An Automated Radar Technique for the Identification of Tropical Precipitation

Xiaoyong Xu Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Norman, Oklahoma

Search for other papers by Xiaoyong Xu 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
Jian Zhang Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Norman, Oklahoma

Search for other papers by Jian Zhang in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

A radar-based automated technique for the identification of tropical precipitation was developed to improve quantitative precipitation estimation during extreme rainfall events. The technique uses vertical profiles of reflectivity to identify the potential presence of warm rain (i.e., tropical rainfall) microphysics and delineates the tropical rainfall region to which the tropical ZR relationship is applied. The performance of the algorithm is examined based on case studies of five storms that produced extreme precipitation in the United States. Results demonstrate relative improvements in radar-based quantitative precipitation estimation through the automated identification of tropical rainfall and the subsequent adaptation of the tropical ZR relation to account for the potential warm rain processes.

Corresponding author address: Xiaoyong Xu, NSSL/WRDD, 120 David L. Boren Blvd., Norman, OK 73072. Email: E-mail xiaoyong.xu@noaa.gov

Abstract

A radar-based automated technique for the identification of tropical precipitation was developed to improve quantitative precipitation estimation during extreme rainfall events. The technique uses vertical profiles of reflectivity to identify the potential presence of warm rain (i.e., tropical rainfall) microphysics and delineates the tropical rainfall region to which the tropical ZR relationship is applied. The performance of the algorithm is examined based on case studies of five storms that produced extreme precipitation in the United States. Results demonstrate relative improvements in radar-based quantitative precipitation estimation through the automated identification of tropical rainfall and the subsequent adaptation of the tropical ZR relation to account for the potential warm rain processes.

Corresponding author address: Xiaoyong Xu, NSSL/WRDD, 120 David L. Boren Blvd., Norman, OK 73072. Email: E-mail xiaoyong.xu@noaa.gov

1. Introduction

Radar-based quantitative precipitation estimation (QPE) is crucial in timely and accurate flood and flash flood identification and warnings. Unfortunately, radar rainfall estimates are affected by source errors such as reflectivity calibration differences, inappropriate radar reflectivity–rainfall rate (ZR) relationships, range degradation, and hail and brightband contaminations (see, e.g., Wilson and Brandes 1979; Zawadzki 1984; Joss and Waldvogel 1990; Hunter 1996; Baeck and Smith 1998). Baeck and Smith (1998) suggested that there could be several contributors, one of which is inappropriate ZR parameters, to radar rainfall underestimation based on reflectivity observations. A number of efforts have been focused on correcting precipitation underestimates by radar (e.g., Wood 1997; Anagnostou and Krajewski 1998; Vieux and Bedient 1998; Vignal et al. 2000; Germann and Joss 2002). In this paper, attention is paid to the underestimation by radar due to using an inappropriate ZR relationship with tropical rainfall. The standard Weather Surveillance Radar-1988 Doppler (WSR-88D) ZR parameters (Z = 300R1.4) are on occasion inappropriate for heavy rain, thus leading to the occurrence of significant underestimates of rainfall (e.g., Baeck and Smith 1998; Vieux and Bedient 1998). These conditions are most prevalent in tropical rainfall regimes where a deep warm cloud layer exists and warm rain processes prevail. To compensate for this, a tropical ZR relationship can and should be utilized to enhance the warm cloud rainfall estimation in place of the standard ZR. Failure to switch to a tropical ZR can lead to a failure to detect a situation leading to flash flood. Wood (1997) used the tropical ZR relationship to improve precipitation estimates during a heavy rain event in southeast Texas. In a flash flood study, Vieux and Bedient (1998) reduced the underestimates significantly with the tropical ZR relationship. Aiken (2000) discussed the performance of the Raleigh, North Carolina, WSR-88D during Hurricanes Fran and Floyd, and suggested that the precipitation algorithms where the tropical ZR was used performed exceptionally well during these hurricanes. These studies applied a uniform tropical ZR to totally replace the uniform standard Z–R. However, there is usually a coexistence of different precipitation types within a rainfall event (e.g., Houze 1989; Tokay and Short 1996; Uijlenhoet et al. 2003). The uniform tropical ZR probably results in errors in the radar rainfall estimate due to its inappropriateness to other precipitation classes (e.g., convective or stratiform). A potentially more accurate and physically correct approach would be to apply different ZR relationships based on real-time precipitation echo classification.

Many studies in the past have shown significant variability in the relationship between radar reflectivity and surface rainfall rate (e.g., Zawadzki 1984; Austin 1987; Uijlenhoet et al. 2003). The variability originates from differences in the microphysical properties and vertical structure of raining clouds associated with different precipitation systems. Houze (1993), using cloud models and radar observations, presented significant differences in the microphysical properties and vertical structure of raining clouds associated with convective and stratiform regimes. Tokay and Short (1996) showed significant differences in the multiplier coefficient of ZR relationships associated with convective and stratiform rain types. This follows that the uncertainty in the ZR conversion can potentially be reduced by applying different ZR relationships for different precipitation systems. Studies (e.g., Steiner et al. 1995; Anagnostou and Krajewski 1999) have shown improvement in radar rainfall accuracy by applying a rainfall convective–stratiform classification.

The authors have examined numerous vertical profiles of reflectivity (VPRs) characteristics and found that the underestimations by the convective or stratiform ZR were usually associated with the VPRs that have reflectivity values increasing monotonically with decreasing height at the lower levels (tropical VPRs). On the other hand, for VPRs that have maximum reflectivity in their midlevels (continental), the underestimate is not significant. The findings indicated that the VPR structure, especially at the lower levels, could provide some guidance for appropriate ZR relationships for estimating surface precipitation from radar data. The following study builds upon the convective–stratiform precipitation typing technique in Zhang et al. (2008, hereafter Z08) with the addition of a new tropical precipitation delineation scheme. Subsequently, three different ZR relationships are applied based on the classifications of the precipitation processes. The objective of this paper is to show improved accuracy in radar-based QPE through the introduction of the new tropical precipitation delineation scheme.

