (the abscissas represent the value of reflectivity in dBZ the vertical axis represents the altitude in km the longdashed line is the 0∘C isotherm altitude the red solid line is the VPR before bright band correction (d) Scatter plot of gauge versus radar-derived 1-h accumulated precipitation (the abscissas represent gauge results inmm the vertical axis represents radar-derived 1-h accumulatedprecipitation in mm and the diagonal line represents the ideal line)

As the VPR shown in Figure 8(c) implies no bright bandis observed for this event The value of h 0∘C is 50426m theVPR slope is negative and tropical precipitation is possiblypresent

For comparison between the three experiments a scatterplot of gauge versus radar-derived 1-h accumulated precipita-tion is provided in Figure 8(d) It is clear that the precipitationof all three experiments corresponds well with observations

Experiment II performs better than Experiment I and Exper-iment III performs the best of all particularly for large rainfallintensity Thus the rainfall estimated by Experiment III isfound to correspond most closely to the observed rainfall

5 Conclusions

This paper introduced a method for improving the precisionof radar QPE by using an FL algorithm to divide events


into stratiform and convective precipitation types and thenfurther identifying tropical rain based on the VPR character-istics First the proposed method corrects any bright bandsthat are present Volume scan reflectivity data from Dopplerradar are used to recognize tropical VPRs using the precip-itation classification technique of Xu et al [33] Volume scanreflectivity data from Doppler radar are then interpolatedto the Cartesian coordinate system over the same latitudeand longitude range (001∘) and the rain is divided intostratiform precipitation and convective precipitation usingan FL algorithm [30] The precipitation is then consideredto be tropical rainfall if it has a tropical VPR and the gridat an altitude of 1 km meets the following criteria (1) thereflectivity is greater than 25 dBZ and (2) the surface wetbulb temperature is greater than 2∘C The reflectivity at 1 kmis used to assign an adaptive Z-R relationship Three setsof experiments were presented in this paper In ExperimentI for each weather event only the default WSR-88D Z-Rrelationship (119885 = 300R14) was used In Experiment II therain type was identified as convective or stratiform using anFL algorithm then the relationships 119885 = 300R14 and 119885 =200R16 were applied for convective and stratiform rainfallcalculations respectively In Experiment III stratiform andconvective rainfall in tropical rainfall events were furtherdivided into stratiform and tropical-stratiform rainfall andconvective and tropical-convective rainfall respectively Therelationships adopted were the same as those used in Exper-iment II except that the tropical rainfall relationship was 119885= 307R166 A long sequence (three months) of data and twoadditional study cases were conducted in order to determinethe impact on QPE of the method devised for this study Theconclusions of the study can be summarized as follows

(1) Although most radar QPEs were lower than thedirectly observed precipitation for all the rainfallevents analyzed in this study the results of all threeexperiments corresponded closely to the observedprecipitation Separating the precipitation into con-vective and stratiform types using the FL algorithmenabled better performances compared to simplyusing the default Z-R relationship Separating theprecipitation into convective stratiform and tropicalrain produced results that were the closest to theobserved precipitation not only during the wholeprecipitation processes but also individually withinthe heavy precipitation portions of these events

(2) Given that a large-scale precipitation event usuallycontains multiple types of precipitation large esti-mation errors can be introduced if a single Z-Rrelationship is used for the entire process The FLalgorithm used here can skillfully separate convec-tive precipitation from stratiform precipitation andthus improve the accuracy of radar QPE to someextent Nevertheless the problem of underestimationremains This study built upon the characteristics oftropical precipitation and the convective stratiformprecipitation classification techniques using an FLalgorithm the resulting method was shown to have

improved accuracy not only during the whole pre-cipitation processes but also individually within theheavy precipitation portions of these events

(3) The proposed tropical rain identification schemeproduced similar results when applied to the casesof heavy rainfall in Hefei (valid at 0800UTC onJuly 22 2009) and in Liuzhou (valid at 1300UTC onJune 12 2008) In future work the newly proposedscheme will be further tested to evaluate its generalapplicability to different times and locations

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This research was supported by the National Science Founda-tion of China (nos 41175092 and 41375109)

References

[1] K P Georgakakos ldquoCovariance propagation and updating inthe context of real-time radar data assimilation by quantitativeprecipitation forecastmodelsrdquo Journal ofHydrology vol 239 no1ndash4 pp 115ndash129 2000

