Laserfiche WebLink
<br />zero and they were used in calculating average values for the vertical and upwind arrays. <br />The radar did scan the volumes of space in question. With the sensitivity of the WSR-88D, <br />non-detection of any return is equivalent to trivial snowfall rates within a 100-km range for <br />any reasonable combination of a and ~ values in equation (1). Although the current software <br />cannot handle groups of gages, a future version will be modified to do so. For example, gages <br />at similar ranges could be grouped <br /> <br />9. Application of Optimization Scheme to Cleveland and Denver Data <br /> <br />The optimization scheme just discussed was applied to the Cleveland and Denver data sets <br />presently available. As noted in section 5, the data sets consist of 2 wet months from the <br />Cleveland area and 3 dry months from the Denver area. <br /> <br />9.1 Cleveland Observations <br /> <br />0.18 <br /> <br /> <br />The Belfort gage nearest the Cleveland radar was located at the 36-km range and the 640 <br />azimuth (table 5), just beyond the ground clutter region according to the clutter bypass map <br />used all winter. A total of 143 hours with detectable hourly snowfall (at least 0.005 inch) was <br />available with simultaneous radar observations. The optimization scheme produced its <br />lowest CTF with an a value of 318 and a ~ value of 1.5. These values are similar to the <br />default a = 300 and ~ = 1.4 used in the WSR-88D rainfall algorithm. <br /> <br />A plot of the resulting hourly data is given on figure 3 with a 1:1 (solid) line and linear least <br />squares regression (dashed) line. The two lines almost overlap, indicating the optimization <br />scheme produces radar-estimated hourly snowfall accumulations that are well matched with <br />hourly gage accumulations. Inspection of figure 3 shows the high frequency of hours with <br />only 0.005 or 0.010 inch of snowfall as previously discussed. Radar-estimated snowfall <br />accumulations ranged from just above zero to about 0.03 inch h-1 with gage observations of <br />0.005 or 0.010 inches. This result suggests the radar would not seriously overpredict in the <br />Cleveland area during ,the many hours with very light snowfall using the relationship: <br /> <br />0.00 <br />0.00 0.03 0.06 0.09 0.12 0.15 0.18 <br />Radar-est. hourly amounts (in) <br /> <br />........ <br />.f: 0.15 <br />'-' <br /> <br />.2 <br />c: <br />g 0.12 <br />E <br />o <br />>-0.09 <br />'!: <br />:J <br />o <br />.s::. <br />0.06 <br /> <br />CD <br />Dl <br />~ 0.03 . <br /> <br /> <br />Ze = 318 s'" <br />N = 143 <br />R = 0.73 <br />Range = 36 km <br /> <br />Figure 3. - Scatter plot of 143 pairs of hourly-observed snowfall for Cleveland gage No.1 versus radar-estimated <br />snowfall directly overhead using the relation Ze = 3188'.5. The regression line is dashed, and the 1:1 line is solid. <br /> <br />25 <br />