• Adler, R. F., and R. A. Mack, 1986: Thunderstorm cloud top dynamics as inferred from satellite observations and a cloud top parcel model. J. Atmos. Sci.,43, 1945–1960.

    • Crossref
    • Export Citation
  • ——, M. J. Markus, and D. D. Fenn, 1985: Detection of severe Midwest thunderstorms using geosynchronous satellite data. Mon. Wea. Rev.,113, 769–761.

    • Crossref
    • Export Citation
  • Crum, T. D., and R. L. Alberty, 1993: The WSR-88D and the WSR-88D operational support facility. Bull. Amer. Meteor. Soc.,74, 1669–1687.

    • Crossref
    • Export Citation
  • Doswell, C. A., III, 1985: The operational meteorology of convective weather. Volume II: Storm scale analysis. NOAA Tech. Memo. ERL ESG-15, 240 pp. [Available from NSSL, 1313 Halley Cir., Norman, OK 73069.].

  • Eilts, M. D., and R. J. Doviak, 1987: Oklahoma downbursts and their asymmetry. J. Climate Appl. Meteor.,26, 69–78.

    • Crossref
    • Export Citation
  • Howard, K. W., J. J. Gourley, and R. A. Maddox, 1997: Uncertainties in WSR-88D measurements and their impacts on monitoring thunderstorm life cycles. Wea. Forecasting,12, 166–174.

    • Crossref
    • Export Citation
  • Johnson, J. T., P. L. MacKeen, A. Witt, E. D. Mitchell, G. J. Stumpf, M. D. Eilts, and K. W. Thomas, 1998: The Storm Cell Identification and Tracking (SCIT) algorithm: An enhanced WSR-88D algorithm. Wea. Forecasting,13, 263–276.

    • Crossref
    • Export Citation
  • Sohl, C. J., E. M. Quetone, and L. R. Lemon, 1996: Severe storm warning decisions: Operational impact of multiple radars. Preprints, 18th Conf. on Severe Local Storms, San Francisco, CA, Amer. Meteor. Soc., 561–564.

  • Waldvogel, A., W. Schmid, and P. Grimm, 1979: Criteria for the detection of hail cells. J. Appl. Meteor.,18, 1521–1525.

    • Crossref
    • Export Citation
  • Witt, A., M. D. Eilts, G. J. Stumpf, J. T. Johnson, E. D. Mitchell, and K. W. Thomas, 1998: An enhanced hail detection algorithm for the WSR-88D. Wea. Forecasting,13, 286–303.

    • Crossref
    • Export Citation
  • Zaras, D. S., R. A. Rabin, R. A. Maddox, and P. L. MacKeen, 1998:Integration of GOES-8/9 and multiple WSR-88D for monitoring long-lived severe convective storms. Preprints, Second Symp. on Integrated Observing Systems, Phoenix, AZ, Amer. Meteor. Soc., 16–19.

  • View in gallery
    Fig. 1.

    Radar beam geometry (2D) for range versus height. The elevation angles for the center of each beam are shown. Regions in which no data are gathered are indicated by black shading: (a) VCP-11 and (b) VCP-21.

  • View in gallery
    Fig. 2.

    Same as Fig. 1 when data are combined from two adjacent radars; example is for targets located somewhere along straight line connecting the radars. Regions in which no data are collected are in black; regions sampled by one radar are shown in gray; regions sampled by both radars are shown in white: (a) both radars in VCP-21, (b) both in VCP-11, (c) one in each VCP.

  • View in gallery
    Fig. 3.

    Radar-detected heights for features located at 4, 10, and 16 km above ground level (AGL). Solid red line shows height detected by single radar. Dashed blue line shows detected height when data from adjacent radars 300 km apart are used in combination. Targets are located somewhere along straight line connecting the two radars:(a) both radars in VCP-21 and (b) one radar in each VCP.

  • View in gallery
    Fig. 4.

    The track and evolution of the Ardmore supercell during the period studied. Letters indicate the time at 1-h intervals with A being 1900 UTC.

  • View in gallery
    Fig. 5.