2. Algorithm description

a. Identification of tropical VPRs

The VPRs can provide important information for radar QPE applications such as the delineation of the brightband layer (e.g., Z08), the identification of potential warm rain processes, and the adjustment for the precipitation efficiency below cloud base. The shapes of VPRs are a potential indicator of the precipitation microphysical process (Houze 1993). Examinations of the VPR characteristics where heavy rainfall was being significantly underestimated when using convective ZR showed that the reflectivity increased monotonically below 0°C height with a maximum reflectivity at the lowest level. This concurs with the results of Zipser and Lutz (1994), who documented that the tropical oceanic profile has a maximum reflectivity at the lowest level and a very rapid decrease in reflectivity with height beginning just above the freezing level. Further, Kucera et al. (1996) found that the volume average reflectivity profiles during the Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE) increased slightly from the 2-km altitude to the surface.

Based on the examination of the VPR characteristics where heavy rainfall was significantly underestimated by the convective or stratiform ZR, two classes of VPR structures are considered to be tropical VPRs where the tropical ZR would likely perform better than the convective Z–R: 1) The reflectivity monotonically increases as the height decreases with a maximum at the lowest level (e.g., Fig. 1a), and 2) there exists a bright band and reflectivity increases (Fig. 1b) or remains constant (Fig. 1c) with the decrease in height below the bottom of the bright band. In this research, the identification (or diagnosis) of tropical VPRs was based on the hourly averaged stratiform VPRs and the 0°C height derived from a background temperature profile. The temperature data are obtained from hourly analyses of the Rapid Update Cycle (RUC; Benjamin et al. (2004) model. For each radar volume scan, two mean VPRs are computed from base-level reflectivity data near the radar one for convective and the other for stratiform precipitation (see Z08). A running hourly average is applied to the volume scan mean VPRs to obtain hourly averaged VPRs. For warm rain process dominated storms, the reflectivity typically increases as height decreases or has a lower-level echo centroid since drops grow fastest as they start falling and colliding at the bottom of the cloud with the largest drops in the lower levels. Hence, the vertical structure of warm rain process dominated storm echoes is very similar to the characteristics of the tropical VPRs at the lower levels (i.e., reflectivity increases monotonically as height decreases). An examination of the environmental characteristics associated with tropical VPRs revealed a deep layer of moisture (usually surface to 500 hPa) in addition to a pronounced absence of cloud to ground (CG) lightning strikes. Therefore, the identified tropical VPRs likely represent the presence of potential warm rain microphysics. Further, a VPR is a moving average in time and is constrained at a set distance from the radar (Z08). Subsequently, a tropical VPR may contain the presence of a bright band and extend significantly vertically, reflecting the coexistence of stratiform and ice-phase processes along with the warm rain process. Several studies (Houze 1989; Tokay and Short 1996) have strongly suggested the coexistence of different microphysical processes within tropical rainfall. The warm rain processes of nucleation, condensation, and coalescence provide the major source of water for raindrops via the continuation of the coalescence process. Microphysical processes such as freezing and riming might occur when updrafts carry liquid hydrometeors well above the freezing level where ice nucleation processes, wind shear, detrainment, and gravitational settling create the stratiform environment, as characterized by ice crystal growth, and aggregation above the 0°C isotherm and melting and evaporation below. Those hydrometeors that return to the surface as precipitation are almost always in the liquid phase, especially over the tropical oceans (Tokay and Short 1996).

The automated identification scheme for tropical VPRs can be summarized as a sequence of the following processing steps.

  1. Look for the local maximum reflectivity Zlmax near the freezing level in the VPR. The search starts from 500 m above the 0°C height at the radar site and continues downward. A 500-m cushion is used to account for uncertainties in the model 0°C height due to infrequent and sparse upper-air sounding observations (e.g., Z08). The VPR is identified as the “tropical VPR” if the Zlmax is located at the bottom of the VPR.

  2. If the Zlmax is not located at the bottom of the VPR, determine if there exists a bright band in the VPR using the method developed by Z08.

  3. If there is not a bright band, search for the global minimum reflectivity, Zgmin, below the level where the Zlmax is located. Once the Zgmin is found, the following criterion is checked ZlmaxZgmin < α. Here, the parameter α is an adaptable threshold (default α = 0.8 dB); α can be relaxed to a larger value (e.g., 1.6 dB) when Zgmin is not lower than the reflectivity at the starting height in step 1. If the aforementioned criterion is not satisfied, then VPR is labeled as the “nontropical VPR.” If the criterion is met, search for the new Zlmax below the original Zlmax and for the new Zgmin below the new Zlmax. This process is repeated until the bottom of the VPR is reached. If the bottom of the VPR can be reached, the VPR is identified as the “tropical VPR.” A sensitivity study of the VPR’s shape was completed at an interval of 0.1 dB. The optimal value for separating the tropical VPRs from the nontropical VPRs was determined to be 0.8 dB (or 1.6 dB). Further refinement of this threshold may be desirable pending the real-time performance of the identification across different geographical regimes.

  4. If there is a bright band, search for the new Zlmax below the bottom of the bright band. The rest of the process is the same as step 3.

Several small adjustments are made in determining the brightband bottom in order to account for the VPR variability. A flow chart of the tropical VPR identification scheme is given in Fig. 2.

b. Delineation of a tropical rainfall region

Ideally, the ZR relation is chosen based on drop size distribution (DSD) observations to minimize any inappropriate ZR interference. That, however, is difficult to accomplish in real-time operational applications due to the shortage of DSD data. As alternatives, various techniques were developed to classify precipitation types based on three-dimensional radar reflectivity observations (Steiner et al. 1995; Z08) and atmospheric environmental data (Z08). Different ZR relationships were applied based on precipitation types and more accurate QPE was obtained (Steiner et al. 1995). In the current study, the Z08 precipitation-typing technique was used to delineate convective and stratiform precipitation and a new tropical precipitation delineation scheme was added. The combined technique produces a two-dimensional field in which each grid cell has a precipitation-type flag.