[2] Z Pu and W-K Tao ldquoMesoscale assimilation of TMI rainfalldata with 4DVAR sensitivity studiesrdquo Journal of the Meteoro-logical Society of Japan vol 82 no 5 pp 1389ndash1397 2004

[3] C S Gardner and W Yang ldquoOperational C-band dual-polarization radar QPE for the subtropical complex terrain ofTaiwanrdquo Advances in Meteorology no 11 pp 1ndash15 2016

[4] M Gocic S Shamshirband Z Razak D Petkovic S Ch andS Trajkovic ldquoLong-term precipitation analysis and estimationof precipitation concentration index using three support vectormachine methodrdquo Advances in Meteorology vol 2016 ArticleID 7912357 11 pages 2016

[5] J Kalogiros M Anagnostou F S Marzano et al ldquoMobile radarnetwork measurements for flood applications during the fieldCampaign of HydroRad Projectrdquo in Advances in MeteorologyClimatology and Atmospheric Physics pp 137ndash143 Springer2013

[6] J S Marshall and W M K Palmer ldquoThe distribution ofraindrops with sizerdquo Journal of the Atmospheric Sciences vol 5no 4 pp 165ndash166 1948

[7] D P Jorgensen and P T Willis ldquoA Z-R relationship forhurricanesrdquo Journal of Applied Meteorology vol 21 no 3 pp356ndash366 1982

[8] P Tabary A-A Boumahmoud H Andrieu et al ldquoEvaluationof two lsquointegratedrsquo polarimetric Quantitative Precipitation Esti-mation (QPE) algorithms at C-bandrdquo Journal of Hydrology vol405 no 3-4 pp 248ndash260 2011

[9] S M Hunter ldquoWSR-88D radar rainfall estimation capabilitieslimitations and potential improvementsrdquo National WeatherDigest vol 20 no 4 pp 26ndash38 1996

[10] C W Ulbrich and L G Lee ldquoRainfall measurement errorby WSR-88D radars due to variations in Z-R law parametersand the radar constantrdquo Journal of Atmospheric and OceanicTechnology vol 16 no 8 pp 1017ndash1024 1999


[11] B E Vieux and P B Bedient ldquoEstimation of rainfall forflood prediction from WSR-88D reflectivity a case study 17-18October 1994rdquoWeather and Forecasting vol 13 no 2 pp 407ndash415 1998

[12] F Tian M Cheng Y Zhang et al ldquoAn investigation into theeffect of rain gauge density on estimating the areal rainfall usinga radar-gauge calibration algorithmrdquoActaMeteorological Sinicavol 68 no 5 pp 717ndash730 2010

[13] X Wei Research on self-adaptive optimized algorithm for radarquantificationally estimating precipitation [PhD dissertation]Nanjing University of Information Science amp Technology 2015

[14] Q Zeng H He Y Zhang et al ldquoParameterizations for theraindrop size distribution and its application in radar rainfallestimationrdquo Meterology and Disaster Reduction Research vol38 no 4 pp 46ndash53 2015

[15] E N Anagnostou ldquoA convectivestratiform precipitation clas-sification algorithm for volume scanning weather radar obser-vationsrdquoMeteorological Applications vol 11 no 4 pp 291ndash3002004

[16] J Gao and D J Stensrud ldquoAssimilation of reflectivity data in aconvective-scale cycled 3DVAR framework with hydrometeorclassificationrdquo Journal of the Atmospheric Sciences vol 69 no 3pp 1054ndash1065 2012

[17] M Xue F Kong K WThomas et al ldquoPrediction of convectivestorms at convection-resolving 1 km resolution over continentalunited states with radar data assimilation an example case of26 May 2008 and precipitation forecasts from spring 2009rdquoAdvances in Meteorology vol 2013 Article ID 259052 9 pages2013

[18] J Sun ldquoConvective-scale assimilation of radar data progressand challengesrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 131 no 613 pp 3439ndash3463 2006

[19] YQi J Zhang and P Zhang ldquoA real-time automated convectiveand stratiform precipitation segregation algorithm in nativeradar coordinatesrdquoQuarterly Journal of the RoyalMeteorologicalSociety vol 139 no 677 pp 2233ndash2240 2013

