E. Brandes. J. Vivekanandan and J. Wilson are conducting precipitation estimation research in three areas: (1) radar reflectivity-based techniques for estimating rainfall, (2) dual- polarimetric methods, and (3) satellite precipitation estimation techniques. All storm types (convective and stratiform, summer and winter, and orographic) are being considered. Goals are being met by conducting modest but highly-focused efforts in which polarimetric radars are collocated with operational WSR- 88D's. Ultimately this research will be coupled with efforts to improve quantitative precipitation forecasts.
The program began in the summer of 1996 when NCAR's S-band, dual- polarization radar (S-Pol) was placed within 2 km of the WSR-88D that services the Denver, Colorado metropolitan area. Observations were available from 113 raingages operated by the Urban Drainage and Flood Control District in Denver. The dataset includes 3 major rainfall events including the Buffalo Creek flash flood of 13 July 1996. During the spring of 1997 S-Pol was placed within 10 km of the Wichita, Kansas WSR-88D. Radar measurements and observations from a 70 raingage network are available for study. A storm on 13 June 1997 struck the community of Andover, Kansas with a flash flood. Rainfall totals derived from the specific differential phase measurements (KDP) from S-Pol show that more than 200 mm of rain fell (see figure below). A field program was conducted in central Florida during the summer of 1998 (PRECIP98) in collaboration with the National Aeronautical and Space Administration's Tropical Rainfall Measurement Mission. This important dataset from a semi-tropical environment includes hydrometeor-type information from a polarimetric radar, raindrop distrometers, and storm penetrating aircraft.
Early studies have compared rainfall estimates made with radar reflectivity measurements from S-Pol and the WSR-88D's. High correlation coefficients (rho) between radar estimates of rainfall (R) and gage observations (G) (see Table 1) attest to the radar capability to depict the rainfall distribution. This capability is offset by large storm-to-storm bias errors indicated by the sum of the gage observations divided by the sum of the radar estimates at these gages (G/R).
Importantly, radar reflectivity measurements from collocated radars produce similar rainfall patterns and have bias errors that are closely related. The biases are thought to be largely meteorological in origin-most likely due to fluctuations in drop-size distributions. Current research seeks to determine the bias source by making detailed comparisons between the full suite of polarimetric radar measurements, rain rate histories at gage sites, dropsize information, and in situ observations of hydrometeor types. Another important research task is the documentation of benefits that might be gained if the WSR-88D's are modified for polarimetric measurements. Research shows that rainfall estimates derived from the specific differential phase (KDP) for well-calibrated radars are comparable to that for radar reflectivity. Advantages of the propagation differential phase measurement is a reduced sensitivity to beam blockage and to anomalous propagation. Further, the rainfall estimates derived from KDP can be compared to those made with radar reflectivity to determine system calibration errors. This research is supported by a USWRP grant from the National Science Foundation and by funds from the National Weather Service Office of Systems Development.
Comparison of rainfall estimates made with radar reflectivity (R) at horizontal polarization with raingage observations (G). Gage observations are in mm; a hail threshold of 51 (53) dBZ was used for 1996 (1997).
| S-Pol | KFTG | ||||||
|---|---|---|---|---|---|---|---|
| date | No. gages | (G) | Gmax | G/R | rho (G:R) | G/R | rho (G:R) |
| 15 June 1996 | 49 | 9.46 | 21.08 | 1.88 | 0.86 | 1.49 | 0.87 |
| 9-10 July 1996 | 45 | 13.39 | 44.96 | 1.67 | 0.90 | 1.42 | 0.87 |
| 13 July 1996 | 38 | 7.95 | 39.88 | 0.78 | 0.78 | 0.56 | 0.77 |
| -- | -- | -- | -- | 1.38a | 0.70a | 1.07a | 0.60a |
| S-Pol | KICT | ||||||
|---|---|---|---|---|---|---|---|
| date | No. gages | (G) | Gmax | G/R | rho (G:R) | G/R | rho (G:R) |
| 2 May 1997 | 36 | 16.24 | 22.6 | 1.00 | 0.89 | 0.92 | 0.88 |
| 7-8 May 1997 | 43 | 17.78 | 35.0 | 0.86 | 0.86 | ND | ND |
| 18-19 May 1997 | 56 | 39.14 | 80.5 | 1.02 | 0.86 | ND | ND |
| 25-26 May 1997 | 46 | 19.10 | 50.0 | 0.88 | 0.84 | 0.87 | 0.91 |
| 26-27 May 1997 | 34 | 5.74 | 25.4 | 1.41 | 0.82 | 1.24 | 0.87 |
| 29-30 May 1997 | 51 | 42.05 | 99.1 | 1.38 | 0.90 | 1.13 | 0.93 |
| 4 June 1997 | 50 | 22.25 | 43.9 | ND | ND | 1.71 | 0.95 |
| 12 June 1997 | 47 | 16.98 | 29.3 | 0.93 | 0.91 | 0.82 | 0.88 |
| 13 June 1997 | 50 | 42.11 | 133.3 | 1.00 | 0.92 | 0.99 | 0.92 |
| 16 June 1997 | 50 | 11.09 | 29.2 | 1.13 | 0.95 | 1.03 | 0.93 |
| 16-17 June 1997 | 13 | 13.54 | 49.5 | 1.14 | 0.95 | 1.19 | 0.91 |
| -- | -- | -- | -- | 1.06b | 0.90b | 1.05c | 0.92c |