Editorial Type:
Article
Article Type:
Research Article

A Physically Based Precipitation–Nonprecipitation Radar Echo Classifier Using Polarimetric and Environmental Data in a Real-Time National System

Lin Tang Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, and NOAA/OAR/National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by Lin Tang in
Current site
Google Scholar
PubMed Close
,
Jian Zhang NOAA/OAR/National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by Jian Zhang in
Current site
Google Scholar
PubMed Close
,
Carrie Langston Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, and NOAA/OAR/National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by Carrie Langston in
Current site
Google Scholar
PubMed Close
,
John Krause Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, and NOAA/OAR/National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by John Krause in
Current site
Google Scholar
PubMed Close
,
Kenneth Howard NOAA/OAR/National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by Kenneth Howard in
Current site
Google Scholar
PubMed Close
, and
Valliappa Lakshmanan Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, and NOAA/OAR/National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by Valliappa Lakshmanan in
Current site
Google Scholar
PubMed Close
Print Publication:
01 Oct 2014
DOI:
https://doi.org/10.1175/WAF-D-13-00072.1
Page(s):
1106–1119
Full access

Abstract

Polarimetric radar observations provide information regarding the shape and size of scatterers in the atmosphere, which help users to differentiate between precipitation and nonprecipitation radar echoes. Identifying and removing nonprecipitation echoes in radar reflectivity fields is one critical step in radar-based quantitative precipitation estimation. An automated algorithm based on reflectivity, correlation coefficient, and temperature data is developed to perform reflectivity data quality control through a set of physically based rules. The algorithm was tested with a large number of real data cases across different geographical regions and seasons and showed a high accuracy (Heidke skill score of 0.83) in segregating precipitation and nonprecipitation echoes. The algorithm was compared with two other operational and experimental reflectivity quality control methodologies and showed a more effective removal of nonprecipitation echoes and a higher computational efficiency. The current methodology also demonstrated a satisfactory performance in a real-time national multiradar and multisensor system.

Corresponding author address: Lin Tang, CIMMS, University of Oklahoma, 120 David Boren Blvd., Norman, OK 73072. E-mail: lin.tang@noaa.gov

Abstract

Polarimetric radar observations provide information regarding the shape and size of scatterers in the atmosphere, which help users to differentiate between precipitation and nonprecipitation radar echoes. Identifying and removing nonprecipitation echoes in radar reflectivity fields is one critical step in radar-based quantitative precipitation estimation. An automated algorithm based on reflectivity, correlation coefficient, and temperature data is developed to perform reflectivity data quality control through a set of physically based rules. The algorithm was tested with a large number of real data cases across different geographical regions and seasons and showed a high accuracy (Heidke skill score of 0.83) in segregating precipitation and nonprecipitation echoes. The algorithm was compared with two other operational and experimental reflectivity quality control methodologies and showed a more effective removal of nonprecipitation echoes and a higher computational efficiency. The current methodology also demonstrated a satisfactory performance in a real-time national multiradar and multisensor system.

Corresponding author address: Lin Tang, CIMMS, University of Oklahoma, 120 David Boren Blvd., Norman, OK 73072. E-mail: lin.tang@noaa.gov
Keywords: Precipitation; Algorithms; Radars/Radar observations

1. Introduction

A significant challenge in radar-derived quantitative precipitation estimation (QPE) is separating precipitation from nonprecipitation (NP) radar returns. NP echoes include electromagnetic interferences (EI) from man-made telecommunication transmitters and sun spikes (SS); ground clutter from anomalous propagation (AP) of the radar beam due to specific atmospheric temperatures and/or moisture gradients (Grecu and Krajewski 2000); biological scatterers (birds, bats, and insects) (Lakshmanan et al. 2010); ground clutter from surface structures, such as towers and buildings (Bachmann and Tracy 2009); and sea clutter (Gray and Larsen 2005). An effective quality control (QC) of radar reflectivity (Z) fields is critical in assuring realistic depiction of storm structure and accurate precipitation estimation from the radar data.

Before polarimetric radar techniques became widely available, heuristic rule–based (e.g., Zhang et al. 2004) and neural network–based [quality control neural network (QCNN)] (e.g., Lakshmanan et al. 2007, 2010) reflectivity QC techniques using Z, radial velocity V, spectrum width συ, and atmospheric environmental data were developed to segregate precipitation and nonprecipitation (P–NP) echoes. These techniques can remove a significant amount of NP returns under various scenarios. However, nocturnal biological echoes (“blooms”) during the peak of bird migration seasons, spring and fall, remained a major issue, especially when they are connected with precipitation echoes. On the other hand, light rain and snow echoes were sometimes removed as a result of the techniques trying to minimize bloom echoes. Manual QC on radar reflectivity field has been practiced at commercial weather companies and in operations such as the National Weather Service River Forecast Centers (http://www.emc.ncep.noaa.gov/mmb/ylin/pcpanl/stage4/). However, manual QC is an expensive resource and prohibitive for timely warnings of severe weather and flash floods. An automated approach is still desirable for real-time applications.

The polarimetric upgrade of the U.S. Weather Surveillance Radar-1988 Doppler (WSR-88D) network opened a new era for the radar data applications in many disciplines (Kumjian 2013b). With the upgrade, new variables of differential reflectivity ZDR, differential phase ΦDP and specific differential phase KDP, and correlation coefficient became available. These variables measure the difference and correlation between the intensities and phases of the power return signals in the horizontal and vertical channels, and provide useful information about the shape, size, and orientation of the scatterers in a given resolution volume. Based on these new measurements, hydrometeor classification algorithms (HCAs) with varying levels of complexity have been developed for radars of different wavelengths (e.g., Straka and Zrnić 1993; Vivekanandan et al. 1999; Straka et al. 2000; Liu and Chandrasekar 2000; Zrnić et al. 2001; Lim et al. 2005; Gourley et al. 2007; Marzano et al. 2008; Park et al. 2009; and many others). Park et al. (2009) developed a fuzzy logic–based HCA for WSR-88D that classifies radar echoes into 10 different hydrometeor and nonhydrometeor categories. This algorithm has been implemented as a part of the National Weather Service operational radar product baseline, and the output hydrometeor classification is used to determine which R(Z), R(Z, ZDR), or R(KDP) relationships (here R represents the precipitation rate) should be applied in the operational polarimetric QPE (Berkowitz et al. 2013). The HCA provides very good separation of P–NP echoes in general, although some issues remain due to nonexclusive membership functions in the fuzzy logics for different echo types (e.g., Liu and Chandrasekar 2000). For instance, the HCA may sometimes misclassify biological or clutter echoes as “big drops” (BD; Fig. 1b). It also has some difficulties with the identification of blooms or AP echoes beyond the range where the beam intersects the melting layer (Fig. 1b). The HCA involves all polarimetric radar variables, including Z, V, συ, , ZDR, and ΦDP, in multiple sophisticated procedures and requires significant computational resources.

Fig. 1.
Fig. 1.