The new tropical precipitation delineation is based on tropical VPRs and the two-dimensional radar reflectivity grid that will be used for rainfall estimation. The 2D reflectivity field is basically the unblocked radar reflectivity observations closest to the ground [namely, “hybrid scan reflectivity”; see Fulton et al. (1998)]. A grid cell is assigned a precipitation flag marked as tropical precipitation if the following three criteria are all satisfied: 1) the grid is covered by the tropical VPR influencing area whose center is positioned at a radar site with a certain radius (default = 250 km), 2) the hybrid scan reflectivity is greater than a threshold (default = 30 dBZ), and 3) the surface wet-bulb temperature is higher than 2°C to assure that the surface precipitation is not snow. Here, the radius is an adaptable parameter and is generally set to a value larger than 80 km, the maximum range used to calculate the VPRs (see Z08). The delineation scheme is not sensitive to the radius since those grids beyond the radius can be processed later by the outward extension (see below). The 250-km range is used in this research since it is a mean effective radius for rainfall measurement by the operational radars. The 30-dBZ threshold is chosen based on the following consideration the major difference between a convective ZR curve and a tropical ZR curve occurs for reflectivities larger than 30 dBZ. The two curves show very small discrepancies for reflectivities lower than 30 dBZ. Grid cells outside of the areas of influence of the tropical rainfall VPRs are also identified as tropical precipitation if they adjoin those that are already labeled as tropical precipitation and meet criteria 2 and 3. The search starts from the grid cells on the boundary of the tropical VPR influences regions and proceeds outward until the boundaries of the analysis domain are reached. The grid cells labeled as tropical precipitation compose the tropical rainfall region. A flow chart of the delineation of the tropical precipitation region is given in Fig. 3.

c. Rainfall classification-based Z–R transformation

Based on the precipitation-type classification, different ZR relations are used to compute the rain rate from different microphysical processes. In the new algorithm developed in this study, the formula Z = 230R1.25 is utilized for the tropical rainfall region. This ZR relation was derived for oceanic tropical precipitation observed during the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE; Austin et al. 1976; Hudlow 1979). For convective echo, the relationship used is Z = 3001.4 (Hunter 1996; Fulton et al. 1998; Ulbrich and Lee 1999), and for stratiform echo, Z = 200R1.6 [the Marshall–Palmer (1948) relation] is applied. The current study does not consider other ZR relations such as snow.

3. Case studies

Five cases (see Table 1) were examined to assess the improvement of the new tropical rain identification scheme in comparison with analyses using just convective and stratiform. Two sets of QPEs were generated for each case; one uses two precipitation types of the Z08 classification scheme (convective and stratiform, or “CS” hereafter) and the associated ZR relationships (see section 2c), while the other uses convective, stratiform, and the new tropical precipitation types (“CST” hereafter) and the associated ZR relationships (see section 2c).

The performances of the two configurations are assessed by comparing their gridded rainfall accumulations with corresponding gauge rainfall reports. The gauge rainfall reports come from the Hydrometeorological Automated Data System (HADS; information online at http//www.nws.noaa.gov/oh/hads/). The performance of each configuration in the rainfall estimation is quantified based on the following statistics 1) the mean bias, bias = 〈AR〉/〈AG〉; 2) the radar–gauge root-mean-square error, RMSE = 〈|ARAG|21/2; and 3) the correlation coefficient, CC = (〈ARAG〉 − 〈AR〉〈AG〉)/[(〈A2R〉 − 〈AR2)(〈A2G〉 − 〈AG2)]0.5, where AR and AG represent radar and gauge accumulations for any given radar–gauge (R–G) pair, and the angle brackets indicate averaging over all such pairs. A bias of greater (less) than 1.0 indicates that the radar has overestimated (underestimated) the rainfall assuming the gauge report is accurate. In this work, a valid R–G pair is defined as a pair in which the gauge accumulation value is 0.8 mm or more. Smaller accumulations are omitted in order to reduce errors caused by the minimum precipitation threshold of the rain gauge (e.g., Klazura et al. 1999; Anagnostou and Krajewski 1999). The performance of the CST is compared against that of the CS. These comparisons are used to demonstrate improvements when using the CST configuration over using the CS configuration.

a. Hurricane Isabel on 18–19 September 2003

Hurricane Isabel was considered one of the most significant tropical cyclones to affect North Carolina and Virginia. Isabel made landfall in eastern North Carolina around 0100 UTC 18 September 2003 and produced heavy rains over large portions of eastern North Carolina and Virginia between 0100 UTC 18 September and 0500 UTC 19 September 2003. The maximum rainfall amount in a continuous 24-h period reached as high as 160 mm. Figure 4 presents example VPR plots for this case. All VPRs presented in Fig. 4 showed a monotonic increase in reflectivity with the decrease of height at the lower level and were subsequently identified as being the tropical VPRs. The warm rain microphysical process could be an important contributor to the increase in reflectivity at the lower level. Figure 5 depicts the radar precipitation-type classification results using the CST at the time corresponding to Fig. 4. The tropical rainfall region covers the most area of the whole domain of interest. Figure 6 shows the biases (= AR/AG) for R–G pairs. Underestimates of rainfall (bias <1) by radar take place over almost the whole domain of interest when using the CS (Figs. 6a and 6c). After applying the CST, the underestimation area is reduced with more consistency (i.e., bias is close to 1) between radar estimations and gauges (Figs. 6b and 6c). The improvements coming from the CST with respect to the CS are also illustrated by the changes in the radar–rain gauge difference statistics (Fig. 7). From the top row in Fig. 7, it can be seen that the biases from the CS and the CST are 0.75 and 0.99, respectively, in the 24-h rain accumulation estimates for 0100 UTC 18 September and 0100 UTC 19 September 2003. The corresponding RMSEs are 21.7 and 13.5 mm, respectively. For the accumulation from 0500 UTC 18 September to 0500UTC 19 September 2003 (Figs. 7c and 7d), the biases produced by the two schemes are 0.72 and 0.95, respectively. The corresponding RMSEs are 22.9 and 13.5 mm, respectively. For the above two periods, the correlation coefficients (CCs) of the CST are slightly higher than those of the CS. The CST performs better than does the CS for this event, indicating that the tropical rainfall identification component in the CST would improve rainfall estimates for hurricanes and other similar precipitations in which the efficient warm rain processes dominate.