[20] Y Qi J Zhang Q Cao Y Hong and X-M Hu ldquoCorrection ofradarQPE errors for nonuniformVPRs inmesoscale convectivesystems using TRMM observationsrdquo Journal of Hydrometeorol-ogy vol 14 no 5 pp 1672ndash1682 2013

[21] Y Qi and J Zhang ldquoCorrection of radar QPE errors associatedwith low and partially observed brightband layersrdquo Journal ofHydrometeorology vol 14 no 6 pp 1933ndash1943 2013

[22] Y Qi J Zhang B Kaney C Langston and K HowardldquoImproving WSR-88D radar QPE for orographic precipitationusing profiler observationsrdquo Journal of Hydrometeorology vol15 no 3 pp 1135ndash1151 2014

[23] J Zhang K Howard C Langston et al ldquoMulti-Radar Multi-Sensor (MRMS) quantitative precipitation estimation initialoperating capabilitiesrdquo Bulletin of the American MeteorologicalSociety vol 97 no 4 pp 621ndash638 2016

[24] S Chumchean A Seed and A Sharma ldquoAn operationalapproach for classifying storms in real-time radar rainfallestimationrdquo Journal of Hydrology vol 363 no 1ndash4 pp 1ndash172008

[25] P P Mapiam and N Sriwongsitanon ldquoClimatological Z-Rrelationship for radar rainfall estimation in the upper Ping riverbasinrdquo ScienceAsia vol 34 no 2 pp 215ndash222 2008

[26] TN RaoDN Rao KMohan and S Raghavan ldquoClassificationof tropical precipitating systems and associated Z-R relation-shipsrdquo Journal of Geophysical Research Atmospheres vol 106 no16 pp 17699ndash17711 2001

[27] M I Biggerstaff and S A Listema ldquoAn improved scheme forconvectivestratiform echo classification using radar reflectiv-ityrdquo Journal of AppliedMeteorology vol 39 no 12 pp 2129ndash21502000

[28] D D Churchill and R A Houze Jr ldquoDevelopment andstructure of winter monsoon cloud clusters on 10 December1978rdquo Journal of the Atmospheric Sciences vol 41 no 6 pp 933ndash960 1984

[29] M Steiner R A Houze and S E Yuter ldquoClimatologicalcharacterization of three-dimensional storm structure fromoperational radar and rain gauge datardquo Journal of AppliedMeteorology vol 34 no 9 pp 1978ndash2007 1995

[30] Y Yang X Chen and Y Qi ldquoClassification of convec-tivestratiform echoes in radar reflectivity observations using afuzzy logic algorithmrdquo Journal of Geophysical Research Atmo-spheres vol 118 no 4 pp 1896ndash1905 2013

[31] L L Yang Y Yang Y C Qi X X Qiu and Z Q Gong ldquoA casestudy of radar-derived quantitative precipitation estimationrdquoAdvancedMaterials Research vol 726ndash731 pp 4541ndash4546 2013

[32] B Wang J Zhang W Xia et al ldquoAnalysis of radar and gaugerainfall during the warm season in Oklahomardquo in Proceedingsof the 22nd Conference on Hydrology New Orleans La USA2008

[33] X Xu K Howard and J Zhang ldquoAn automated radar tech-nique for the identification of tropical precipitationrdquo Journal ofHydrometeorology vol 9 no 5 pp 885ndash902 2008

[34] Y Qi J Min J Zhang et al ldquoStudy of radar quantitativeprecipitation estimation in summer of the Yangtze River-Huaihe River valleyrdquo in Proceedings of the 27th Annual Meetingof ChineseMeteorological Society on Prectice of Radar TechnologyDevelopment and Application 2010

[35] L Zhang L Chu F Ye et al ldquoThe automatic identification ofrainfall type by using radar datardquo Transactions of AtmosphericSciences vol 35 no 1 pp 95ndash102 2012

[36] Q Cao Y Hong J J Gourley et al ldquoStatistical and phys-ical analysis of the vertical structure of precipitation in themountainous west region of the united states using 11+ yearsof spaceborne observations from TRMM precipitation radarrdquoJournal of Applied Meteorology and Climatology vol 52 no 2pp 408ndash424 2013

[37] Q Cao and Y Qi ldquoThe variability of vertical structure ofprecipitation inHuaihe River Basin of China implications fromlong-term spaceborne observations with TRMM precipitationradarrdquo Water Resources Research vol 50 no 5 pp 3690ndash37052014