(a) Base reflectivity and (b) corresponding hydrometeor classification fields on 0.5° tilt of KEWX observations valid at 0719:46 UTC 6 May 2012.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

Lakshmanan et al. (2013) developed a dual-polarization (dp) radar reflectivity quality control using a neural network approach (dpQCNN) to segregate NP from precipitation echoes. With inputs of the six variables (Z, V, συ, , ZDR, and ΦDP), the dpQCNN calculates a probability of precipitation for each pixel utilizing either a pre-trained neural network or a simple classifier based on Z, ZDR, and thresholds. A clustering postprocess is applied to group the range gates into precipitation or NP entities, and the NP entities are removed. It was shown that dpQCNN provides a good Heidke skill score (HSS) of 0.8 (Lakshmanan et al. 2013). However, because of the statistical scoring process and characteristics of the neural network model, dpQCNN’s performance is highly dependent on the datasets available for the training and neural network approaches in general may have limited ability to explicitly identify possible causal relationships (Tu 1996). The dpQCNN also requires all six of the polarimetric radar variables, and the computational expenses can be relatively high.

It is commonly recognized that the correlation coefficient shows distinctly different characteristics between P–NP echoes under most of the situations (e.g., Balakrishnan and Zrnić 1990; Doviak and Zrnić 1993; Zrnić and Ryzhkov 1999; Melnikov and Zrnić 2007; Kumjian 2013a). The quantity measures the diversity of the scatterers contributing to the two polarized radar returns within a given resolution volume. Generally, its value varies from less than 0.5 to 1 (Melnikov and Zrnić 2007) with uniform scatterers such as raindrops of similar sizes producing close to 1 (Kumjian 2013a) and nonuniform scatterers such as birds and insects producing values less than 0.75 (Zrnić and Ryzhkov 1999). Taking advantage of this unique feature, the current study aims at a simple -based P–NP segregation (dpQC) algorithm that can be applied in the real-time national multiradar and multisensor (MRMS) QPE system (Zhang et al. 2014). The MRMS QPE system ingests 146 WSR-88D and 30 Environment Canada weather radars and produces a suite of precipitation products at 1-km resolution and a 2-min update cycle. The large volumes of data and rapid product update rate require high computational efficiencies for the algorithms in the system.

The dpQC is a physically based scheme that applies a set of explicit meteorological principles according to , Z, and atmospheric thermodynamic fields. It combines a simple ρHV filter that separates P (high ρHV values) and NP (low values) areas, and applies a set of heuristic rules that handle exceptions to the basic filter. Such exceptions include areas of hail, nonuniform beam filling (NBF), and a melting layer (ML) that have low values, and random clutter and biological pixels with high values. The dpQC uses 3D reflectivity structure and environmental data to protect hail, NBF, and ML areas from the simple filter, and it uses spatial filters and vertical and horizontal consistency checks to remove random NP pixels that exhibit high values. The dpQC is computationally efficient, since the simple filter can flag a large portion of the data that does not require further processing. Other filters and spatial analysis steps are arranged sequentially, with less data in each step. The algorithm is based on general and explicit meteorological principles and is not sensitive to specific cases. The transparency of the algorithm makes the dpQC easy to maintain and update in a real-time environment.

The rest of this paper is organized as follows. Section 2 provides a detailed description of the dpQC methodology, and section 3 presents case study results and discussions. Conclusions and remarks are given in section 4.

2. Methodology

Figure 2 shows an overview flowchart of the dpQC algorithm, which consists of seven steps, including 1) a basic filter to set apart the “most likely” precipitation areas; 2) delineation of hail and NBF areas; 3) delineation of the ML area; 4) a texture filter to remove NP echoes associated with noisy structure; 5) a spike filter to remove NP echoes associated with the SS and EI; 6) AP removal through a vertical gradient and a horizontal smoothness check; and 7) a final cleanup procedure that removes speckles and fills in small holes in precipitation areas. The input data include Z, , and temperature profiles at the radar site. The output is the precipitation reflectivity field. Details of each step are presented below.

Fig. 2.
Fig. 2.

Flowchart of the dpQC algorithm.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

a. The basic filter

It is commonly known that pure rain and pure snow are associated with high ρHV values close to 1 and that NP scatterers generally produce low values. Figure 3a shows a base reflectivity map from KEWX at 0551:39 UTC 6 May 2012. The single-polarization radar QCNN (Lakshmanan et al. 2007) has difficulties in separating the precipitation from the NP echoes (Fig. 3c) because the reflectivity and velocity features of birds and insects are very similar to those of stratiform precipitation. On the other hand, values of the two groups are distinctly different (Fig. 3b). A simple filter with a threshold of 0.95 was remarkably effective in removing the blooms (Fig. 3d). The residual NP echoes in Fig. 3d are likely from random high values associated with insects (Doviak and Zrnić 1993) and from high pixels due to the correction bias from the noise variations (Melnikov and Zrnić 2007).

Fig. 3.
Fig. 3.

(a) Raw Z, (b) , (c) Z fields after the QCNN process, and (d) Z field after the basic filter (threshold = 0.95) from the 0.5° tilt of KEWX observations valid at 0551:39 UTC 6 May 2012.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

Figure 4 shows normalized histograms of , in which an experienced radar meteorologist manually separates the precipitation (indicated as black bars) and NP (indicated as gray bars) echoes. According to the case of KARX (Fig. 4a), the values associated with NP echoes spread widely between 0.3 and 0.95, while precipitation echoes have values generally greater than 0.95. However, this feature is not persistent across all precipitation regimes. In the case of a convective storm observed by KVNX at 0736 UTC 20 May 2011, the histograms (Fig. 4b) of P–NP echoes had a very noticeable overlapping, with values of the NP echoes spreading across the whole range of 0–1. At the same time, some precipitation echoes had values lower than 0.80 due to large hailstones. These precipitation echoes require special treatments in the dpQC as described below.

Fig. 4.
Fig. 4.

Histogram plots of the values associated with NP (gray) and P echoes (black), observed by (a) KARX at 1933 UTC 26 Jun 2012 and (b) KVNX at 0736 UTC 20 May 2011.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

b. Hail and NBF

Low values are often found in hail regions of convective storms or in areas of mixed hydrometeor types (rain/hail, rain/snow, etc.). They indicate more hydrometeor diversity and likely mixed-phase precipitation (e.g., Balakrishnan and Zrnić 1990; Ryzhkov et al. 2005, 2013a,b). A noticeable decrease in is also found when large gradients of occur within a pulse volume, a phenomenon known as NBF. The decrease caused by NBF increases with range as the radar beam broadens progressively (Ryzhkov 2007). In dpQC, these precipitation echoes are delineated and excluded from the NP decision based only on the general filter (Fig. 2).