b. Texas rainfall event during 26–27 March 2007

The rainfall rates (∼10 mm h−1) in central Texas on 26–27 March were extraordinary, and resulted in widespread flooding with 80 mm of the rainfall total coming during a 24-h period. Figure 8 gives some VPRs for this case at 2300 UTC on 26 March 2007. The VPRs from the Austin, Texas (KEWX), and Fort Hood, Texas (KGRK), radars showed the tropical structures, but the VPRs from the Midland, Texas (KMAF), and Laughlin Air Force Base (AFB), Texas (KDFX), radars did not. Based on the different structures of the VPRs, the tropical rainfall region is delineated around the KEWX and KGRK radars (Fig. 9). The use of CS alone resulted in a significant underestimation in the observed rainfall by radar (Figs. 10a and 10c). The use of the CST leads to a bias of closer to 1 (1.01) and a reduced RMSE (10.8 mm) compared to the bias (0.70) and RMSE (14.2 mm) of the CS. Though the gauge has a slightly lower correlation to the CST estimation (0.81) than to the CS estimation (0.83), both of the correlations are relatively high. Overall, improved consistence between the radar estimation and the gauge was introduced after applying the tropical rainfall identification module. However, considerable underestimation remains with this case even with the tropical identification. The significant underestimations in the central Texas event are potentially a result of reflectivity data dropouts from Dyess AFB, Texas (KDYX), during the period 1200–1800 UTC 26 March 2007.

c. Tropical Storm Barry during 1–3 June 2007

Tropical Storm Barry resulted in copious amounts of rainfall across the southeastern United States on 1–3 June 2007, in which approximately 130 mm of rain fell during a 24-h time period. The VPRs associated with Tropical Storm Barry depict a classic transition from convective to tropical precipitation processes (Fig. 11). Figure 12 shows a scatterplot of radar versus gauge results for this case. The gauge showed a high correlation (∼0.92) to either the CS or CST results. For the radar 24-h accumulation ending at 1200 UTC of 3 June 2007 by the CS, the average bias and RMSE are 0.62 and 9.2 mm, respectively. While with CST there remained significant underestimation in some areas, overall improved radar–gauge consistence was realized. The bias and RMSE are improved to 0.79 and 13.0 mm, respectively. The possible main reason for the remaining underestimation is that the current tropical ZR relationship is not fully representative of the actual rainfall process, similar to the central Texas event.

d. Heavy rainfall in NE Oklahoma on 11 June 2007

The overnight rainfall event in northeastern Oklahoma and southeastern Kansas on 11 June 2007 was very impressive with rain rates approaching 25 mm h−1. The VPRs at the Tulsa, Oklahoma (KINX), and Oklahoma City, Oklahoma (KTLX), radars showed typical tropical structures (fFigures not shown). Figures 13 and 14 show the improvement of the consistency of the radar-estimated rainfall and the rain gauge accumulations when the CST is used. Using the CST removed biases lower than 1 in most of the radar–gauge pairs, and for all of the pairs with higher amounts. After using the CST scheme, biases near 1.0 dominate in the region of interest (Figs. 13b, 13d and 13f). For the 1-h accumulation between 0900 and 1000 UTC, the bias is improved to a high degree from 0.57 by the CS to 1.02 by the CST. The corresponding RMSE is lowered from 8.4 to 3.9 mm. A very substantial improvement in terms of bias and RMSE is also indicated for the 1-h accumulation for 1200–1300 UTC and the 6-h amount from 0900 to 1500 UTC. A very high correlation (>0.9) is shown between the gauge and the radar estimates of either the CS or the CST.

e. North Texas flooding on 18 June 2007

During the north Texas flooding event of 18 June 2007, 200 mm of rain was reported at Gainesville, Texas, in a 12-h period. The high rainfall rate could have been caused by warm rain processes. The VPRs from the Fort Worth, Texas (KFWS), radar indicated that the storm had a lower-level echo centroid structure during this rainfall event (Fig. 15). This low-echo centroid structure was possibly related to the warm rain microphysical process. The CS grossly underestimates the precipitation (Fig. 16a). The bias of the CS in the 12-h accumulation for 0100–1300 UTC is as low as 0.65. The corresponding RMSE reaches 21.2 mm. With the CST, the bias is improved to 1.05 and the RMSE is reduced to 15.1 mm. Figure 16b also shows the correlation coefficient increase by the CST method with respect to CS (Fig. 16a). For the extreme accumulation (point A in Fig. 16), the radar estimate agreed well with the gauge when the CST configuration was employed. However, a significant increase in the rainfall amounts and better radar–gauge correspondence was not obtained in other locations near the KFWS radar (points B and C in Fig. 16). As with previous cases, missing data likely contributed to the poor correspondence. The impact of these missing data on the CST performance is being further investigated.

4. Summary

The use of a convective or stratiform ZR relationship may underestimate heavy precipitation in warm rain process–dominated storms. A radar-based fully automated technique for the identification of tropical rainfall was developed to improve quantitative precipitation estimation during extreme rainfall events. The technique used vertical profiles of reflectivity to identify the potential presence of warm rain (e.g., tropical rainfall) microphysics and delineates the tropical rainfall region to which the tropical ZR relationship is applied. The technique comprises three modules 1) the identification of tropical VPRs, 2) the delineation of a tropical rainfall region, and 3) ZR transformation based on precipitation typing. A new convective–stratiform–tropical (CST) classification algorithm is configured by adding the new tropical precipitation delineation scheme to the Z08 convective–stratiform (CS) typing technique.

The performance of the new classification algorithm was examined based on case studies of five storms that produced extreme precipitation. Table 2 summarizes the performances of radar rainfall estimates before (CS configuration) and after (CST configuration) the new tropical rainfall identification scheme is applied. The results demonstrate substantial improvements in radar-based quantitative precipitation estimation through the automated adaptation of the tropical ZR relationships. The improvement is operationally significant given that extreme rainfall events often associated with flash floods and hurricane landfalls are associated with highly efficient warm rain processes.

