Exaggerated Storm Total Precipitation Radar Estimates From Bright Band Interaction

William D. Marino
National Weather Service

Grand Rapids, Michigan

1. Introduction

The "bright band" is an area of enhanced reflectivity, seen on a PPI display, which typically results from the radar beam passing through a layer of melting snow. "Bright band" enhancement usually occurs during the cool season, in stratiform precipitation events. The existence of a "bright band" can cause exaggerated radar rainfall estimates on the WSR-88D Storm Total Precipitation (STP) product. On May 8, 1994, just such an event occurred in Lower Michigan. On this day, the WSR-88D STP product estimated a maximum total rainfall of 4.7 inches within the White Lake, Michigan (DTX) WSR-88D radar umbrella. The maximum observed rainfall was 0.76 inches. The purpose of this paper is to present a case where the Hybrid Scan Construction of the STP, interacting with the melting level, was responsible for approximately a 500 percent error in the maximum radar rainfall estimate.

2. Data

The valid period of the WSR-88D STP estimate was from 1105 UTC 7 May to 0627 UTC 8 May 1994. Actual rainfall totals were measured at the official rain gage at Detroit Metropolitan Airport (DTW) in west central Wayne County, plus 12 climate/agricultural gages and two supplemental rain gages including the Pontiac, Michigan Automated Surface Observing System (PTK ASOS).

The WSR-88D radar is located in White Lake, Michigan in west-central Oakland County. It is located on a hill with an elevation of 1500 ft AGL, which is about 150 ft above the mean terrain in the area.

Rain gage observations were taken at 1200 UTC each day except for DTW, which measures rainfall every three hours on the synoptic hours, and one supplemental gage measured at midnight LST. Most of the rainfall from this event occurred from about 1600 UTC on May 6, 1994 until about 0400 UTC on May 8, 1994. The Pontiac ASOS hourly rainfall accumulation (Figure 1) indicated most of the rain occurred between 1730 UTC and 2330 UTC. The PTK ASOS also showed that light rain continued to fall until 0330 UTC on May 8, 1994.

The Flint, Michigan (FNT) radiosonde observation on May 8, 1994 at 0000 UTC was used to estimate the height of the melting level. This observation also coincided with the time frame that the heaviest rainfall rates were occurring.

3. Synoptic Overview

At 0600 UTC, on May 7, 1994 a stationary front was located from southeast Missouri to southeast Ohio. By 1200 UTC, a low (as seen in the MSLP field) had developed over central Illinois. A warm front extended from the low, into southern West Virginia (Figure 2). The low pressure wave then moved east along the front to south central Ohio by 1800 UTC.

*Figure 1. ASOS observations at Pontiac, Michigan (PTK) on May 7, 1994.

*Figure 2. Surface map (MSLP isobars) 1200 UTC 8 May 1994.

During this entire scenario, a 60 to 80 mile wide area of light rain was occurring along the entire length of the frontal system. Near the low pressure wave, the rain area then extended north for about 300 miles. That area of rain also extended northwest of the system for another 150 miles.

A few light rain showers developed ahead of the main precipitation area during the morning hours of May 7. However, the main rain area moved into southeast Michigan just before 1600 UTC. The maximum rainfall rates occurred as the main surface wave passed south of the Pontiac area during the early afternoon on May 7. From the late afternoon of May 7, until the early morning hours of May 8, a trailing light rain area remained over southeast Lower Michigan.

The surface geostrophic wind plots (not shown) indicated the presence of a cold conveyor belt over southeast Lower Michigan. Maximum low level winds extended on an axis from near Lake St. Clair, across the northern suburbs of Detroit, to near Jackson in south central Lower Michigan. It was just north of this area that the maximum rainfall occurred.

The 0000 UTC 8 May 1994 FNT radiosonde observation indicated that the melting level was at 5000 ft AGL. Temperatures at the surface, in the rain area, were mostly from the mid 40s to around 50F. This allowed the precipitation to fall to the ground as rain for the entire event.

4. The Bright Band

The bright band is a region of relatively high equivalent reflectivity that usually appears as an elevated layer at the height where falling ice particles begin to melt and thus become water coated. This bright band region depicts the melting layer and is often about 3000 feet in depth (Green 1993).

As frozen precipitate fall into an area with temperatures in the -5C to 0C range there is a rapid increase in the coalescence of individual snow crystals. Then, as snowflakes exit sub-freezing air, they gradually melt to raindrops and increase their fall speed. Ice particles exhibit about one-fifth of the reflectivity as the equivalent amount of liquid water (Green 1993). So, when the snowflakes fall into near-freezing air, the bigger flakes begin to exhibit greater radar reflectivity. As the particles fall below the melting level they become coated with liquid water. This causes a fivefold increase in the efficiency of the particles to return energy to the radar (Green 1993). This is the primary cause of the bright band. After the particles completely melt into rain, their fall speed increases rapidly resulting in a decrease of precipitation partiale concentration. Radar reflectivity again decreases (Martner, et al. 1993).