    Storm-top height determined from the SCIT algorithm for both the Fort Worth radar (KFWS) shown by solid teal and blue lines and the Norman radar (KTLX) shown by dashed orange line. The elevation angles and heights of the radar beam centers are also shown for both radars (thin lines). Periods when tornadoes were occurring are indicated. Note that top heights are not always exactly on center of beam heights; this is because SCIT-determined echo top location is not always directly above SCIT-determined cell position (see Johnson et al. 1998 for details). Height uncertainties are illustrated by arrows at left side of figure; red plus black arrows are uncertainty in KFWS data and red arrows are uncertainty if data from both radars were analyzed.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 198 52 2
PDF Downloads 57 24 1

Echo Height Measurements with the WSR-88D: Use of Data from One Versus Two Radars

R. A. MaddoxNOAA/Environmental Research Laboratories, National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by R. A. Maddox in
Current site
Google Scholar
PubMed
Close
,
D. S. ZarasNOAA/Environmental Research Laboratories, National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by D. S. Zaras in
Current site
Google Scholar
PubMed
Close
,
P. L. MacKeenNOAA/Environmental Research Laboratories, National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by P. L. MacKeen in
Current site
Google Scholar
PubMed
Close
,
J. J. GourleyNOAA/Environmental Research Laboratories, National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by J. J. Gourley in
Current site
Google Scholar
PubMed
Close
,
R. RabinNOAA/Environmental Research Laboratories, National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by R. Rabin in
Current site
Google Scholar
PubMed
Close
, and
K. W. HowardNOAA/Environmental Research Laboratories, National Severe Storms Laboratory, Norman, Oklahoma

Search for other papers by K. W. Howard in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

The new Doppler radars of the National Weather Service (i.e., the WSR-88D radars) are operated continuously in a volume scanning mode (called Volume Coverage Pattern, VCP) with the elevation tilt angles fixed for several VCPs. Because of the fixed VCPs, the radar data can be used to determine the heights of precipitation echo features only to limits of accuracy that depend upon the elevation angles used in the VCP, the radar beamwidth, and the range of echoes. Data from adjacent WSR-88D radars, if used simultaneously, could reduce significantly the height uncertainties inherent in single radar measurements. This is illustrated for idealized situations and also for an event involving a long-lived, tornadic thunderstorm. The use of coordinated scan strategies and combined data analysis procedures for adjacent WSR-88D radars during significant thunderstorm events should be considered for operational applications.

* Additional affiliation: University of Oklahoma and Cooperative Institute for Mesoscale Meteorological Studies, Norman, Oklahoma.

Visiting Scientist, National Weather Service Forecast Office, Tucson, Arizona.

Corresponding author address: Dr. R. A. Maddox, ℅ NWS Office, 520 N. Park Ave., Suite 304, Tucson, AZ 85719-5035.

Email: rmaddox@azstarnet.com

Abstract

The new Doppler radars of the National Weather Service (i.e., the WSR-88D radars) are operated continuously in a volume scanning mode (called Volume Coverage Pattern, VCP) with the elevation tilt angles fixed for several VCPs. Because of the fixed VCPs, the radar data can be used to determine the heights of precipitation echo features only to limits of accuracy that depend upon the elevation angles used in the VCP, the radar beamwidth, and the range of echoes. Data from adjacent WSR-88D radars, if used simultaneously, could reduce significantly the height uncertainties inherent in single radar measurements. This is illustrated for idealized situations and also for an event involving a long-lived, tornadic thunderstorm. The use of coordinated scan strategies and combined data analysis procedures for adjacent WSR-88D radars during significant thunderstorm events should be considered for operational applications.

* Additional affiliation: University of Oklahoma and Cooperative Institute for Mesoscale Meteorological Studies, Norman, Oklahoma.

Visiting Scientist, National Weather Service Forecast Office, Tucson, Arizona.

Corresponding author address: Dr. R. A. Maddox, ℅ NWS Office, 520 N. Park Ave., Suite 304, Tucson, AZ 85719-5035.

Email: rmaddox@azstarnet.com

1. Introduction

Operational weather radars are used routinely to measure the heights to which precipitation echoes, particularly thunderstorms, reach in the vertical. The height of the top of convective clouds is related to updraft strength and thus provides some measure of probable storm intensity (e.g., Doswell 1985, cf. chap. III E). Furthermore, the height to which strong radar reflectivities grow is often related to the occurrence of hail at the surface (Waldvogel et al. 1979; Witt et al. 1998). Rapid decreases in the height of intense reflectivity features have been related to the occurrence of microbursts at the surface (e.g., Eilts and Doviak 1987).