Through the case studies, the CST demonstrated its suitability to improving the radar-based rainfall accumulation estimation across different time periods (e.g., 1, 6, 12, and 24 h). We will further refine the CST scheme and perform extensive database studies. The refinement of the tropical rainfall identification scheme will include the introduction of the sounding data and the lightning data along with model analysis fields. Further sensitive studies will be conducted to refine the adjustment of some parameters (e.g., the α, the influencing radius, and the 30-dBZ threshold discussed in section 2).

In the past, the National Severe Storms Laboratory (NSSL) has conducted an operational demonstration of the polarimetric utility of the Norman, Oklahoma (KOUN), radar. The polarimetric variables could give insight into the microphysical properties of rainfall and the type of DSDs (e.g., Ryzhkov and Zrnić 1996; Ryzhkov et al. 2005). The polarimetric measurements revealed that the tropical rainfall, identified by the algorithm developed in this work, was often characterized by the dominance of small drops in the raindrop spectrum. The corresponding Z–ZDR scattergrams show very “flat,” that is, very small differential reflectivity (ZDR), values for a given reflectivity (Z) value. For a given reflectivity Z, the DSD with the dominance of small drops generally means a higher rain-rate R. Hence, the conventional ZR relation generally underestimates rainfall with the dominance of small drops in the DSD, indicating that the automated identification of tropical rainfall is essentially reasonable in this work. The further analysis on the DSDs of tropical rainfall using the polarimetric radar will be performed in future studies.

Acknowledgments

Major funding for this research was provided through collaboration with the Central Weather Bureau of Taiwan, Republic of China, and partial funding was provided under NOAA–University of Oklahoma Cooperative Agreement NA17RJ1227, U.S. Department of Commerce. The authors also thank Wenwu Xia and Carrie Langston for assembling and processing much of the data. Sincere thanks to the three anonymous reviewers for their helpful comments on the original manuscript.

REFERENCES

  • Aiken, R., 2000: Performance of WSR-88D during Hurricanes Fran and Floyd rainfall measurements Presentation on 2000 TAC Report. NOAA/National Weather Service/Radar Operations Center, Wilmington, NC. [Available online at http://www.roc.noaa.gov/app/tac/past_mtgs/May_2000_Report.asp.].

  • Anagnostou, E. N., and Krajewski W. F. , 1998: Calibration of the WSR-88D precipitation processing subsystem. Wea. Forecasting, 13 , 396406.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Anagnostou, E. N., and Krajewski W. F. , 1999: Real-time radar rainfall estimation. Part II: Case study. J. Atmos. Oceanic Technol., 16 , 198205.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Austin, P., 1987: Relation between measured radar reflectivity and surface rainfall. Mon. Wea. Rev., 115 , 10531070.

  • Austin, P. M., Geotis S. G. , Cunning J. B. , Thomas J. L. , Sax R. I. , and Gillespie J. R. , 1976: Raindrop size distributions and Z-R relationsip for GATE. Preprints, 10th Conf. on Hurricanes and Tropical Meteorology, Charlottesville, VA, Amer. Meteor. Soc., 165–166.

  • Baeck, M., and Smith J. , 1998: Rainfall estimation by the WSR-88D for heavy rainfall events. Wea. Forecasting, 13 , 416436.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and Coauthors, 2004: An hourly assimilation–forecast cycle: The RUC. Mon. Wea. Rev., 132 , 495518.

  • Fulton, R., Breidenbach J. , Seo D-J. , Miller D. , and O’Bannon T. , 1998: The WSR-88D rainfall algorithm. Wea. Forecasting, 13 , 377395.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Germann, U., and Joss J. , 2002: Mesobeta profiles to extrapolate radar precipitation measurements above the Alps to the ground level. J. Appl. Meteor., 41 , 542557.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Houze R. A. Jr., , 1989: Observed structure of mesoscale convective systems and implications for large-scale heating. Quart. J. Roy. Meteor. Soc., 115 , 425461.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Houze R. A. Jr., , 1993: Cloud Dynamics. Academic Press, 573 pp.

  • Hudlow, M., 1979: Mean rainfall patterns for the three phases of GATE. J. Appl. Meteor., 18 , 16561669.

  • Hunter, S. M., 1996: WSR-88D radar rainfall estimation Capabilities, limitations and potential improvements. Natl. Wea. Dig., 20 , 4. 2638.

    • Search Google Scholar
    • Export Citation
  • Joss, J., and Waldvogel A. , 1990: Precipitation measurement and hydrology. Radar in Meteorology Battan Memorial and 40th Anniversary Radar Meteorology Conference, D. Atlas, Ed., Amer. Meteor. Soc., 577–606.

    • Crossref
    • Export Citation
  • Klazura, G. E., Thomale J. M. , Kelly D. S. , and Jendrowski P. , 1999: A comparison of NEXRAD WSR-88D radar estimates of rain accumulation with gauge measurements for high- and low-reflectivity horizontal gradient precipitation events. J. Atmos. Oceanic Technol., 16 , 18421850.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kucera, P. A., Short D. A. , and Thiele O. W. , 1996: An analysis of rainfall intensity and vertical structure from shipborne radars TOGA COARE. Preprints, Seventh Conf. on Mesoscale Processes, Reading, United Kingdom, Amer. Meteor. Soc., 131–134.

  • Marshall, J. S., and Palmer W. Mc K. , 1948: The distribution of raindrops with size. J. Meteor., 5 , 165166.

  • Ryzhkov, A. V., and Zrnić D. S. , 1996: Rain in shallow and deep convection measured with a polarimetric radar. J. Atmos. Sci., 53 , 29892995.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryzhkov, A. V., Schuur T. J. , Burgess D. W. , Heinselman P. L. , Giangrande S. E. , and Zrnić D. S. , 2005: The Joint Polarization Experiment. Bull. Amer. Meteor. Soc., 86 , 809824.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Steiner, M., Houze R. A. Jr., and Yuter S. E. , 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34 , 19782007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tokay, A., and Short D. A. , 1996: Evidence from tropical raindrop spectra of the origin of rain from stratiform versus convective clouds. J. Appl. Meteor., 35 , 355371.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Uijlenhoet, R., Steiner M. , and Smith J. A. , 2003: Variability of raindrop size distribution in a squall line and implications for radar rainfall estimation. J. Hydormeteor., 44 , 4361.

