WMOD00425 - Laserfiche WebLink

Home Browse Search

GCIP taskings. Much of the following was taken from Super and Holroyd (1997b). Enhance- ments to the SAA since its development and initial implementation will be elaborated in section 6. As stated in the Introduction, the SAA endeavored to provide quantitative precipitation estimation (QPE) for snow rather than rain. The role of the PPS is to estimate rain only. Radar remote sensing of frozen precipitation presents challenges different from sensing rain. Snow has more complicated shapes than rain as well as a different dielectric constant, resulting in lesser reflectivities. Furthermore, if the snow is melting, a spuriously strong reflectivity layer known as the "bright band" will result. Estimates are therefore valid for dry snow only and not for equivalent radar reflectivity (Ze) measurements contaminated by such bright band effects. The SAA provides estimates of both SWE and SD over I-hour (hr), 2-hr, 3-hr, 6-hr, and 24-hr periods. Despite the limitations, the relationship between equivalent radar reflectivity Ze and snowfall rate S may be approximated by the same power law used for rainfall (with S substituted for rainfall rate R), Z = aSP e where a and p are empirically determined coefficients. We calibrate to Ze rather than Z as in the rainfall equation, because of the non-Rayleigh scattering of incident radar energy by snow. The SAA related Ze to hourly S accumulations sampled by Universal gauges and snow boards within -60 km of each radar. All matching of Ze and S used the single "nearest neighbor" range bin directly over the surface observing site. The algorithm uses equation (1) to convert Ze values (in decibels) to S for each range bin location and volume scan, to avoid a bias caused by averaging Ze over time. We determine the power (P) and coefficient (a) by minimizing a criterion function (absolute difference between gauge measurements and radar estimates) as pis incremented from 1.0 to 3.0. It is difficult to obtain accurate hourly S accumulations for reasons discussed by Super <md Holroyd (1996). These reasons are related primarily to serious wind-caused undercatch by most existing recording gauges. Accordingly, special observations were made during the 1995-96 and 1996-97 winters in sheltered locations. These locations were geographically diverse, including Albany, New York; Cleveland, Ohio; Minneapolis, Minnesota; and Denver and Grand Junction, Colorado. An optimization scheme, based on the work of Smith et al. (1975) and described in Super and Holroyd (1996), was used to determine appropriate Ze-S relationships. The optimization scheme yielded p values near 2.0 for all three sites, essentially the same as the theoretical value derived by Matrosov (1992) for S-band radar, and similar to the 2.21 value from re-analysis of snow particle-size distribution data by Sekhon and Srivastava (1970). If the scheme failed to optimize, an alternative iterative process was used instead. Final results were reported (Super and Holroyd, 1998) as follows: With Ze in millimeters6 (mm) m-3 and S in mm hr-I, values of a were 120 for Albany, 130 for Denver, 260 for Cleveland, and 180 for Minneapolis. An a value of 150 was 10 (1) I I I I I I I I I I I I I I I I I I I