[38] J Zhang and Y Qi ldquoA real-time algorithm for the correction ofbrightband effects in radar-derived QPErdquo Journal of Hydrome-teorology vol 11 no 5 pp 1157ndash1171 2010

[39] J Zhang C Langston and K Howard ldquoBrightband identifica-tion based on vertical profiles of reflectivity from theWSR-88DrdquoJournal of Atmospheric ampOceanic Technology vol 25 no 10 pp1859ndash1872 2008

[40] L Tang J ZhangC Langston J Krause KHoward andV Lak-shmanan ldquoA physically based precipitation-nonprecipitationradar echo classifier using polarimetric and environmental datain a real-time national systemrdquoWeather and Forecasting vol 29no 5 pp 1106ndash1119 2014

[41] Y Qi J Zhang P Zhang and Q Cao ldquoVPR correction ofbright band effects in radar QPEs using polarimetric radarobservationsrdquo Journal of Geophysical Research Atmospheres vol118 no 9 pp 3627ndash3633 2013


[42] Y Xiao L Liu Z Li et al ldquoAutomatic recognition andremoval of the bright band using radar reflectivity datardquo PlateauMeteorology vol 29 no 1 pp 197ndash205 2010

[43] X Gao J Liang and C Li ldquoRadar quantitative precipitationestimation techniques and effect evaluationrdquo Journal of TropicalMeteorology vol 28 no 1 pp 77ndash88 2012

[44] Y Tang J Gan L Zhao and K Gao ldquoOn the climatologyof persistent heavy rainfall events in Chinardquo Advances inAtmospheric Sciences vol 23 no 5 pp 678ndash692 2006

[45] Q Zhang C Liu C-Y Xu Y Xu and T Jiang ldquoObserved trendsof annual maximumwater level and streamflow during past 130years in the Yangtze River basin Chinardquo Journal of Hydrologyvol 324 no 1ndash4 pp 255ndash265 2006

[46] Q Zhang C-Y Xu Z Zhang Y D Chen C-L Liu and HLin ldquoSpatial and temporal variability of precipitation maximaduring 1960ndash2005 in the Yangtze River basin and possibleassociation with large-scale circulationrdquo Journal of Hydrologyvol 353 no 3-4 pp 215ndash227 2008

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of


EarthquakesJournal of


Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining


Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of


Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering


GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of


OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in


MineralogyInternational Journal of


MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Geological ResearchJournal of


Geology Advances in


Convective rainfall systems are associatedwith strong verticalvelocity fields small areal coverage and high rainfall inten-sities and an echo that can be extended to a relatively highaltitude andwith a high center ofmass Conversely stratiformrainfall systems have relatively weak vertical velocity fieldsgreat horizontal homogeneity and low rainfall intensityAlthough the intensity of stratiform rainfall is much weakerthan that in adjacent convective cells stratiform rain typicallycovers a larger area and always contributes with a significantportion (40ndash50) of the total rainfall even in convective-dominant systems [15] Tropical rain systems generally occurin lower altitudes and very humid air with condensationoccurring continuously throughout the air descentThemainfeature of tropical rain systems is that their reflectivityincreases monotonically as height decreases yielding a max-imum reflectivity at the lowest levels of the vertical profileof reflectivity (VPR) resulting in considerable error whenconducting precipitation estimation using the default Z-Rrelationship ofWSR-88DMoreover multiple types of precip-itation may coexist within a wide range of rainfall events