    • Search Google Scholar
    • Export Citation
  • Ulbrich, C. W., and Lee L. G. , 1999: Rainfall measurement error by WSR-88D radars due to variation in ZR law parameters and the radar constant. J. Atmos. Oceanic Technol., 16 , 10171024.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vieux, B., and Bedient P. , 1998: Estimation of rainfall for flood prediction from WSR-88D reflectivity: A case study, 17–18 October 1994. Wea. Forecasting, 13 , 407415.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vignal, B., Galli G. , Joss J. , and Germann U. , 2000: Three methods to determine profiles of reflectivity from volumetric radar data to correct precipitation estimates. J. Appl. Meteor., 39 , 17151726.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wilson, J. W., and Brandes E. A. , 1979: Radar measurement of rainfall: A summary. Bull. Amer. Meteor. Soc., 60 , 10481058.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wood, L. T., 1997: Using the tropical Z-R relationship to improve precipitation estimates during a heavy rain event in southeast Texas. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 208–209.

  • Zawadzki, I., 1984: Factors affecting the precision of radar measurements of rain. Preprints, 22nd Int. Conf. on Radar Meteorology, Zurich, Switzerland, Amer. Meteor. Soc., 251–256.

  • Zhang, J., Langston C. , and Howard K. , 2008: Brightband identification based on vertical profiles of reflectivity from WSR-88D. J. Atmos. Oceanic Technol., 25 , 18591872.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zipser, E. J., and Lutz K. R. , 1994: The vertical profile of radar reflectivity of convective cells: A strong indicator of storm intensity and lightning probability? Mon. Wea. Rev., 122 , 17511759.

    • Crossref
    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Conceptual models of tropical VPRs: (a) reflectivity monotonically increases as the height decreases with a maximum at the lowest level and there exists a bright band where (b) reflectivity increases or (c) remains constant with the decrease of height below the bottom of the bright band. The solid line is the VPR. The dashed line is the 0°C isotherm.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 2.
Fig. 2.

A flow chart of the automated scheme used to identify tropical VPRs.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 3.
Fig. 3.

A flow chart of the delineation of tropical rainfall regions based on tropical VPRs.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 4.
Fig. 4.

The VPRs (bold black curves): (a)–(c) the Wakefield, VA (KAKQ); Morehead City, NC (KMHX); and Raleigh–Durham, NC (KRAX) radars, respectively, at 1600 UTC 18 Sep 2003; (d)–(f) same as in (a)–(c) but at 1800 UTC. The locations of the radars are shown in Fig. 5. From top to bottom the horizontal lines represent the heights of the −20°, −10°, 0°, 10°, and 20°C isotherms at the radar sites, respectively. The 20°C isotherm is not shown in (c), (e), or (f) because there the surface temperature is lower than 20°C.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 5.
Fig. 5.

(a), (b) Hybrid scan reflectivity and (c), (d) echo classification for Hurricane Isabel at 1600 and 1800 UTC on 18 Sep 2003. For the echo classification fields, the light blue area represents the stratiform precipitation, the red represents convective rainfall, and the green area is the tropical rainfall. The large letters represent the radar names and their locations. The domain is 37°N, 82°W (NW corner) to 32°N, 75°W (SE corner).

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 6.
Fig. 6.

Distribution of the bias (= AR/AG) of individual R–G pairs for 24-h accumulations ending at (a), (b) 0100 and at (c), (d) 0500 UTC on 19 September 2003. Radar estimations used in (a), (c) and (b), (d) columns are from the CS and CST schemes, respectively. The color reflects the variation of the bias. The warm colors (orange/red) are for biases of less than 1 (radar underestimate). The cold colors (blue/purple) are for biases greater than 1 (radar overestimate). The white colors are for biases very close to 1; i.e., the radar agrees well with the gauge. The size of the circle represents the gauge amount. The domain is the same as in Fig. 5.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 7.
Fig. 7.

Scatterplot of radar vs gauge results for the Hurricane Isabel case in 24-h accumulations ending at (a), (b) 0100 and (c), (d) 0500 UTC on 19 September 2003. Radar estimations used in (a), (c) and (b), (d) columns are from the CS and CST schemes, respectively. The one-to-one line (dashed line) is the line of perfect correlation. The R–G pairs (data points) are all of the radar–gauge pairs in which the gauge accumulations are greater than 0.8 mm. The linear fit (solid line) is the linear regression line. The bias, RMSE, and CCs are also shown in the plots.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 8.
Fig. 8.

Same as in Fig. 4 but for VPRs valid at 2300 UTC on 26 Mar 2007: (a), (b) results from the KEWX and KGRK radars, respectively, are identified as tropical; (c), (d) results from the KMAF and KDFX radars, respectively, are nontropical. The locations of the radars are shown in Fig. 9.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 9.
Fig. 9.

Same as in Fig. 5 but for (a) hybrid scan reflectivity and (b) echo classification from the TX rainfall case at 2300 UTC on 26 Mar 2007. The domain is from 34°N, 103.5°W (NW corner) to 26.5°N, 93.5°W (SE corner). The yellow represents the stratiform area with a hybrid scan height above the frozen level. The purple is for hail.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 10.
Fig. 10.

As in Fig. 7 but for TX rainfall case in 24-h accumulations ending at 1200 UTC on 27 Mar 2007.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 11.
Fig. 11.

Same as in Fig. 4 but for VPRs from the Key West, FL (KBYX), radar valid at (a) 0100 UTC (convective) and (b) 1600 UTC (tropical) on 1 Jun 2007, and at (c) 1400 UTC on 2 Jun 2007 (tropical).

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 12.
Fig. 12.

As in Fig. 7 but for the Tropical Strom Barry case in 24-h accumulations between 1200 UTC 2 Jun and 1200 UTC 3 Jun 2007.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 13.
Fig. 13.

As in Fig. 6 but for the NE OK heavy rain case in 1-h accumulations ending at 1000 UTC, at 1300 UTC, and in 6-h accumulations ending at 1500 UTC on 11 Jun 2007. The domain is from 38°N, 98°W (NW corner) to 36°N, 94°W (SE corner).

