WMOD00564 - Laserfiche WebLink

Home Browse Search

The SAA uses a range correction scheme to adjust for severe underestimation as a factor of range. A range correction factor (see Section 5) was det,ermined using the vertical gradient of Level II Ze and S as observed by the lowest 5 antenna tilts in a circlle of about 35 km radius from the Minneapolis radar. The correction factor is of the form of a multiplier factor, F = Cl + C2*R + C3*R2, for range R and empirical coefficients Cl, C2, and C3. The relation was found to work well (see Section 5.2) to about 200 km (the beam center height above the radar is 4.1 km all 200 km, 5.1 km at 230 km) in deep storms. Beyond that range and in shallower storms at closer range, the correction becomes inadequate as the radar beam climbs above the precipitation and the reflectivities become less than the minimum threshold, currently 4 dBZ, making correction impossible. The range correction used for Minneapolis appeared to be working well for all locations. A linear plot (sc~e Section 5.4) of the ratio between the radar estimate of precipitation and that measured by cooperative gages showed the expected scatter but no obvious range bias. Preliminary indications from this large data set are that the range correction coefficients are appropriate and not in need of adjustment. The 5-radar composite showed that Duluth was. seriously underestimating precipitation because of likely calibration errors. Also, Bismark may have had smaller calibration errors. The composite products showed no discontinuity along the lines of equal distance between the radars for Aberdeen, Grand Forks, and Minneapolis. During the winter of 1999-2000, the operational test will be continued and expanded westward to include five WSR-88Ds near the cities of Minot, North Dakota; Rapid City, South Dakota; Glasgow, Montana; Billings, Montana; and Great Falls, Montana. The following default adaptable parameters are being used for the northern plains states: minimum reflectivity = 4 dBZ, maximum reflectivity = 40 dBZ: alpha = 150, beta = 2.0, range correction factor for R> 35 km: F = 1.04607 - 0.OO29590*R + 0.0000506*R2 for range R. 3.2 An Arctic Airmass Snowfall The storm of 2-3 January 1999, at Minneapolis and Aberdeen, was during an arctic outbreak. The SAA operating on NIDS data greatly underestimated, by a factor of about 2 or 3, the snowfall in both liquid equivalent and depth. Our analysis contained these points: I. In a preliminary attempt to reduce the virga contamination, the minimum threshold had been increased to +10 dBZ. The weak reflectivities from the arctic outbreak were thereby excluded from the accumulations. Therefore, the minimum threshold has been changed back to +4 dBZ since mid-January 1999 to produce more accurate accumulations fiOr storms in arctic air masses. That will increase the virga problem, which is recognizable by a ring of supposed accumulations at far ranges while there is no accumulation near the radar. 2. The storm clouds were quite shallow, with a very steep vertical profile of reflectivity (VPR). Depending on the depth of the precipitating clouds, even the lowest tilt beam would begin to have beam filling problems fairly near the radar, and would overshoot the clouds at greater ranges. 3. Rawinsonde data showed that the region for the rapidly growing dendrites (where temperatures were -13 to -17 oC and relative humidities near 100 percent) was shallow and just above the surface. So the radar was mostly scanning above the rapid growth zone resulting in underestimation by the SAA (see point 4 below). This explains why the radar was seeing small dBZ values and yet it was sometimes snowing at moderate to heavy rates. 3