The new Doppler radars of the National Weather Service [i.e., the Weather Surveillance Radar-1988 Doppler (WSR-88D) radars; see Crum and Alberty (1993)] are operated continuously in a volume scanning mode (called Volume Coverage Pattern, VCP) with successive elevation angles (often referred to as “tilts”) fixed for several VCPs. Those used when thunderstorms are in progress are either VCP-11 (14 elevation scans in 5 min) or VCP-21 (9 elevation scans in 6 min). These VCPs are illustrated in Fig. 1. The angular separation between successive tilts causes considerable gaps in the vertical sampling of precipitation systems. For example, in VCP-11 (Fig. 1a), there is considerable separation between the radar beams at higher tilts from 25- to 100-km range. When the radars scan in VCP-21 (Fig. 1b), there are very large vertical gaps in coverage from 25- to beyond 150-km range.

Current WSR-88D software (i.e., Build 10) in the operational system has increased use of “feature” height information, which is incorporated into several algorithms, for example, the Hail Detection Algorithm and Storm Cell Identification and Tracking (SCIT) algorithms; see Witt et al. (1998) and Johnson et al. (1998). The time trend of the height of cell features is a much-used display capability of the operational system. Forecasters using the system can monitor evolution of the strength or height of radar detected reflectivity attributes of thunderstorm cells identified by the SCIT algorithm, for example, maximum reflectivity, height of maximum reflectivity, echo top height, etc. The trends, if accurate, can provide useful information for diagnosing some storm structures and aspects of their evolution.

The “height” of features identified by various WSR-88D data processing algorithms are defined to be the height above the radar level (ARL) of the center of the relevant elevation scan radar beam (see, e.g., Howard et al. 1997). This is equivalent to assuming that the echo fills at least half of the beam volume; for cases where it does not, the uncertainties are less than those discussed here and of opposite sign. Since heights are defined to be center-of-beam height, there is uncertainty that is usually equal to half of the beam diameter. Fixed mode scanning of the operational radars precludes vertical scanning along a radial (i.e., RHIs are not possible) that would allow more precise determination of the vertical position of echo features, for example, echo top or height of maximum reflectivity. The radar data processing procedures provide height information (see, e.g., Witt et al. 1998), but the VCP-fixed separations between beam centers for each successive tilt angle mean that the height of features is determined imprecisely.

Large height uncertainties of WSR-88D radar measurements can make interpretation of heights and their trends difficult or impossible in some situations. This is especially true if the radars are operating in VCP-21. Sohl et al. (1996) have discussed advantages of using data from several radars to support warning decisions in National Weather Service (NWS) operations. They show several examples of storms whose structures could not be accurately determined unless the forecaster examined WSR-88D displays from several adjacent radars. In this paper we explore ways in which the uncertainties of WSR-88D height information could be reduced if data from adjacent radars were analyzed simultaneously.

2. Height uncertainties

Howard et al. (1997) show magnitudes of inherent uncertainties for both VCP-11 and 21 (see their Fig. 3), and discuss the implications for diagnosing the heights and trends of storm features with operational WSR-88D data. When the radar is operated in VCP-21, the height uncertainties can be as large as 7–8 km at ranges around 100–125 km. Emphasis is on VCP-21 because this scanning strategy is used by many central U.S. NWS offices during significant severe thunderstorm episodes. A survey of science operations officers at 20 offices, mostly in the central United States, indicates that about 25% of their offices tend to operate primarily in VCP-21 and that about 25% switch between VCP-21 and VCP-11 during severe thunderstorm situations. Thus, severe weather decisions are often based upon data obtained from WSR-88D radars operating in VCP-21.

The uncertainties inherent in WSR-88D height computations could be reduced if data from at least two adjacent radars were analyzed simultaneously. For example, beam coverage that results when data from two radars 300 km apart are combined is shown in Fig. 2. As in Fig. 1, the black regions have no radar data available. Comparison of the panels of Fig. 2 with those of Fig. 1 illustrates the dramatic increase in the sampling, or coverage, of the radar measurements in the vertical.

