WMOD00562 - Laserfiche WebLink

Home Browse Search

Table 2. - R values between hourly SWE from gage measurements and radar estimates from array averages (column 6) and single range bins (column 7), and R values between array average and single bin SWE estimates (column 8). Gage versus Radar R Values Bins Array Average In Array Single versus Site Gage Range Array Hours Average Bins Single Bins CLE 1 36 15 143 0.73 0.68 to 0.73 0.95 to 0.99 CLE 2 61 9 235 0.80 0.76 to 0.81 0.95 to 0.99 CLE 3 87 9 187 0.68 0.64 to 0.67 0.95 to 0.99 DEN 1 25 21 65 0.80 0.75 to 0.81 0.95 to 0.99 DEN 2 24 27 48 0.71 0.60 to 0.73 0.95 to 0.99 DEN 3 49 15 83 0.79 0.75 to 0.78 0.95 to 0.99 The R values were examined for the central bin in each of the 6 arrays, located directly over each gage. These "nearest neighbor" bins had R values ranging between 0.989 and 0.993, so they explained between 98 and 99 percent of the array average variances. Based on this result alone, it can be argued that there is little point in using array averages instead of just the nearest neighbor range bin for snowfall estimation. Use of the nearest neighbor has the further advantage of minimizing the spatial averaging problem discussed in section 3.2. Further evidence indicating that the nearest neighbor range bin provides almost the same results as the 9-km2 or larger arrays is provided in table 3. The table shows that both radar-estimated average snowfall accumulations and the standard errors of estimate are quite similar for the arrays and nearest neighbor with the possible exception of Denver gage 3, which has the greatest difference between radar-estimated average snow accumulations. The individual bin values were plotted for this gage, and it was discovered that the four bins making up the northeast comer of the 3-km by 5-degree array had markedly greater average snowfall accumulations (average = 0.0276 inch) than the other 11 bins (average = 0.0182 inch). Examination of the terrain suggests the four greater-value bins may be subject to ground clutter contamination. But whatever the cause of the gradient, use of the nearest neighbor provided a better estimate directly over the gage in this case. Based on the evidence and discussion presented, it was decided to abandon use of arrays of range bins for estimating snowfall accumulations. All additional work reported herein simply used the nearest neighbor single range bin directly over each gage. Reasons for not using a wind advection scheme to select range bins to relate to gages are discussed in section 4.2. 4.2 Wind Advection Scheme A horizontal wind advection scheme was developed for use as part of the SAA as discussed by Super and Holroyd (1996). V AD (Velocity Azimuth Display) winds were calculated for each volume scan at a radius of 10 km from the radar for 305-m (lOOO-ft) intervals above the radar. The V AD winds and 4 different assumed fall speeds of 0.5, 1.0, 1.5 and 2.0 m S-1 were used to calculate the reverse trajectories of snowflakes from each range bin position on the ground. These fall speeds extend over the likely range for the larger snowflakes, which provide essentially all of the radar reflectivity return, from large individual flakes to graupel.. It is recognized that a fairly wide range of fall speeds may exist for larger 9