Hail cores and NBF are generally found in intense precipitation events and are associated with strong echoes and high echo tops atypical of a clutter. When a data bin is associated with low (<0.95), the corresponding echo tops of 18 (ETOP18dBZ) and 0 dBZ (ETOP0dBZ) are checked, and the following criteria are applied to identify areas of hail, NBF, and a mixture of hydrometeors of different shapes and sizes:
e1

If a strong reflectivity echo (>45 dBZ) is associated with a reduction and the echo top of 18 dBZ is higher than 8.0 km [Eq. (1)], then the low is possibly associated with hailstones. NBF is commonly observed when a heavy precipitation core takes only a portion of the radar resolution volume or when a nonuniform mixture of hail and rain produces a gradient of precipitation within the beam (Ryzhkov 2007). The is reduced due to the spread or diversity of values within the sampling volume (Kumjian 2013c). In Eq. (2), the algorithm searches for any potential storm cores between a given low bin and the radar site, and the range of the storm core is denoted rstorm_core. A storm core is defined as any accumulated range of high reflectivity (Z > 45 dBZ) echoes that is longer than 1 km. A low bin is considered a possible NBF if Eq. (2) is met. The echoes identified as possible hail or NBF are exempt from the filter and texture filter (Fig. 2). Note that these criteria delineate not only areas of hail and NBF but also areas of mixed hydrometeors of different shapes and sizes. Figure 5 shows a precipitation event observed from KVNX at 0355:54 UTC 1 May 2012. The ρHV field (Fig. 5b) showed low values below 0.95 in areas of hail cores, NBF, and where mixed hydrometeor types might have coexisted. These areas are marked (maroon in Fig. 5d) and exempted from the filters. After the dpQC, these precipitation echoes were retained, while the biological echoes close to the radar were removed (Fig. 5d).

Fig. 5.
Fig. 5.

(a) Raw Z, (b) , and (c) fields from 0.5° tilt of the KVNX observations at 0355:54 UTC 1 May 2012. (d) The Z field after dpQC, where maroon shows low pixels from possible hail, NBF, and hydrometeors of mixed sizes/shapes.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

c. The ML

Relatively low is one of the polarimetric radar signatures of the ML, which consists of melting snow aggregates (stratiform precipitation) or melting graupel or hail (convective precipitation) (e.g., Brandes and Ikeda 2004; Giangrande et al. 2008). ML detection algorithms of various complexities have been proposed before (e.g., Gourley and Calvert 2003; Brandes and Ikeda 2004; Tabary et al. 2006; Giangrande et al. 2008) utilizing different combinations of polarimetric radar variables. The dpQC adopts a simple ML delineation based on the field and the temperature profile at the radar site. A “first guess” ML is assumed to be the 1-km-thick layer below the 0°C height in the atmosphere (Fabry and Zawadzki 1995; Zhang et al. 2008; Zhang and Qi 2010; Andrić et al. 2013). Radar bins along each radial are segregated into three groups (i.e., below, in, and above the first-guess ML), and the average from each group is calculated as follows:
e3
Here H0°C is the atmospheric 0°C height (km), h(i) is the height of the ith bin, and n is the total number of bins in a given group. The data below, within, and above the first-guess ML are checked and a “real” ML is found when the following condition is met:
e4

If a real ML is found, then the ML boundaries in a radial are determined by searching for the significant radial gradients of near the top and bottom of the first-guess ML. All echoes in the real ML with a value below 0.7 are then classified as NP and removed. The rest of the echoes are retained as “likely precipitation” for further processes (Fig. 2).

Figure 6 shows an example of the ML lineation during a winter precipitation event, where a black circle indicates the relatively low values associated with hydrometeors of mixed phases (Fig. 6b). A simple filter with a threshold of 0.95 would remove some of these precipitation echoes (Fig. 6c). But with the ML delineation (Fig. 6d), the precipitation echoes are retained.

Fig. 6.
Fig. 6.

(a) Raw Z and (b) fields from 1.45° tilt of the KCLE observations at 1508:34 UTC 26 Jan 2012. The black circle indicates the areas of low values in the ML, which would be removed by the filter with (c) a 0.95 threshold but would not be by (d) an adjusted filter with a 0.70 threshold after the ML delineation.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

d. The texture filter

The texture filter follows the filter to identify NP echoes associated with noisy local texture. Similar to the texture of Z and ΦDP used in the operational HCA (Park et al. 2009), texture in this study is defined as the standard deviation of in a 1-km radial segment. The texture characterizes small-scale fluctuations of along the radial. An echo is classified as NP when its texture is greater than 0.10. Data bins in areas of hail, NBF, and ML are exempted from the texture filter. After the basic and texture filters, the majority of NP echoes with low (<0.95) and noisy texture are removed.

e. The spike filter

While most SS and EI echoes are associated with low (<0.95), random pixels of high values are frequently found in such “spike” shaped echoes, making the filter ineffective. The predominant feature of SS–EI echoes is their continuity along the radial direction and discontinuity in the azimuthal direction. Further, the spikes usually occur in the lowest tilt and lack vertical consistencies. In the spike filter, the number of bins with Z > 0 dBZ are counted for each radial on the lowest two tilts (denoted as N1 for the lowest tilts and N2 for the second lowest tilts). If N1 is greater than 70% of the whole radial and N2 is less than 10% of N1, then the whole radial on the lowest tilt is identified as SS–EI. Further, if an echo of Z > 0 dBZ extends 30 km or longer in range but less than three radials in azimuth, then the echo is also identified as SS–EI and is removed.

f. Vertical gradient and horizontal smoothness checks

The vertical gradient check handles AP ground clutter and clear-air echoes around the radar. Some AP echoes are associated with values higher than 0.95 as shown in Fig. 7b and fail the basic filter. However, AP echoes are usually near the ground and lack vertical consistency. A vertical gradient check proposed by Zhang et al. (2004) is adopted in the current study to identify AP. A specific vertical gradient of reflectivity (VDZ) is defined as the difference between the Z under examination and the corresponding Z at the same range/azimuth bin on an upper reference tilt. If the echo intensity decreases significantly with height (e.g., >50 dBZ km−1), then the echo is identified as AP. The vertical gradient check is not fully effective when the beam is so severely ducted that large areas of strong AP occur on several of the lowest tilts. Further, it cannot be applied at very far ranges where the height difference between the lowest two tilts becomes too large, and shallow precipitation may only be observed in the first tilt. Given the random distribution of the high values in AP, dpQC applies a horizontal filter to further reduce the AP echoes. For a given nonmissing reflectivity bin (Z = Z0), all the neighboring data within a window of 1.25 km × 1.5 azimuth degrees are recorded. If more than one-half of these neighboring bins have a missing reflectivity, or the averaged reflectivity of nonmissing bins are less than 25% of Z0, then the given bin is considered random noise or AP and removed. Figure 7c shows an example where the dpQC successfully identified AP echoes associated with high values.

Fig. 7.
Fig. 7.

(a) Raw Z, (b) , and (c) Z fields after dpQC from 0.5° tilt of KOUN observations at 0208:28 UTC 11 May 2010. The red circles outline areas of severe AP.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

Clear-air echoes are sometimes associated with unrealistic high values (>1.0; Fig. 8b) due to low signal-to-noise ratios (SNR). Inaccurate measurements of the noise power in the horizontal and vertical channels may be one of the main factors causing the anomalously high (Melnikov and Zrnić 2007). Other factors such as inherent positive biases in the correlation coefficient estimator could also contribute to a greater than one (Ivić and Melnikov 2013). Vertical gradient and horizontal smoothness checks are able to identify the clutter that usually appears in areas around the radar. It should be noted that this step might become unnecessary after the ongoing research of correction approaches demonstrates success in operational radars. But the procedure may still be useful for processing historical WSR-88D radar data.