A very simple situation is assumed. The radars are at the same elevation; there is no terrain blockage of any beams; the beams are refracted as if propagating in a “standard atmosphere;” and features being monitored are located somewhere along the line connecting the two radars. Of course in actual application, horizontally varying terrain elevation makes the situation more complicated; the atmosphere is rarely standard; and storms seldom move directly from one radar toward an adjacent radar.

The height of echo features detected is at the center of the beam, as in WSR-88D algorithms. Finally, it is assumed that both radars can detect features of interest. While this is sometimes true, the rapidly increasing sample volume of the radar beam (e.g., the sample volume is about 0.13 km3 at 25-km range but is about 8.55 km3 at 200-km range) affects strongly the attributes and structures of storms that can be resolved. Thus, these analyses present a “best case” scenario of how height measurements might be improved by using data from multiple radars.

Figure 3 illustrates the radar detected heights for features at 4, 10, and 16 km above the radar, where it is assumed that these features can be detected at all ranges. For both radars in VCP-21 (Fig. 3a), it is clear that substantial reductions in the height uncertainty could, in some situations, be realized if data from both radars were used (e.g., examine the detected heights of the 10- and 16-km features from about 75- to 100-km range). The main advantage for lower-level features, that is, the feature at 4 km in this figure, is that the range over which monitoring is possible is much extended. For both radars operating in VCP-11 (not shown) the potential reductions in height uncertainty are fairly small out to ranges of about 175 km but then become quite substantial (i.e., reductions of 3–4 km in the height uncertainty at ranges from 225 to 250 km). If the radars are operating in different VCPs (Fig. 3b), the potential decrease in height uncertainty is substantial for features at heights of 10 and 16 km at almost all ranges.

3. Case example

On 7 May 1995, a severe, supercell thunderstorm developed west of Fort Worth, Texas, and moved rapidly north-northeastward at about 22 m s−1, eventually decaying in south-central Oklahoma, see Fig. 4. This long-lived, tornadic storm produced four tornadoes, $103 million in property damage, and three deaths (Zaras et al. 1998). The storm was within the range of both the Fort Worth (KFWS) and the Norman (KTLX) WSR-88D radars for extended periods of its life. Both radars were operating in VCP-21 during this severe weather event. Although the storm passed just to the east of the Department of Defense WSR-88D radar at Frederick, Oklahoma, the data were unfortunately not recorded.

The time trend of the cell-top height as produced by the SCIT1 algorithm is shown in Fig. 5 from both the KTLX and KFWS WSR-88D radars. (Both radars are assumed to be at same elevation; in reality the KTLX radar is 156 m higher than the KFWS radar.) There are periods during this storm for which SCIT data are available from only one of the radars. This figure illustrates very clearly the uncertainties present in the storm-top height trends produced by the algorithm. For example, from 1900 to 2000 UTC the uncertainty in the KFWS height measurements changes little and is large because the storm is moving nearly tangent to the radar scans at a range of about 100 km (refer back to Figs. 1b and 4). Whereas, from 2000 to 2100 UTC the trend from KFWS showing the increase in top heights reflects primarily the increasing height of the 6° beam above the ground as the storm moves away from the radar. Similarly, for most periods the top height trends from KTLX show the decreasing heights of the beam centers as the storm moves toward the radar, rather than real attributes of internal storm behavior.

It is very hard to determine much about the behavior of the top height of the Ardmore storm with certainty except that after 2230 UTC it was decreasing rapidly. At a number of times during the period when tornadoes were occurring, the top height trends from the two radars were of opposite sign. During the period of significant tornado occurrence, which was a critical period since the storm was moving from the area of responsibility of the Fort Worth office into that of the Norman office, it is not possible to use the WSR-88D radar data to determine accurately the top height of the thunderstorm (see Zaras et al. 1998 also).

The combined data indicate that the top of the storm remained somewhere in the range from 13 to 15 km above ground level (AGL) during the period from 2100 through 2230 UTC (the tornadic period). Prior to 2100 UTC the trends are only due to the storm’s moving away from Fort Worth and all that is certain is that the top is above 10 km. When both radars scan the storm, the top height is estimated with less uncertainty (refer to Table 1 and note the “uncertainty” arrows shown in Fig. 5). The rapid fluctuations around 2140 UTC in the Fort Worth top height appear to result from either SCIT cell identification problems or invalid SCIT associations of the tops of new, nearby cells with the primary echo.