The schematic in Figure 3, done for the "Lake Ontario Winter Storms (LOWS) Project" (Martner et al. 1993), show a southwest to northeast cross section through a warm front. Note the relationship between the bright band and the melting level. At location "A", the precipitation falls as all snow where the temperature through the entire layer is <0C. At location "B", there is some melting, enough at mid levels to melting the snow to rain before it refreezes into sleet. This could create bright band contamination of radar precipitation estimates depending on the distance from the radar and the height of the melting level. Location "C" has enough warming to melt the snow. Cold air is not deep enough to refreeze the precipitation to sleet, but the surface air temperature is cold enough for the occurrence of freezing rain. In this instance there could be substantial bright band contamination of radar reflectivity values. Finally, at location "D", there is enough warming to melt the snow to rain but no refreeze occurs. This was believed to be the situation across in southeast Michigan on May 7, 1994.

5. Hybrid Scan Construction

The STP algorithm uses hybrid scan construction to utilize, for each azimuth and range, the best possible reflectivity values. The radar uses the four lowest elevation tilts for its conversion of radar reflectivities to precipitation amounts. The intent is to keep the sample area around 3000 feet above the ground.

*Figure 3. Schematic showing the location of the bright band in a typical precipitation event (Martner, et al. 1993)

Figure 4 depicts tilt angles and the height above the ground at various ranges from the radar. From zero to 11 nm from the radar, the 3.4 degree tilt is used in the hybrid scan construction. Its vertical depth covers zero to 5000 ft AGL. From 11 nm to 19 nm, the 2.4 degree tilt is utilized, resulting in an elevation from 4000 to 7000 feet AGL. At 19 miles to 27 miles, the 1.5 degree angle is used, where the heights are also from about 4000 feet to 7000 feet AGL. Finally from 27 nm to the end of the scan at 125 miles, both the 1.5 and 2.5 degree angles are used. At 27 miles this scan is from 4,000 ft AGL to 8,000 ft, but by 125 miles it is from 11000 ft to about 35,000 ft. On the 27 to 125 nm scan, the greater reflectivity from either the 1.5 degree or .5 degree are used for precipitation estimation. This is called bi-scan maximization (FMH11 Part C, 1991).

*Figure 4. WSR-88D Hybrid Scan (dark shaded area is where the precipitation is measured).

6. Results

Comparing the actual rainfalls to the radar estimated rainfall (Figure 5a), there is a fivefold increase in the radar estimated rainfall in the circular shaped heavy rain bands. That is the signature of the bright band on the STP product. If the radar estimates for this event over southern Lower Michigan were taken as truth, estimates of 4.7 inches of rainfall would have had serious operational ramifications for forecasters.

*Figure 5a. Observed precipitation amounts from May 7, 1994.

Note that the hybrid scan construction's first break in tilt angle is at 11 nm. It is also at 11 nm where the end of the first excessive radar rainfall estimate ring is (Figure 5b). The second break in the hybrid scan construction is at 19 nm, which again, coincides with the end of the next excessive rainfall ring. At 25 nm note the WSR-88D STP product suddenly begins to underestimate rainfall totals. This is to close to where the final scan angles break off, and represents where the radar beam is passing through frozen precipitate.

*Figure 5b. WSR-88D storm total precipitation estimates 11055 UTC 7 May to 0626 UT 8 May 1994.

With the 0000 UTC 8 May 1994 FNT radiosonde observation indicating that the melting level was at 5000 ft, it is not surprising that the STP product maximum precipitation bands are located as they are. The bright band on the reflectivity cross-section from 1952 UTC (Figure 6) is clearly indicated from 5000 ft AGL to near the ground. So, the maximum precipitation cylinders are exactly where the "bright band" reflectivity areas intersect the hybrid scan tilts. Where the tilt used in the hybrid scan tilts is above the freezing level, the estimated precipitation amounts are less than the actual rainfall amounts. This is because precipitation at those altitudes is actually snow. The location of observed mixed precipitation coincides with the location of the greatest radar estimated rainfall. Finally, where the observed precipitation was all rain in the hybrid scan, STP precipitation estimates were fairly reliable (area over northern Ohio not shown on Figure 5b).

Figure 6. WSR-88D reflectivity cross-section 1952 UTC 7 May 1994.

7. Conclusions

The May 7, 1994 case demonstrated that the "bright band" can cause significant overestimation of radar-derived precipitation estimates. These excessive precipitation estimates occur when the WSR-88D radar hybrid scan construction (used in the radar precipitation estimate algorithms) intersects the melting level. This occurs mostly when the melting level is between the ground and 6000 feet AGL. When using the WSR-88D STP product forecasters should know if the precipitation is convective or not and if the melting level, anywhere over the radar umbrella, would interact with the STP hybrid scan construction to contaminate rainfall estimates and/or other related products. Understanding the purpose of the WSR-88D hybrid scan construction and knowledge of the radar reflectivity characteristics of melting precipitation are keys to ascertaining the accuracy of the radar precipitation estimates.

8. Acknowledgments

The author wishes to acknowledge the efforts of Mark Walton and Dick Wagenmaker in the review and editing of this manuscript.

9. References

DOC/NOAA/NWS, 1991: Federal Meteorological Handbook No 11 (Interim Version One), Part C, WSR-88D Products and Algorithms, Doppler Radar Meteorological Observations, 3-6 - 3-17, 5-1 - 5-7.

Green, D.G., 1990: Principles of Weather Radar, NWS Operations Training Facility, Norman, Oklahoma. WSR-88D OPERATIONS COURSE, 73pp.

Martner, B.E., J.B. Snider, R.J. Zamora, 1993: A remote-sensing view of a freezing-rain storm. Mon. Wea. Rev., 121, 2562-2577.

National Weather Service Regional Office