Fig. 8.
Fig. 8.

(a) Raw Z, (b) , and (c) Z fields after dpQC from 0.5° tilt of the KTLX observations at 1231:51 UTC 8 Mar 2013.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

g. The speckle filter and hole filling

This step performs two tasks: removing clear-air echoes of minor size and filling in small voids in precipitation regions. If the residual echoes of 10 dBZ or higher in the whole volume scan accumulate to less than 10 km2 in size, then the whole volume scan is assigned missing reflectivity. This process effectively cleans up clear-air pixels with biased high values in the upper tilts. Finally, a median filter is applied to fill in small voids within precipitation echoes where sporadic precipitation pixels may have exhibited low ρHV values and are removed by the basic filter in the earlier steps.

3. Evaluations of dpQC

Parameters of the dpQC algorithm were determined via a trial and error process based on 25 cases that represented various P–NP scenarios from different regions of the United States and across different seasons. There were a total of 138 h of data from 29 WSR-88D, as shown in Table 1. Experienced radar meteorologists carefully evaluated the performance of dpQC for each of the 25 cases (Table 1) and the parameters were tuned to assure correct P–NP classifications across all cases. Further, dpQC is evaluated using a set of independent cases (Table 2) along with dpQCNN (Lakshmanan et al. 2013). A composite reflectivity (CREF) field is generated from each volume scan in the independent cases, and radar meteorologists manually separated the CREF field into P–NP groups. The dpQC results are compared with the manually generated “truth” fields and the HSS is calculated as follows:
e5
where a represents the number of hits, b the false alarms, c the misses, and d the correct nulls. The dpQC algorithm performs well with an HSS of 0.83, compared with dpQCNN of 0.80. The dpQC algorithm (dpQCNN) takes on average 3.15 s (17.76 s) of CPU time and 83 MB (199 MB) of memory to process a volume scan on a workstation with four 2.27-GHz processors and 12-GB double data rate type 3 random access memory (DDR3 RAM).
Table 1.

List of cases used for the development and tuning of dpQC. BI: biological clutter; WF: wind farm; GC: ground clutter.

Table 1.
Table 2.

Testing cases for evaluating the QC algorithm.

Table 2.

The dpQC algorithm has been implemented in the MRMS system (Zhang et al. 2014) to process 146 WSR-88D since 5 December 2012. While computationally efficient, the dpQC scheme has been very effective in removing biological and other NP echoes and retaining light precipitation. Figure 9 shows example CREF mosaic fields from the real-time MRMS system before and after the dpQC. The raw CREF image (Fig. 9a) showed large areas of blooms because it was late night (~0330 local time) during the peak bird migration season (early October). The dpQCNN was able to remove the majority of the blooms, although some blooms connected with the precipitation were left in (red circles in Fig. 9b). Meanwhile a light rain near the KDMX radar (white circle in Fig. 9b) was removed. This may be an indication that the relative weights given to the reflectivity and features in dpQCNN require further training. The dpQC scheme was able to remove the blooms and to retain the light precipitation in this case (Fig. 9c) because of the distinctively different features in areas of blooms versus in light rain.

Fig. 9.
Fig. 9.

Real-time CREF fields (a) before any QC, (b) after dpQCNN, and (c) after dpQC valid at 0830 UTC 10 Oct 2013. The white circles indicate an area of light precipitation and the red circles are the areas of blooms.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

The dpQC scheme has been evaluated on a daily basis for over a year in the real-time MRMS system, and its performance was found to be very satisfactory. A few minor issues remain, though. Because of the difference in the maximum ranges of (300 km) and reflectivity (460 km) fields, the filter does not help beyond 300 km. AP echoes at these far ranges are still a challenge. One such example is shown in Figs. 10a–c. Fortunately, the data beyond the 300-km range are not critical for inland WSR-88D because the network is relatively dense and the average spacing between the neighboring radars is less than 250 km. For coastal radars, satellite data may be used to screen some of the residual AP echoes. It is noteworthy that the dpQC was able to handle most of the SS and EI even when they extend beyond the 300-km range (Figs. 10d–f), because the radial continuity–azimuthal discontinuity check plays an important role in the spike filter.

Fig. 10.
Fig. 10.

(a),(d) Raw Z; (b),(e) ; and (c),(f) Z fields after dpQC from 0.5° tilt of (a)–(c) KJAX observations at 0843:14 UTC 23 Sep 2012 and (d)–(f) KICX observations at 0131:47 UTC 11 Mar 2013. The white circles indicate areas outside the max range of available . Without data, the dpQC was able to remove (d)–(f) the SS but not (a)–(c) the AP echoes.

Citation: Weather and Forecasting 29, 5; 10.1175/WAF-D-13-00072.1

4. Conclusions

A physically based P–NP echo classifier was developed using polarimetric radar variables and atmospheric environmental data. The algorithm applied a set of explicit meteorological principles that segregate P–NP echoes in the radar reflectivity field based on the correlation coefficient, 3D reflectivity, and temperature structure. The new algorithm was evaluated using 16 independent events and showed a high accuracy (HSS of 0.83) in segregating P–NP echoes. When compared with a more complex QC algorithm that uses all polarimetric variables and a neural network approach, the dpQC showed a similar HSS (0.83 vs 0.80), but with a significantly higher computational efficiency (3.15 vs 17.76s in CPU time usage and 83 vs 199 MB in RAM usage for processing a volume scan of data). Because of its computational efficiency and transparency of the algorithm, the dpQC is easy to implement and to maintain for large radar networks in a real-time environment.

Acknowledgments

Funding for this research was partially provided under the agreement between the National Oceanic and Atmospheric Administration (NOAA) and the Federal Aviation Administration's Aviation Weather Research Program, and partially provided by NOAA/Office of Oceanic and Atmospheric Research under NOAA-University of Oklahoma Cooperative Agreement NA11OAR4320072, U.S. Department of Commerce. The authors thank Dr. Alexander Ryzhkov and Dr. Yadong Wang for their comments. We would also like to express our appreciation to three anonymous reviewers for their helpful suggestions.

REFERENCES

  • Andrić, J., Kumjian M. R. , Zrnić D. S. , Straka J. M. , and Melnikov V. M. , 2013: Polarimetric signatures above the melting layer in winter storms: An observational and modeling study. J. Appl. Meteor. Climatol., 52, 682–700, doi:10.1175/JAMC-D-12-028.1.

    • Search Google Scholar
    • Export Citation

  • Bachmann, S. M., and Tracy M. D. , 2009: Data driven adaptive identification and suppression of ground clutter for weather radar. 25th Conf. on Int. Interactive Information and Processing Systems (IIPS) for Meteorology, Oceanography, and Hydrology, Phoenix, AZ, Amer. Meteor. Soc., 11B.3. [Available online at https://ams.confex.com/ams/89annual/techprogram/paper_146356.htm.]