It is important for the reader to understand that the cell of interest has been subjectively identified, and the simplified trends shown in Fig. 5 are from SCIT output for a number of different storm cells. During the period that the supercell was monitored by the Fort Worth radar the SCIT algorithm actually identified it as more than five different “cells” and the Norman radar identified the storm as two separate cells. Thus, the trends shown were not available at either forecast office. Additionally, since SCIT tracks cell features only for the latest 10 volume scans, only limited segments of the lifetime of this long-lived thunderstorm could be displayed on the operational trend plots. Regardless, the data are more representative of the behavior of the actual storm when information from both radars is combined and displayed for the entire life of the thunderstorm.

4. Discussion

It has been illustrated that height uncertainties inherent in volume scan WSR-88D data can be reduced significantly for VCP-21, and for VCP-21 and VCP-11 combined, if the observations from at least two adjacent radars are analyzed simultaneously. The reductions for VCP-11 are also significant but less than for the VCP-21 scan strategy. The determination and interpretation of evolving vertical storm structures is extremely difficult because WSR-88D trends in height measurements result from effects due to storm motion, storm evolution, and volume scan uncertainties that are all concatenated in time and space. The situation could be considerably improved if data from adjacent radars were analyzed simultaneously. We realize that such analyses would be very complex and require the development of algorithms employing much different logic than those of the current, single-radar-based algorithms.

A number of investigators have used satellite data to estimate the height and trend of the tops of severe thunderstorms (e.g., Adler et al. 1985; Adler and Mack 1986). Zaras et al. (1998) examined satellite data for the Ardmore storm case discussed in section 3 and concluded that the height of the storm remained relatively constant from 1900 to 2200 UTC (refer back to Fig. 5). Thus, the possibility of incorporating satellite data into severe storm detection algorithms should also be explored. Since computing power available in NWS operations continues to increase rapidly, we suggest that it is not premature to begin exploring the feasibility of algorithms that process data from multiple radars, and from other sensing systems, in real time.

Acknowledgments

We thank Les Lemon, Kim Elmore, Mike Eilts, and Doug Forsyth for their help and suggestions, which have improved this paper. We also thank Loretta McKibben, who has patiently worked with us to produce the final version of Fig. 5.

REFERENCES

  • Adler, R. F., and R. A. Mack, 1986: Thunderstorm cloud top dynamics as inferred from satellite observations and a cloud top parcel model. J. Atmos. Sci.,43, 1945–1960.

    • Crossref
    • Export Citation
  • ——, M. J. Markus, and D. D. Fenn, 1985: Detection of severe Midwest thunderstorms using geosynchronous satellite data. Mon. Wea. Rev.,113, 769–761.

    • Crossref
    • Export Citation
  • Crum, T. D., and R. L. Alberty, 1993: The WSR-88D and the WSR-88D operational support facility. Bull. Amer. Meteor. Soc.,74, 1669–1687.

    • Crossref
    • Export Citation
  • Doswell, C. A., III, 1985: The operational meteorology of convective weather. Volume II: Storm scale analysis. NOAA Tech. Memo. ERL ESG-15, 240 pp. [Available from NSSL, 1313 Halley Cir., Norman, OK 73069.].

  • Eilts, M. D., and R. J. Doviak, 1987: Oklahoma downbursts and their asymmetry. J. Climate Appl. Meteor.,26, 69–78.

    • Crossref
    • Export Citation
  • Howard, K. W., J. J. Gourley, and R. A. Maddox, 1997: Uncertainties in WSR-88D measurements and their impacts on monitoring thunderstorm life cycles. Wea. Forecasting,12, 166–174.

    • Crossref
    • Export Citation
  • Johnson, J. T., P. L. MacKeen, A. Witt, E. D. Mitchell, G. J. Stumpf, M. D. Eilts, and K. W. Thomas, 1998: The Storm Cell Identification and Tracking (SCIT) algorithm: An enhanced WSR-88D algorithm. Wea. Forecasting,13, 263–276.