Radar research scientists worldwide have undertakenextensive studies to improve the accuracy of radar QPE [16ndash23] Typically precipitation is first classified into varioustypes and corresponding multiple Z-R relationships are thenused [19 24ndash26] Precipitation can generally be classified intoeither convective or stratiform regimes which has been aparticular focus of many studies [27ndash29] Such basic classi-fications typically adopt a fixed threshold or set of boundaryconditions for rainfall recognition However these methodsare known to be sensitive to the used threshold values andbecause there is a considerable overlap between stratiformand convective regimes in many respects using fixed bound-ary conditions or thresholds often leads to false positivesMoreover it is often difficult to directly identify the boundarybetween stratiform and convective regimes [30] Yang et al[30] used the reference values of Steiner et al [29] to derive astatistical relationship between four characteristic parametersand two rain types in order to develop the probability distri-bution of convective and stratiform rainfall based on a fuzzy-logic (FL) algorithm which was able to improve the accuracyof QPE [31] Despite efforts such as these many techniquescan still underestimate rainfall to some extent Based on thestratiform and convective precipitation classification manyresearchers have suggested focusing on tropical rain as anavenue to mitigate the underestimation of QPE and somestudies have been able to reduce QPE underestimation byemploying an automated radar technique to identify tropicalprecipitation in order to divide precipitation into tropicalrain and stratiform and convective types [32 33] Qi et al [34]converted reflectivity into precipitation utilizing the tropicalrain Z-R relationship instead of the default WSR-88D Z-Rrelationship which improved the accuracy to some extentSimilarly Zhang et al [35] improved the accuracy of radar-derived QPE by using an automatic identification methodto identify tropica1 stratiform and convective precipitationtypes based on radar observations and model outputs Theythen applied various Z-R relationships in order to convertreflectivity into rain rates based on precipitation types The

results showed that their method improved the accuracy ofradar-derived QPE

In this study precipitation is first divided into stratiformand convective precipitation types using the FL algorithmsproposed by Yang et al [30] and then tropical rain is identi-fied based on the VPR characteristics [36 37] as discussed byXu et al [33] Tropical rainfall is further divided into tropical-stratiform and tropical-convective rainfall and the respectiveadaptive Z-R relationships are then applied for QPE Athree-month dataset is used to compare the newly proposedtropical rain identification scheme with the results obtainedwith the default WSR-88D Z-R relationship and a convec-tivestratiform rain identification scheme Section 2 describesthe precipitation classificationmethod and the correspondingZ-R relationships and discusses the benefits of precipitationclassification Section 3 illustrates the evaluation results ofusing long-term datasets using descriptive statistics Two casestudies are also provided in Section 4 to evaluate the perfor-mance of the proposed precipitation classification methodFinally Section 5 presents the main results of the study

2 Algorithm Description

Underestimation of rainfall can occur when precipitation isdivided into general stratiform and convective types Thisstudy aims to improve the accuracy of radar QPE by buildingupon the convective stratiform and tropical precipitationtechnique described by Xu et al [33] The proposed methoduses volume scan reflectivity data from a Doppler radarto recognize tropical VPRs by first dividing rain data intostratiform and convective precipitation (using an FL algo-rithm) and then classifying tropical rain events as stratiformconvective or tropical precipitation Figure 1 shows a flowchart illustrating this process of precipitation classification

21 Identification of Tropical VPRs This study uses VPRsto identify the potential presence of tropical rain As afirst step volume scan reflectivity data from Doppler radarare simply classified into assumed stratiform or convectivetypes based on the vertically integrated liquid (VIL) watercontent In particular the data are classified as belonging tothe stratiform type when the VIL value is below 65 kgm2[38] otherwise they are classified as convective Stratiformreflectivity observations larger than 10 dBZ from all tilts inan annular region between two predefined ranges 1199031 and1199032 (with the empirical values of 20 and 80 km resp) areused to obtain an averaged VPR [39] The number of datasamples within a 20-km radius of the radar is limited owingto a terrain clearance rule incorporated into a quality control(QC) scheme [40] whereby any radar bin whose bottomis within 50m of the ground is removed Furthermore theregion of interest has to be far enough from the radar receiverto avoid the cone of silence in the receiverrsquos immediatevicinity The accuracy of precipitation classification is alsolimited by the decrease in vertical resolution beyond an 80-km radius

The second step combines the output of the first step withthe method of bright band identification proposed by Zhang


Start



Correct BB


Polar VPR Hail



Yes

Yes

Yes

Yes

Yes

YesYes

No

No

No

No

NoNo

Do not correct BB

VIL lt 65 kgm2


Ref lt 50


3 of 5 120572(1205724 1205720) lt 0

dBZ


(1) ref gt 25


dBZ


(1) ref gt 25


dBZ

Pconv gt 05

(Pconv) with














(2)













120583119896119862 (119909 119886 119887) =


(119896 = 1 2 119873)(5)

and stratification

120583119896119878 (119909 119886 119887) =


(119896 = 1 2 119873) (6)