  • Balakrishnan, N., and Zrnić D. S. , 1990: Use of polarization to characterize precipitation and discriminate large hail. J. Atmos. Sci., 47, 1525–1540, doi:10.1175/1520-0469(1990)047<1525:UOPTCP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Berkowitz, D. S., Schultz J. A. , Vasiloff S. , Elmore K. L. , Payne C. D. , and Boettcher J. B. , 2013: Status of dual pol QPE in the WSR-88D network. 27th Conf. on Hydrology, Austin, TX, Amer. Meteor. Soc., 2.2. [Available online at https://ams.confex.com/ams/93Annual/webprogram/Paper221525.html.]

  • Brandes, E. A., and Ikeda K. , 2004: Freezing-level estimation with polarimetric radar. J. Appl. Meteor. Climatol., 43, 1541–1553, doi:10.1175/JAM2155.1.

    • Search Google Scholar
    • Export Citation

  • Doviak, R. J., and Zrnić D. S. , 1993: Doppler Radar and Weather Observations. Academic Press, 562 pp.

  • Fabry, F., and Zawadzki I. , 1995: Long-term radar observations of the melting layer of precipitation and their interpretation. J. Atmos. Sci., 52, 838–851, doi:10.1175/1520-0469(1995)052<0838:LTROOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Giangrande, S. E., Krause J. M. , and Ryzhkov A. V. , 2008: Automatic designation of the melting layer with a polarimetric prototype of the WSR-88D radar. J. Appl. Meteor. Climatol., 47, 1354–1364, doi:10.1175/2007JAMC1634.1.

    • Search Google Scholar
    • Export Citation

  • Gourley, J. J., and Calvert C. M. , 2003: Automated detection of the bright band using WSR-88D data. Wea. Forecasting, 18, 585–599, doi:10.1175/1520-0434(2003)018<0585:ADOTBB>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Gourley, J. J., Tabary P. , and Parent-du-Chatelet J. , 2007: A fuzzy logic algorithm for the separation of precipitating from nonprecipitating echoes using polarimetric radar observations. J. Atmos. Oceanic Technol., 24, 1439–1451, doi:10.1175/JTECH2035.1.

    • Search Google Scholar
    • Export Citation

  • Gray, W., and Larsen H. , 2005: Radar rainfall estimation in the New Zealand context. Atmos. Sci. Lett., 6, 31–34, doi:10.1002/asl.87.

  • Grecu, M., and Krajewski W. F. , 2000: An efficient methodology for detection of anomalous propagation echoes in radar reflectivity data using neural networks. J. Atmos. Oceanic Technol., 17, 121–129, doi:10.1175/1520-0426(2000)017<0121:AEMFDO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Ivić, I. R., and Melnikov V. , 2013: Multi-lag hybrid correlation coefficient estimator. 36th Conf. on Radar Meteorology, Breckenridge, CO, Amer. Meteor. Soc., 248. [Available online at https://ams.confex.com/ams/36Radar/webprogram/Paper228447.html.]

  • Kumjian, M. R., 2013a: Principles and applications of dual-polarization weather radar. Part I: Description of the polarimetric radar variables. J. Oper. Meteor., 1, 226–242, doi:10.15191/nwajom.2013.0119.

    • Search Google Scholar
    • Export Citation

  • Kumjian, M. R., 2013b: Principles and applications of dual-polarization weather radar. Part II: Warm- and cold-season applications. J. Oper. Meteor., 1, 243–264, doi:10.15191/nwajom.2013.0120.

    • Search Google Scholar
    • Export Citation

  • Kumjian, M. R., 2013c: Principles and applications of dual-polarization weather radar. Part III: Artifacts. J. Oper. Meteor., 1, 265–274, doi:10.15191/nwajom.2013.0121.

    • Search Google Scholar
    • Export Citation

  • Lakshmanan, V., Fritz A. , Smith T. , Hondl K. , and Stumpf G. J. , 2007: An automated technique to quality control radar reflectivity data. J. Appl. Meteor. Climatol., 46, 288–305, doi:10.1175/JAM2460.1.

    • Search Google Scholar
    • Export Citation

  • Lakshmanan, V., Zhang J. , and Howard K. , 2010: A technique to censor biological echoes in weather radar images. J. Appl. Meteor. Climatol., 49, 453–462, doi:10.1175/2009JAMC2255.1.

    • Search Google Scholar
    • Export Citation

  • Lakshmanan, V., Christopher C. , Krause J. , and Tang L. , 2013: Quality control of weather radar data using polarimetric variables. J. Atmos. Oceanic Technol., 31, 1234–1249, doi:10.1175/JTECH-D-13-00073.1.

    • Search Google Scholar
    • Export Citation

  • Lim, S., Chandrasekar V. , and Bringi V. N. , 2005: Hydrometeor classification system using dual-polarization radar measurements: Model improvements and in situ verification. IEEE Trans. Geosci. Remote Sens., 43, 792–801, doi:10.1109/TGRS.2004.843077.

    • Search Google Scholar
    • Export Citation

  • Liu, H., and Chandrasekar V. , 2000: Classification of hydrometeors based on polarimetric radar measurements: Development of fuzzy logic and neuro-fuzzy systems, and in situ verification. J. Atmos. Oceanic Technol., 17, 140–164, doi:10.1175/1520-0426(2000)017<0140:COHBOP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Marzano, F. S., Scaranari D. , Montopoli M. , and Vulpiani G. , 2008: Supervised classification and estimation of hydrometeors from C-band dual-polarized radars: A Bayesian approach. IEEE Trans. Geosci. Remote Sens., 46, 85–98, doi:10.1109/TGRS.2007.906476.

    • Search Google Scholar
    • Export Citation

  • Melnikov, V. M., and Zrnić D. S. , 2007: Autocorrelation and cross-correlation estimators of polarimetric variables. J. Atmos. Oceanic Technol., 24, 1337–1350, doi:10.1175/JTECH2054.1.

    • Search Google Scholar
    • Export Citation

  • Park, H. S., Ryzhkov A. V. , Zrnić D. S. , and Kim K.-E. , 2009: The hydrometeor classification algorithm for the polarimetric WSR-88D: Description and application to an MCS. Wea. Forecasting, 24, 730–748, doi:10.1175/2008WAF2222205.1.

    • Search Google Scholar
    • Export Citation

  • Ryzhkov, A. V., 2007: The impact of beam broadening on the quality of radar polarimetric data. J. Atmos. Oceanic Technol., 24, 729–744, doi:10.1175/JTECH2003.1.

    • Search Google Scholar
    • Export Citation

  • Ryzhkov, A. V., Schuur T. J. , Burgess D. W. , Giangrande S. E. , and Zrnić D. S. , 2005: The Joint Polarization Experiment: Polarimetric rainfall measurements and hydrometeor classification. Bull. Amer. Meteor. Soc., 86, 809–824, doi:10.1175/BAMS-86-6-809.

    • Search Google Scholar
    • Export Citation

  • Ryzhkov, A. V., Kumjian M. R. , and Ganson S. M. , 2013a: Polarimetric radar characteristics of melting hail. Part I: Theoretical simulations using spectral microphysical modeling. J. Appl. Meteor. Climatol., 52, 2849–2870, doi:10.1175/JAMC-D-13-073.1.

