Laserfiche WebLink
<br />exceeded this value. The suggestion of table 10 is that the a coefficient for the Denver area <br />should be about half that of the Cleveland value of318, but the exponents should be similar, <br />near 1.5. Of course, this preliminary conclusion is based on limited data without heavy <br />snowfalls. <br /> <br />, <br />The optimization scheme software cannot presently handle input data from more than one <br />gage with its corresponding radar observations. However, it was decided to apply the scheme <br />to gages No.1 and 2 at 24- and 25-km ranges using their average a and p values from table <br />10; that is, 150 and 1.45, respectively. The resulting radar estimates and gage observations <br />were then grouped together to increase the population to 113 pairs. The resulting R value <br />was a respectable 0.76 and the regression line essentially matched the 1:1 line. The average <br />a = 150 and p = 1.45 were used with equation (1) to estimate snowfall with the vertical array <br />over gage No.3, which resulted in an average radar-estimated snowfall of 0.0211 inch h-\ <br />slightly above the value of table 10. The corresponding R value was 0.79. These results <br />indicate no significant decrease in radar estimation to at least a 49-km range; that is, no <br />range effect. <br /> <br />With the absence of any discernible range effect out to at least gage No.3, the average a and <br />p values for gages No.1 to 3 of table 10 were used in equation (9) to estimate snowfall <br />accumulations from the vertical arrays over gages No.1 to 3. <br /> <br />Z = 155 81.6 <br />e <br /> <br />(9) <br /> <br />The resulting 196 data pairs are plotted on figure 8 along with the regression line, which <br />almost matches the 1:1 line. The correlation coefficient is 0.76. The average radar-estimated <br />and gage-observed hourly snowfall proved to be identical at 0.0205 inch. These results <br />suggest that equation (9) is a good first approximation to use with winter storms in the <br />Denver area. But as with the similar Cleveland area figures, a marked tendency is seen on <br />figure 8 for the radar estimates to be too high for ,low snowfalls and too low for higher <br />observed accumulations. <br /> <br />Little can be said about gage No.4 observations with only 34 hours of detectable snowfall at <br />a low average value of 0.015 inch. This broad mountain valley site was chosen because it <br />was surrounded by mountains but had no ground clutter for several kilometers around it. <br />This location seemed to offer a good opportunity to evaluate mountain valley sites where <br />people tend to live. However, it has become apparent that the location is in a distinct "rain <br />(and snow) shadow" because of downslope motion from higher terrain. <br /> <br />Gage No.5 was deliberately chosen as a mountain site in the midst of a very cluttered area <br />according to the clutter bypass map used by the Denver WSR-88D all winter. Its data can <br />be used to evaluate whether the Denver radar might be used to estimate snowfall over the <br />Rocky Mountains. Examination of PPI radar displays from De~ver generally reveals a large <br />area with very little reflectivity from low tilts over the Rocky Mountains because of beam <br />blockage and clutter suppression. The altitude of gage No.5 is above the center of the 0.50 <br />beam (for standard refraction), the only reflectivity data so far examined, and little useful <br />information was expected from this lowest tilt beam. Future plans include examination of <br />radar returns from higher tilts. <br /> <br />33 <br />