120583119896119878 (119909 119886 119887) = 1 minus 120583119896119862 (119909 119886 119887) (7)


119875119890 = (sum119873119896=1119882119896 sdot 119875119896119890)(sum119873119896=1119882119896) (8)




(9)






BIAS = 1119899119899sum119894=1

(119877119886 (119894) minus 119877119892 (119894)) (10)





RMSE = (sum119899119894=1 (119877119886 (119894) minus 119877119892 (119894))2119899 )12

(12)










reflectivity


Approximately6min





33∘N

32∘30998400N

32∘N

31∘30998400N

31∘N

30∘30998400N

116∘E 116∘30998400E 117∘E 117∘30998400E 118∘E 118∘30998400E









Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

05

00

minus05

minus10

minus15

BIA

S

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

030

020

010

000

RAE

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

30

25

20

15

10

05

00

RMSE







Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

BIA

S20

10

00

minus10

minus20

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

DefaultRA

E

040

030

020

010

000

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

RMSE

60

40

20

00

(c)





3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60



Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


Start



Correct BB


Polar VPR Hail



Yes

Yes

Yes

Yes

Yes

YesYes

No

No

No

No

NoNo

Do not correct BB

VIL lt 65 kgm2


Ref lt 50


3 of 5 120572(1205724 1205720) lt 0

dBZ


(1) ref gt 25


dBZ


(1) ref gt 25


dBZ

Pconv gt 05

(Pconv) with














(2)













120583119896119862 (119909 119886 119887) =


(119896 = 1 2 119873)(5)

and stratification

120583119896119878 (119909 119886 119887) =


(119896 = 1 2 119873) (6)




120583119896119878 (119909 119886 119887) = 1 minus 120583119896119862 (119909 119886 119887) (7)


119875119890 = (sum119873119896=1119882119896 sdot 119875119896119890)(sum119873119896=1119882119896) (8)




(9)






BIAS = 1119899119899sum119894=1

(119877119886 (119894) minus 119877119892 (119894)) (10)





RMSE = (sum119899119894=1 (119877119886 (119894) minus 119877119892 (119894))2119899 )12

(12)










reflectivity


Approximately6min





33∘N

32∘30998400N

32∘N

31∘30998400N

31∘N

30∘30998400N

116∘E 116∘30998400E 117∘E 117∘30998400E 118∘E 118∘30998400E









Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

05

00

minus05

minus10

minus15

BIA

S

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

030

020

010

000

RAE

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

30

25

20

15

10

05

00

RMSE







Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

BIA

S20

10

00

minus10

minus20

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

DefaultRA

E

040

030

020

010

000

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

RMSE

60

40

20

00

(c)





3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60



Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in










(2)













120583119896119862 (119909 119886 119887) =


(119896 = 1 2 119873)(5)

and stratification

120583119896119878 (119909 119886 119887) =


(119896 = 1 2 119873) (6)




120583119896119878 (119909 119886 119887) = 1 minus 120583119896119862 (119909 119886 119887) (7)


119875119890 = (sum119873119896=1119882119896 sdot 119875119896119890)(sum119873119896=1119882119896) (8)




(9)






BIAS = 1119899119899sum119894=1

(119877119886 (119894) minus 119877119892 (119894)) (10)





RMSE = (sum119899119894=1 (119877119886 (119894) minus 119877119892 (119894))2119899 )12

(12)










reflectivity


Approximately6min





33∘N

32∘30998400N

32∘N

31∘30998400N

31∘N

30∘30998400N

116∘E 116∘30998400E 117∘E 117∘30998400E 118∘E 118∘30998400E









Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

05

00

minus05

minus10

minus15

BIA

S

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

030

020

010

000

RAE

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

30

25

20

15

10

05

00

RMSE







Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

BIA

S20

10

00

minus10

minus20

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

DefaultRA

E

040

030

020

010

000

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

RMSE

60

40

20

00

(c)





3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60



Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in



120583119896119878 (119909 119886 119887) = 1 minus 120583119896119862 (119909 119886 119887) (7)


119875119890 = (sum119873119896=1119882119896 sdot 119875119896119890)(sum119873119896=1119882119896) (8)




(9)






BIAS = 1119899119899sum119894=1

(119877119886 (119894) minus 119877119892 (119894)) (10)





RMSE = (sum119899119894=1 (119877119886 (119894) minus 119877119892 (119894))2119899 )12