    • Search Google Scholar
    • Export Citation

  • Ryzhkov, A. V., Kumjian M. R. , Ganson S. M. , and Zhang P. , 2013b: Polarimetric radar characteristics of melting hail. Part II: Practical implications. J. Appl. Meteor. Climatol., 52, 2871–2886, doi:10.1175/JAMC-D-13-074.1.

    • Search Google Scholar
    • Export Citation

  • Straka, J. M., and Zrnić D. S. , 1993: An algorithm to deduce hydrometeor types and contents from multiparameter radar data. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 513–516.

  • Straka, J. M., Zrnić D. S. , and Ryzhkov A. V. , 2000: Bulk hydrometeor classification and quantification using polarimetric radar data: Synthesis of relations. J. Appl. Meteor., 39, 1341–1372, doi:10.1175/1520-0450(2000)039<1341:BHCAQU>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Tabary, P., Le Henaff A. , Vulpiani G. , Parent-du-Chatelet J. , and Gourley J. J. , 2006: Melting layer characterization and identification with a C-band dual-polarization radar: A long-term analysis. Preprints, Fourth European Conf. on Radar in Meteorology and Hydrology, Barcelona, Spain, Servei Meteoròlogic de Catalunya, 1.8. [Available online at http://www.crahi.upc.edu/ERAD2006/proceedingsMask/00005.pdf.]

  • Tu, J. V., 1996: Advantages and disadvantages of using artificial neural networks versus logistic regression for predicting medical outcomes. J. Clin. Epidemiol., 49, 1225–1231, doi:10.1016/S0895-4356(96)00002-9.

    • Search Google Scholar
    • Export Citation

  • Vivekanandan, J., Zrnić D. S. , Ellis S. M. , Oye R. , Ryzhkov A. V. , and Straka J. M. , 1999: Cloud microphysics retrieval using S-band dual-polarization radar measurements. Bull. Amer. Meteor. Soc., 80, 381–388, doi:10.1175/1520-0477(1999)080<0381:CMRUSB>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Zhang, J., and Qi Y. , 2010: A real-time algorithm for the correction of brightband effects in radar-derived QPE. J. Hydrometeor., 11, 1157–1171, doi:10.1175/2010JHM1201.1.

    • Search Google Scholar
    • Export Citation

  • Zhang, J., Wang S. , and Clarke B. , 2004: WSR-88D reflectivity quality control using horizontal and vertical reflectivity structure. 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., 5.4. [Available online at https://ams.confex.com/ams/11aram22sls/techprogram/paper_81858.htm.]

  • Zhang, J., Langston C. , and Howard K. , 2008: Brightband identification based on vertical profiles of reflectivity from the WSR-88D. J. Atmos. Oceanic Technol., 25, 1859–1872, doi:10.1175/2008JTECHA1039.1.

    • Search Google Scholar
    • Export Citation

  • Zhang, J., and Coauthors, 2014: Initial operating capabilities of quantitative precipitation estimation in the multi-radar multi-sensor system. 28th Conf. on Hydrology, Atlanta, GA, Amer. Meteor. Soc., 5.3. [Available online at https://ams.confex.com/ams/94Annual/webprogram/Paper240487.html.]

  • Zrnić, D. S., and Ryzhkov A. V. , 1999: Polarimetry for weather surveillance radars. Bull. Amer. Meteor. Soc., 80, 389–406, doi:10.1175/1520-0477(1999)080<0389:PFWSR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Zrnić, D. S., Ryzhkov A. V. , Straka J. M. , Liu Y. , and Vivekanandan J. , 2001: Testing a procedure for the automatic classification of hydrometeor types. J. Atmos. Oceanic Technol., 18, 892–913, doi:10.1175/1520-0426(2001)018<0892:TAPFAC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
Save
Cover Weather and Forecasting
Weather and Forecasting

  • Andrić, J., Kumjian M. R. , Zrnić D. S. , Straka J. M. , and Melnikov V. M. , 2013: Polarimetric signatures above the melting layer in winter storms: An observational and modeling study. J. Appl. Meteor. Climatol., 52, 682–700, doi:10.1175/JAMC-D-12-028.1.

    • Search Google Scholar
    • Export Citation

  • Bachmann, S. M., and Tracy M. D. , 2009: Data driven adaptive identification and suppression of ground clutter for weather radar. 25th Conf. on Int. Interactive Information and Processing Systems (IIPS) for Meteorology, Oceanography, and Hydrology, Phoenix, AZ, Amer. Meteor. Soc., 11B.3. [Available online at https://ams.confex.com/ams/89annual/techprogram/paper_146356.htm.]

  • Balakrishnan, N., and Zrnić D. S. , 1990: Use of polarization to characterize precipitation and discriminate large hail. J. Atmos. Sci., 47, 1525–1540, doi:10.1175/1520-0469(1990)047<1525:UOPTCP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Berkowitz, D. S., Schultz J. A. , Vasiloff S. , Elmore K. L. , Payne C. D. , and Boettcher J. B. , 2013: Status of dual pol QPE in the WSR-88D network. 27th Conf. on Hydrology, Austin, TX, Amer. Meteor. Soc., 2.2. [Available online at https://ams.confex.com/ams/93Annual/webprogram/Paper221525.html.]

  • Brandes, E. A., and Ikeda K. , 2004: Freezing-level estimation with polarimetric radar. J. Appl. Meteor. Climatol., 43, 1541–1553, doi:10.1175/JAM2155.1.

    • Search Google Scholar
    • Export Citation

  • Doviak, R. J., and Zrnić D. S. , 1993: Doppler Radar and Weather Observations. Academic Press, 562 pp.

  • Fabry, F., and Zawadzki I. , 1995: Long-term radar observations of the melting layer of precipitation and their interpretation. J. Atmos. Sci., 52, 838–851, doi:10.1175/1520-0469(1995)052<0838:LTROOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Giangrande, S. E., Krause J. M. , and Ryzhkov A. V. , 2008: Automatic designation of the melting layer with a polarimetric prototype of the WSR-88D radar. J. Appl. Meteor. Climatol., 47, 1354–1364, doi:10.1175/2007JAMC1634.1.

    • Search Google Scholar
    • Export Citation

  • Gourley, J. J., and Calvert C. M. , 2003: Automated detection of the bright band using WSR-88D data. Wea. Forecasting, 18, 585–599, doi:10.1175/1520-0434(2003)018<0585:ADOTBB>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Gourley, J. J., Tabary P. , and Parent-du-Chatelet J. , 2007: A fuzzy logic algorithm for the separation of precipitating from nonprecipitating echoes using polarimetric radar observations. J. Atmos. Oceanic Technol., 24, 1439–1451, doi:10.1175/JTECH2035.1.

    • Search Google Scholar
    • Export Citation

  • Gray, W., and Larsen H. , 2005: Radar rainfall estimation in the New Zealand context. Atmos. Sci. Lett., 6, 31–34, doi:10.1002/asl.87.

