Laserfiche WebLink
<br />Using the single bin directly above each gage may provide essentially the same results as <br />averaging over approximately 3- by 3-km range bin arrays. Furthermore, use of a single <br />range bin will not eliminate the problem of radar overestimation of small snowfall amounts <br />and underestimation of larger amounts. If that estimation problem is related to spatial <br />variations, the variations are on space scales smaller than a single range bin. <br /> <br /> <br />Future analysis will give further consideration to the radar overestimation/underestimation <br />problem. Perhaps a more complex relationship than equation (1) would be useful, or some <br />improvement to the optimization scheme of section 8 may be in order. <br /> <br />9.2 Denver Observations <br /> <br />The optimization scheme was applied to 3 months of observations by the Denver W8R-88D <br />and gages No.1 to 5 of table 3 in the same manner as done for the Cleveland area <br />measurements. (Gage No.6 provided no useful data during this period.) As previously noted, <br />the Denver area experienced a dry winter, and no major storms occurred during the period <br />being considered. <br /> <br />Table 10. - Summary of results of applying the optimization scheme, to five gages in the Denver area. <br />Only hours with gage amounts of 0.005 or more inches are included. <br /> <br /> Distance Beam Standard Average <br />Gage from Radar Height* Hours of Error S <br />No. (km) (m) Observation a ~ R Estimate (in) (in h'l) <br />1 25 205 65 152 1.20 0.82 0.0119 0.0182 <br />2 24 320 48 148 1.70 0.71 0.0170 0.0238 <br />3 49 575 83 164 1.90 0.78 0.0120 0.0205 <br />4 81 700 34 134 2.50 0.63 0.0113 0.0150 <br />5 91 -275 108 1 2.20 0.89 0.0133 0.0251 <br /> <br /> <br />* Height of 0.50 tilt beam center a,bove gage assuming standard refraction. <br /> <br />Table 10 presents the results of applying the optimization scheme plus the correlation <br />coefficients resulting from using the noted a and ~ values with equation (1) to calculate <br />radar-estimated snowfall accumulations. The a values for gages No.1 to 3 are tightly <br />grouped between 148 and 164. The similar value for gage No.4 may be a fluke, being based <br />on only 34 pairs. The absurdly Iowa value for gage No.5 is discussed below. <br /> <br />The relatively high R values of table 10 are similar to those within 115 km of the Cleveland <br />radar (table 8) with the possible exception of gage No.4. These R values again suggest good <br />gage accuracy because more variability would otherwise be expected. In the absence offorest <br />cover, gages No.1 through 3 were located in backyards of established neighborhoods with <br />wind protection primarily provided by wooden fences and houses. This approach apparently <br />produced decent snowfall measurements. <br /> <br /> <br />It is encouraging that similar a values resulted from three independent runs of the <br />optimization scheme using the 3 gages within 49 km"ofthe radar. This result is in spite of <br />the limited hours with snowfall detected by each gage. Although the ~ values vary from 1.2 <br />to 1.9, this range of exponent has limited influence for S below 0.10 inch h-1 as shown on <br />figure 1. Only 1 percent of the combined 196 hours with snowfall at gages No.1 through 3 <br /> <br />32 <br />