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 14.
Fig. 14.

As in Fig. 7 but for the NE OK heavy rain case in 1-h accumulations ending (a), (b) 1000 UTC (c), (d) at 1300 UTC, and (e), (f) in 6-h accumulations ending at 1500 UTC on 11 Jun 2007.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 15.
Fig. 15.

Same as in Fig. 4 but for VPRs from the KFWS radar valid at (a) 0500, (b) 0600, and (c) 0700 UTC on 18 Jun 2007. They are all identified as tropical.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Fig. 16.
Fig. 16.

As in Fig. 7 except for the Gainesville north TX case in 12-h accumulations from 0100 to 1300 UTC on 18 Jun 2007.

Citation: Journal of Hydrometeorology 9, 5; 10.1175/2007JHM954.1

Table 1.

Extreme rainfall case studies.

Table 1.
Table 2.

Summary of gauge–radar comparisons using the two schemes.

Table 2.
Save
  • Aiken, R., 2000: Performance of WSR-88D during Hurricanes Fran and Floyd rainfall measurements Presentation on 2000 TAC Report. NOAA/National Weather Service/Radar Operations Center, Wilmington, NC. [Available online at http://www.roc.noaa.gov/app/tac/past_mtgs/May_2000_Report.asp.].

  • Anagnostou, E. N., and Krajewski W. F. , 1998: Calibration of the WSR-88D precipitation processing subsystem. Wea. Forecasting, 13 , 396406.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Anagnostou, E. N., and Krajewski W. F. , 1999: Real-time radar rainfall estimation. Part II: Case study. J. Atmos. Oceanic Technol., 16 , 198205.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Austin, P., 1987: Relation between measured radar reflectivity and surface rainfall. Mon. Wea. Rev., 115 , 10531070.

  • Austin, P. M., Geotis S. G. , Cunning J. B. , Thomas J. L. , Sax R. I. , and Gillespie J. R. , 1976: Raindrop size distributions and Z-R relationsip for GATE. Preprints, 10th Conf. on Hurricanes and Tropical Meteorology, Charlottesville, VA, Amer. Meteor. Soc., 165–166.

  • Baeck, M., and Smith J. , 1998: Rainfall estimation by the WSR-88D for heavy rainfall events. Wea. Forecasting, 13 , 416436.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and Coauthors, 2004: An hourly assimilation–forecast cycle: The RUC. Mon. Wea. Rev., 132 , 495518.

  • Fulton, R., Breidenbach J. , Seo D-J. , Miller D. , and O’Bannon T. , 1998: The WSR-88D rainfall algorithm. Wea. Forecasting, 13 , 377395.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Germann, U., and Joss J. , 2002: Mesobeta profiles to extrapolate radar precipitation measurements above the Alps to the ground level. J. Appl. Meteor., 41 , 542557.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Houze R. A. Jr., , 1989: Observed structure of mesoscale convective systems and implications for large-scale heating. Quart. J. Roy. Meteor. Soc., 115 , 425461.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Houze R. A. Jr., , 1993: Cloud Dynamics. Academic Press, 573 pp.

  • Hudlow, M., 1979: Mean rainfall patterns for the three phases of GATE. J. Appl. Meteor., 18 , 16561669.

  • Hunter, S. M., 1996: WSR-88D radar rainfall estimation Capabilities, limitations and potential improvements. Natl. Wea. Dig., 20 , 4. 2638.

    • Search Google Scholar
    • Export Citation
  • Joss, J., and Waldvogel A. , 1990: Precipitation measurement and hydrology. Radar in Meteorology Battan Memorial and 40th Anniversary Radar Meteorology Conference, D. Atlas, Ed., Amer. Meteor. Soc., 577–606.

    • Crossref
    • Export Citation
  • Klazura, G. E., Thomale J. M. , Kelly D. S. , and Jendrowski P. , 1999: A comparison of NEXRAD WSR-88D radar estimates of rain accumulation with gauge measurements for high- and low-reflectivity horizontal gradient precipitation events. J. Atmos. Oceanic Technol., 16 , 18421850.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kucera, P. A., Short D. A. , and Thiele O. W. , 1996: An analysis of rainfall intensity and vertical structure from shipborne radars TOGA COARE. Preprints, Seventh Conf. on Mesoscale Processes, Reading, United Kingdom, Amer. Meteor. Soc., 131–134.

  • Marshall, J. S., and Palmer W. Mc K. , 1948: The distribution of raindrops with size. J. Meteor., 5 , 165166.

  • Ryzhkov, A. V., and Zrnić D. S. , 1996: Rain in shallow and deep convection measured with a polarimetric radar. J. Atmos. Sci., 53 , 29892995.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ryzhkov, A. V., Schuur T. J. , Burgess D. W. , Heinselman P. L. , Giangrande S. E. , and Zrnić D. S. , 2005: The Joint Polarization Experiment. Bull. Amer. Meteor. Soc., 86 , 809824.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Steiner, M., Houze R. A. Jr., and Yuter S. E. , 1995: Climatological characterization of three-dimensional storm structure from operational radar and rain gauge data. J. Appl. Meteor., 34 , 19782007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tokay, A., and Short D. A. , 1996: Evidence from tropical raindrop spectra of the origin of rain from stratiform versus convective clouds. J. Appl. Meteor., 35 , 355371.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Uijlenhoet, R., Steiner M. , and Smith J. A. , 2003: Variability of raindrop size distribution in a squall line and implications for radar rainfall estimation. J. Hydormeteor., 44 , 4361.

    • Search Google Scholar
    • Export Citation
  • Ulbrich, C. W., and Lee L. G. , 1999: Rainfall measurement error by WSR-88D radars due to variation in ZR law parameters and the radar constant. J. Atmos. Oceanic Technol., 16 , 10171024.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vieux, B., and Bedient P. , 1998: Estimation of rainfall for flood prediction from WSR-88D reflectivity: A case study, 17–18 October 1994. Wea. Forecasting, 13 , 407415.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vignal, B., Galli G. , Joss J. , and Germann U. , 2000: Three methods to determine profiles of reflectivity from volumetric radar data to correct precipitation estimates. J. Appl. Meteor., 39 , 17151726.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wilson, J. W., and Brandes E. A. , 1979: Radar measurement of rainfall: A summary. Bull. Amer. Meteor. Soc., 60 , 10481058.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wood, L. T., 1997: Using the tropical Z-R relationship to improve precipitation estimates during a heavy rain event in southeast Texas. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 208–209.