  • Grecu, M., and Krajewski W. F. , 2000: An efficient methodology for detection of anomalous propagation echoes in radar reflectivity data using neural networks. J. Atmos. Oceanic Technol., 17, 121–129, doi:10.1175/1520-0426(2000)017<0121:AEMFDO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Ivić, I. R., and Melnikov V. , 2013: Multi-lag hybrid correlation coefficient estimator. 36th Conf. on Radar Meteorology, Breckenridge, CO, Amer. Meteor. Soc., 248. [Available online at https://ams.confex.com/ams/36Radar/webprogram/Paper228447.html.]

  • Kumjian, M. R., 2013a: Principles and applications of dual-polarization weather radar. Part I: Description of the polarimetric radar variables. J. Oper. Meteor., 1, 226–242, doi:10.15191/nwajom.2013.0119.

    • Search Google Scholar
    • Export Citation

  • Kumjian, M. R., 2013b: Principles and applications of dual-polarization weather radar. Part II: Warm- and cold-season applications. J. Oper. Meteor., 1, 243–264, doi:10.15191/nwajom.2013.0120.

    • Search Google Scholar
    • Export Citation

  • Kumjian, M. R., 2013c: Principles and applications of dual-polarization weather radar. Part III: Artifacts. J. Oper. Meteor., 1, 265–274, doi:10.15191/nwajom.2013.0121.

    • Search Google Scholar
    • Export Citation

  • Lakshmanan, V., Fritz A. , Smith T. , Hondl K. , and Stumpf G. J. , 2007: An automated technique to quality control radar reflectivity data. J. Appl. Meteor. Climatol., 46, 288–305, doi:10.1175/JAM2460.1.

    • Search Google Scholar
    • Export Citation

  • Lakshmanan, V., Zhang J. , and Howard K. , 2010: A technique to censor biological echoes in weather radar images. J. Appl. Meteor. Climatol., 49, 453–462, doi:10.1175/2009JAMC2255.1.

    • Search Google Scholar
    • Export Citation

  • Lakshmanan, V., Christopher C. , Krause J. , and Tang L. , 2013: Quality control of weather radar data using polarimetric variables. J. Atmos. Oceanic Technol., 31, 1234–1249, doi:10.1175/JTECH-D-13-00073.1.

    • Search Google Scholar
    • Export Citation

  • Lim, S., Chandrasekar V. , and Bringi V. N. , 2005: Hydrometeor classification system using dual-polarization radar measurements: Model improvements and in situ verification. IEEE Trans. Geosci. Remote Sens., 43, 792–801, doi:10.1109/TGRS.2004.843077.

    • Search Google Scholar
    • Export Citation

  • Liu, H., and Chandrasekar V. , 2000: Classification of hydrometeors based on polarimetric radar measurements: Development of fuzzy logic and neuro-fuzzy systems, and in situ verification. J. Atmos. Oceanic Technol., 17, 140–164, doi:10.1175/1520-0426(2000)017<0140:COHBOP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Marzano, F. S., Scaranari D. , Montopoli M. , and Vulpiani G. , 2008: Supervised classification and estimation of hydrometeors from C-band dual-polarized radars: A Bayesian approach. IEEE Trans. Geosci. Remote Sens., 46, 85–98, doi:10.1109/TGRS.2007.906476.

    • Search Google Scholar
    • Export Citation

  • Melnikov, V. M., and Zrnić D. S. , 2007: Autocorrelation and cross-correlation estimators of polarimetric variables. J. Atmos. Oceanic Technol., 24, 1337–1350, doi:10.1175/JTECH2054.1.

    • Search Google Scholar
    • Export Citation

  • Park, H. S., Ryzhkov A. V. , Zrnić D. S. , and Kim K.-E. , 2009: The hydrometeor classification algorithm for the polarimetric WSR-88D: Description and application to an MCS. Wea. Forecasting, 24, 730–748, doi:10.1175/2008WAF2222205.1.

    • Search Google Scholar
    • Export Citation

  • Ryzhkov, A. V., 2007: The impact of beam broadening on the quality of radar polarimetric data. J. Atmos. Oceanic Technol., 24, 729–744, doi:10.1175/JTECH2003.1.

    • Search Google Scholar
    • Export Citation

  • Ryzhkov, A. V., Schuur T. J. , Burgess D. W. , Giangrande S. E. , and Zrnić D. S. , 2005: The Joint Polarization Experiment: Polarimetric rainfall measurements and hydrometeor classification. Bull. Amer. Meteor. Soc., 86, 809–824, doi:10.1175/BAMS-86-6-809.

    • Search Google Scholar
    • Export Citation

  • Ryzhkov, A. V., Kumjian M. R. , and Ganson S. M. , 2013a: Polarimetric radar characteristics of melting hail. Part I: Theoretical simulations using spectral microphysical modeling. J. Appl. Meteor. Climatol., 52, 2849–2870, doi:10.1175/JAMC-D-13-073.1.

    • Search Google Scholar
    • Export Citation

  • Ryzhkov, A. V., Kumjian M. R. , Ganson S. M. , and Zhang P. , 2013b: Polarimetric radar characteristics of melting hail. Part II: Practical implications. J. Appl. Meteor. Climatol., 52, 2871–2886, doi:10.1175/JAMC-D-13-074.1.

    • Search Google Scholar
    • Export Citation

  • Straka, J. M., and Zrnić D. S. , 1993: An algorithm to deduce hydrometeor types and contents from multiparameter radar data. Preprints, 26th Int. Conf. on Radar Meteorology, Norman, OK, Amer. Meteor. Soc., 513–516.

  • Straka, J. M., Zrnić D. S. , and Ryzhkov A. V. , 2000: Bulk hydrometeor classification and quantification using polarimetric radar data: Synthesis of relations. J. Appl. Meteor., 39, 1341–1372, doi:10.1175/1520-0450(2000)039<1341:BHCAQU>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Tabary, P., Le Henaff A. , Vulpiani G. , Parent-du-Chatelet J. , and Gourley J. J. , 2006: Melting layer characterization and identification with a C-band dual-polarization radar: A long-term analysis. Preprints, Fourth European Conf. on Radar in Meteorology and Hydrology, Barcelona, Spain, Servei Meteoròlogic de Catalunya, 1.8. [Available online at http://www.crahi.upc.edu/ERAD2006/proceedingsMask/00005.pdf.]

  • Tu, J. V., 1996: Advantages and disadvantages of using artificial neural networks versus logistic regression for predicting medical outcomes. J. Clin. Epidemiol., 49, 1225–1231, doi:10.1016/S0895-4356(96)00002-9.

    • Search Google Scholar
    • Export Citation

  • Vivekanandan, J., Zrnić D. S. , Ellis S. M. , Oye R. , Ryzhkov A. V. , and Straka J. M. , 1999: Cloud microphysics retrieval using S-band dual-polarization radar measurements. Bull. Amer. Meteor. Soc., 80, 381–388, doi:10.1175/1520-0477(1999)080<0381:CMRUSB>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Zhang, J., and Qi Y. , 2010: A real-time algorithm for the correction of brightband effects in radar-derived QPE. J. Hydrometeor., 11, 1157–1171, doi:10.1175/2010JHM1201.1.