(12)










reflectivity


Approximately6min





33∘N

32∘30998400N

32∘N

31∘30998400N

31∘N

30∘30998400N

116∘E 116∘30998400E 117∘E 117∘30998400E 118∘E 118∘30998400E









Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

05

00

minus05

minus10

minus15

BIA

S

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

030

020

010

000

RAE

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

30

25

20

15

10

05

00

RMSE







Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

BIA

S20

10

00

minus10

minus20

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

DefaultRA

E

040

030

020

010

000

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

RMSE

60

40

20

00

(c)





3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60



Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in







reflectivity


Approximately6min





33∘N

32∘30998400N

32∘N

31∘30998400N

31∘N

30∘30998400N

116∘E 116∘30998400E 117∘E 117∘30998400E 118∘E 118∘30998400E









Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

05

00

minus05

minus10

minus15

BIA

S

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

030

020

010

000

RAE

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

30

25

20

15

10

05

00

RMSE







Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

BIA

S20

10

00

minus10

minus20

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

DefaultRA

E

040

030

020

010

000

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

RMSE

60

40

20

00

(c)





3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60



Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

05

00

minus05

minus10

minus15

BIA

S

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

030

020

010

000

RAE

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

30

25

20

15

10

05

00

RMSE







Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

BIA

S20

10

00

minus10

minus20

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

DefaultRA

E

040

030

020

010

000

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

RMSE

60

40

20

00

(c)





3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60



Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

BIA

S20

10

00

minus10

minus20

(a)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

DefaultRA

E

040

030

020

010

000

(b)

Case of rain1 2 3 4 5 6 7 8 9 10 11 12

3cls2cls

Default

RMSE

60

40

20

00

(c)





3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60



Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


3cls2cls

Default

0 30 60 90 120 150 180 210

BIAS

3cls2cls

Default

0 30 60 90 120 150 180 210

3cls2cls

Default

0 30 60 90 120 150 180 210

RMSERAE

0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60

0 10 20 30 40 0 10 20 30 40 0 10 20 30 40

0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Case

1Ca

se 3

Case

6Ca

se 7

15

10

05

00

minus05

minus10

minus15

minus20

060

050

040

030

020

010

000

24

20

16

12

08

04

00

80

60

40

20

00

80

60

40

20

00

80

60

40

20

00

minus20

minus40

070

060

050

040

030

020

010

000

080

060

040

020

000

060

050

040

030

020

010

000

10

8

6

4

2

0

40

20

00

minus20

minus40

minus60

60

30

00

minus30

minus60



Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


Gauge (mm)

Default2cls

3cls

14121086420

Rada

r (m

m)

14

12

10

8

6

4

2

0

14

12

10

8

6

4

2

0

14121086420

(a)

Rada

r (m

m)

Gauge (mm)

Default2cls

3cls

32

28

24

20

16

12

8

4

0

32

28

24

20

16

12

8

4

0

322824201612840

322824201612840

(b)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(c)

Rada

r (m

m)

70

60

50

40

30

20

10

0

70

60

50

40

30

20

10

0

706050403020100

706050403020100

Gauge (mm)

Default2cls

3cls

(d)









4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in





4 Case Studies











333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


333

33

327

324

321

318

315

312

309

306

116

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(a)

333

33

327

324

321

318

315

312

309

30611

6

1164

1168

1172

1176

118

1184

ConvTrop

Stra

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Default2cls

3cls

Rada

r

25

20

15

10

5

0

2520151050

252015105025

20

15

10

5

0

Gauge (mm)

(d)





ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


ConvTrop

Stra

258

255

252

249

246

243

24

237

234

231

108

1084

1088

1092

1096

110

1104

1108

(a)

258

255

252

249

246

243

24

237

234

231

ConvTrop

Stra

108

1084

1088

1092

1096

110

1104

1108

(b)

ℎ(k

m)

8

7

6

5

4

3

2

1

Ref (dBZ)0 5 10 15 20 25 30 35 40 45 50

(c)

Rada

r (m

m)

45

40

35

30

25

20

15

10

5

0

454035302520151050

45403530252015105045

40

35

30

25

20

15

10

5

0

Gauge (mm)

Default2cls

3cls





5 Conclusions








Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in







Competing Interests


Acknowledgments


References


























































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in
















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in

research article radar-derived quantitative precipitation...

Documents