  • Zawadzki, I., 1984: Factors affecting the precision of radar measurements of rain. Preprints, 22nd Int. Conf. on Radar Meteorology, Zurich, Switzerland, Amer. Meteor. Soc., 251–256.

  • Zhang, J., Langston C. , and Howard K. , 2008: Brightband identification based on vertical profiles of reflectivity from WSR-88D. J. Atmos. Oceanic Technol., 25 , 18591872.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zipser, E. J., and Lutz K. R. , 1994: The vertical profile of radar reflectivity of convective cells: A strong indicator of storm intensity and lightning probability? Mon. Wea. Rev., 122 , 17511759.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    Conceptual models of tropical VPRs: (a) reflectivity monotonically increases as the height decreases with a maximum at the lowest level and there exists a bright band where (b) reflectivity increases or (c) remains constant with the decrease of height below the bottom of the bright band. The solid line is the VPR. The dashed line is the 0°C isotherm.

  • Fig. 2.

    A flow chart of the automated scheme used to identify tropical VPRs.

  • Fig. 3.

    A flow chart of the delineation of tropical rainfall regions based on tropical VPRs.

  • Fig. 4.

    The VPRs (bold black curves): (a)–(c) the Wakefield, VA (KAKQ); Morehead City, NC (KMHX); and Raleigh–Durham, NC (KRAX) radars, respectively, at 1600 UTC 18 Sep 2003; (d)–(f) same as in (a)–(c) but at 1800 UTC. The locations of the radars are shown in Fig. 5. From top to bottom the horizontal lines represent the heights of the −20°, −10°, 0°, 10°, and 20°C isotherms at the radar sites, respectively. The 20°C isotherm is not shown in (c), (e), or (f) because there the surface temperature is lower than 20°C.

  • Fig. 5.

    (a), (b) Hybrid scan reflectivity and (c), (d) echo classification for Hurricane Isabel at 1600 and 1800 UTC on 18 Sep 2003. For the echo classification fields, the light blue area represents the stratiform precipitation, the red represents convective rainfall, and the green area is the tropical rainfall. The large letters represent the radar names and their locations. The domain is 37°N, 82°W (NW corner) to 32°N, 75°W (SE corner).

  • Fig. 6.

    Distribution of the bias (= AR/AG) of individual R–G pairs for 24-h accumulations ending at (a), (b) 0100 and at (c), (d) 0500 UTC on 19 September 2003. Radar estimations used in (a), (c) and (b), (d) columns are from the CS and CST schemes, respectively. The color reflects the variation of the bias. The warm colors (orange/red) are for biases of less than 1 (radar underestimate). The cold colors (blue/purple) are for biases greater than 1 (radar overestimate). The white colors are for biases very close to 1; i.e., the radar agrees well with the gauge. The size of the circle represents the gauge amount. The domain is the same as in Fig. 5.

  • Fig. 7.

    Scatterplot of radar vs gauge results for the Hurricane Isabel case in 24-h accumulations ending at (a), (b) 0100 and (c), (d) 0500 UTC on 19 September 2003. Radar estimations used in (a), (c) and (b), (d) columns are from the CS and CST schemes, respectively. The one-to-one line (dashed line) is the line of perfect correlation. The R–G pairs (data points) are all of the radar–gauge pairs in which the gauge accumulations are greater than 0.8 mm. The linear fit (solid line) is the linear regression line. The bias, RMSE, and CCs are also shown in the plots.

  • Fig. 8.

    Same as in Fig. 4 but for VPRs valid at 2300 UTC on 26 Mar 2007: (a), (b) results from the KEWX and KGRK radars, respectively, are identified as tropical; (c), (d) results from the KMAF and KDFX radars, respectively, are nontropical. The locations of the radars are shown in Fig. 9.

  • Fig. 9.

    Same as in Fig. 5 but for (a) hybrid scan reflectivity and (b) echo classification from the TX rainfall case at 2300 UTC on 26 Mar 2007. The domain is from 34°N, 103.5°W (NW corner) to 26.5°N, 93.5°W (SE corner). The yellow represents the stratiform area with a hybrid scan height above the frozen level. The purple is for hail.

  • Fig. 10.

    As in Fig. 7 but for TX rainfall case in 24-h accumulations ending at 1200 UTC on 27 Mar 2007.

  • Fig. 11.

    Same as in Fig. 4 but for VPRs from the Key West, FL (KBYX), radar valid at (a) 0100 UTC (convective) and (b) 1600 UTC (tropical) on 1 Jun 2007, and at (c) 1400 UTC on 2 Jun 2007 (tropical).

  • Fig. 12.

    As in Fig. 7 but for the Tropical Strom Barry case in 24-h accumulations between 1200 UTC 2 Jun and 1200 UTC 3 Jun 2007.

  • Fig. 13.

    As in Fig. 6 but for the NE OK heavy rain case in 1-h accumulations ending at 1000 UTC, at 1300 UTC, and in 6-h accumulations ending at 1500 UTC on 11 Jun 2007. The domain is from 38°N, 98°W (NW corner) to 36°N, 94°W (SE corner).

  • Fig. 14.

    As in Fig. 7 but for the NE OK heavy rain case in 1-h accumulations ending (a), (b) 1000 UTC (c), (d) at 1300 UTC, and (e), (f) in 6-h accumulations ending at 1500 UTC on 11 Jun 2007.

  • Fig. 15.

    Same as in Fig. 4 but for VPRs from the KFWS radar valid at (a) 0500, (b) 0600, and (c) 0700 UTC on 18 Jun 2007. They are all identified as tropical.

  • Fig. 16.

    As in Fig. 7 except for the Gainesville north TX case in 12-h accumulations from 0100 to 1300 UTC on 18 Jun 2007.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 539 287 29
PDF Downloads 214 51 3