    • Search Google Scholar
    • Export Citation

  • Zhang, J., Wang S. , and Clarke B. , 2004: WSR-88D reflectivity quality control using horizontal and vertical reflectivity structure. 11th Conf. on Aviation, Range, and Aerospace Meteorology, Hyannis, MA, Amer. Meteor. Soc., 5.4. [Available online at https://ams.confex.com/ams/11aram22sls/techprogram/paper_81858.htm.]

  • Zhang, J., Langston C. , and Howard K. , 2008: Brightband identification based on vertical profiles of reflectivity from the WSR-88D. J. Atmos. Oceanic Technol., 25, 1859–1872, doi:10.1175/2008JTECHA1039.1.

    • Search Google Scholar
    • Export Citation

  • Zhang, J., and Coauthors, 2014: Initial operating capabilities of quantitative precipitation estimation in the multi-radar multi-sensor system. 28th Conf. on Hydrology, Atlanta, GA, Amer. Meteor. Soc., 5.3. [Available online at https://ams.confex.com/ams/94Annual/webprogram/Paper240487.html.]

  • Zrnić, D. S., and Ryzhkov A. V. , 1999: Polarimetry for weather surveillance radars. Bull. Amer. Meteor. Soc., 80, 389–406, doi:10.1175/1520-0477(1999)080<0389:PFWSR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation

  • Zrnić, D. S., Ryzhkov A. V. , Straka J. M. , Liu Y. , and Vivekanandan J. , 2001: Testing a procedure for the automatic classification of hydrometeor types. J. Atmos. Oceanic Technol., 18, 892–913, doi:10.1175/1520-0426(2001)018<0892:TAPFAC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • <sc>Fig</sc>. 1.
    View in gallery
    Fig. 1.

    (a) Base reflectivity and (b) corresponding hydrometeor classification fields on 0.5° tilt of KEWX observations valid at 0719:46 UTC 6 May 2012.

  • <sc>Fig</sc>. 2.
    View in gallery
    Fig. 2.

    Flowchart of the dpQC algorithm.

  • <sc>Fig</sc>. 3.
    View in gallery
    Fig. 3.

    (a) Raw Z, (b) , (c) Z fields after the QCNN process, and (d) Z field after the basic filter (threshold = 0.95) from the 0.5° tilt of KEWX observations valid at 0551:39 UTC 6 May 2012.

  • <sc>Fig</sc>. 4.
    View in gallery
    Fig. 4.

    Histogram plots of the values associated with NP (gray) and P echoes (black), observed by (a) KARX at 1933 UTC 26 Jun 2012 and (b) KVNX at 0736 UTC 20 May 2011.

  • <sc>Fig</sc>. 5.
    View in gallery
    Fig. 5.

    (a) Raw Z, (b) , and (c) fields from 0.5° tilt of the KVNX observations at 0355:54 UTC 1 May 2012. (d) The Z field after dpQC, where maroon shows low pixels from possible hail, NBF, and hydrometeors of mixed sizes/shapes.

  • <sc>Fig</sc>. 6.
    View in gallery
    Fig. 6.

    (a) Raw Z and (b) fields from 1.45° tilt of the KCLE observations at 1508:34 UTC 26 Jan 2012. The black circle indicates the areas of low values in the ML, which would be removed by the filter with (c) a 0.95 threshold but would not be by (d) an adjusted filter with a 0.70 threshold after the ML delineation.

  • <sc>Fig</sc>. 7.
    View in gallery
    Fig. 7.

    (a) Raw Z, (b) , and (c) Z fields after dpQC from 0.5° tilt of KOUN observations at 0208:28 UTC 11 May 2010. The red circles outline areas of severe AP.

  • <sc>Fig</sc>. 8.
    View in gallery
    Fig. 8.

    (a) Raw Z, (b) , and (c) Z fields after dpQC from 0.5° tilt of the KTLX observations at 1231:51 UTC 8 Mar 2013.

  • <sc>Fig</sc>. 9.
    View in gallery
    Fig. 9.

    Real-time CREF fields (a) before any QC, (b) after dpQCNN, and (c) after dpQC valid at 0830 UTC 10 Oct 2013. The white circles indicate an area of light precipitation and the red circles are the areas of blooms.

  • <sc>Fig</sc>. 10.
    View in gallery
    Fig. 10.

    (a),(d) Raw Z; (b),(e) ; and (c),(f) Z fields after dpQC from 0.5° tilt of (a)–(c) KJAX observations at 0843:14 UTC 23 Sep 2012 and (d)–(f) KICX observations at 0131:47 UTC 11 Mar 2013. The white circles indicate areas outside the max range of available . Without data, the dpQC was able to remove (d)–(f) the SS but not (a)–(c) the AP echoes.

Fig. 1.

(a) Base reflectivity and (b) corresponding hydrometeor classification fields on 0.5° tilt of KEWX observations valid at 0719:46 UTC 6 May 2012.
Fig. 2.

Flowchart of the dpQC algorithm.
Fig. 3.

(a) Raw Z, (b) , (c) Z fields after the QCNN process, and (d) Z field after the basic filter (threshold = 0.95) from the 0.5° tilt of KEWX observations valid at 0551:39 UTC 6 May 2012.
Fig. 4.

Histogram plots of the values associated with NP (gray) and P echoes (black), observed by (a) KARX at 1933 UTC 26 Jun 2012 and (b) KVNX at 0736 UTC 20 May 2011.
Fig. 5.

(a) Raw Z, (b) , and (c) fields from 0.5° tilt of the KVNX observations at 0355:54 UTC 1 May 2012. (d) The Z field after dpQC, where maroon shows low pixels from possible hail, NBF, and hydrometeors of mixed sizes/shapes.
Fig. 6.

(a) Raw Z and (b) fields from 1.45° tilt of the KCLE observations at 1508:34 UTC 26 Jan 2012. The black circle indicates the areas of low values in the ML, which would be removed by the filter with (c) a 0.95 threshold but would not be by (d) an adjusted filter with a 0.70 threshold after the ML delineation.
Fig. 7.

(a) Raw Z, (b) , and (c) Z fields after dpQC from 0.5° tilt of KOUN observations at 0208:28 UTC 11 May 2010. The red circles outline areas of severe AP.
Fig. 8.

(a) Raw Z, (b) , and (c) Z fields after dpQC from 0.5° tilt of the KTLX observations at 1231:51 UTC 8 Mar 2013.
Fig. 9.

Real-time CREF fields (a) before any QC, (b) after dpQCNN, and (c) after dpQC valid at 0830 UTC 10 Oct 2013. The white circles indicate an area of light precipitation and the red circles are the areas of blooms.
Fig. 10.

(a),(d) Raw Z; (b),(e) ; and (c),(f) Z fields after dpQC from 0.5° tilt of (a)–(c) KJAX observations at 0843:14 UTC 23 Sep 2012 and (d)–(f) KICX observations at 0131:47 UTC 11 Mar 2013. The white circles indicate areas outside the max range of available . Without data, the dpQC was able to remove (d)–(f) the SS but not (a)–(c) the AP echoes.
All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 1269 388 65
PDF Downloads 903 216 20