    • Crossref
    • Export Citation
  • Sohl, C. J., E. M. Quetone, and L. R. Lemon, 1996: Severe storm warning decisions: Operational impact of multiple radars. Preprints, 18th Conf. on Severe Local Storms, San Francisco, CA, Amer. Meteor. Soc., 561–564.

  • Waldvogel, A., W. Schmid, and P. Grimm, 1979: Criteria for the detection of hail cells. J. Appl. Meteor.,18, 1521–1525.

    • Crossref
    • Export Citation
  • Witt, A., M. D. Eilts, G. J. Stumpf, J. T. Johnson, E. D. Mitchell, and K. W. Thomas, 1998: An enhanced hail detection algorithm for the WSR-88D. Wea. Forecasting,13, 286–303.

    • Crossref
    • Export Citation
  • Zaras, D. S., R. A. Rabin, R. A. Maddox, and P. L. MacKeen, 1998:Integration of GOES-8/9 and multiple WSR-88D for monitoring long-lived severe convective storms. Preprints, Second Symp. on Integrated Observing Systems, Phoenix, AZ, Amer. Meteor. Soc., 16–19.

Fig. 1.
Fig. 1.

Radar beam geometry (2D) for range versus height. The elevation angles for the center of each beam are shown. Regions in which no data are gathered are indicated by black shading: (a) VCP-11 and (b) VCP-21.

Citation: Weather and Forecasting 14, 3; 10.1175/1520-0434(1999)014<0455:EHMWTW>2.0.CO;2

Fig. 2.
Fig. 2.

Same as Fig. 1 when data are combined from two adjacent radars; example is for targets located somewhere along straight line connecting the radars. Regions in which no data are collected are in black; regions sampled by one radar are shown in gray; regions sampled by both radars are shown in white: (a) both radars in VCP-21, (b) both in VCP-11, (c) one in each VCP.

Citation: Weather and Forecasting 14, 3; 10.1175/1520-0434(1999)014<0455:EHMWTW>2.0.CO;2

Fig. 3.
Fig. 3.

Radar-detected heights for features located at 4, 10, and 16 km above ground level (AGL). Solid red line shows height detected by single radar. Dashed blue line shows detected height when data from adjacent radars 300 km apart are used in combination. Targets are located somewhere along straight line connecting the two radars:(a) both radars in VCP-21 and (b) one radar in each VCP.

Citation: Weather and Forecasting 14, 3; 10.1175/1520-0434(1999)014<0455:EHMWTW>2.0.CO;2

Fig. 4.
Fig. 4.

The track and evolution of the Ardmore supercell during the period studied. Letters indicate the time at 1-h intervals with A being 1900 UTC.

Citation: Weather and Forecasting 14, 3; 10.1175/1520-0434(1999)014<0455:EHMWTW>2.0.CO;2

Fig. 5.
Fig. 5.

Storm-top height determined from the SCIT algorithm for both the Fort Worth radar (KFWS) shown by solid teal and blue lines and the Norman radar (KTLX) shown by dashed orange line. The elevation angles and heights of the radar beam centers are also shown for both radars (thin lines). Periods when tornadoes were occurring are indicated. Note that top heights are not always exactly on center of beam heights; this is because SCIT-determined echo top location is not always directly above SCIT-determined cell position (see Johnson et al. 1998 for details). Height uncertainties are illustrated by arrows at left side of figure; red plus black arrows are uncertainty in KFWS data and red arrows are uncertainty if data from both radars were analyzed.

Citation: Weather and Forecasting 14, 3; 10.1175/1520-0434(1999)014<0455:EHMWTW>2.0.CO;2

Table 1.

Height uncertainties (km) for the Ardmore, OK, supercell.

Table 1.

1

The SCIT algorithm determines the maximum height that echo reflectivity of 30 dBZ or greater reaches for each identified storm cell. This height is referred to as the “storm top” and is determined relative to the elevation of the radar, that is the height is ARL (see Johnson et al. 1998 for details). Some WSR-88D products show heights above mean sea level (MSL) and there is the possibility for confusion. The analysis of data from multiple radars requires that heights be computed consistently and relative to MSL.

Save