Laserfiche WebLink
<br />AUGUST 1984 <br /> <br />. DONEAUD, IONESCU-NISCOV AND MILLER,JR. <br /> <br />1607 <br /> <br />promise for a most appropriate ATI reflectivity <br />threshold (MART) for rain rate computation (if the <br />small clusters are deleted) seems to be 25 dBz [see <br />also Eq. (3) for b < 1]. For such .a threshold, if the <br />clusters up to 10 km2 h are deleted, the rain rate can <br />be considered in a first approximation to be indepen- <br />dent of the ATI, and roughly 4.0mm h-I. In such a <br />case, the accuracy of the estimates, according to the <br />standard deviation of the mean value (-1.55 mm <br />h-I), will range between :t40%. This accuracy is <br />comparable with that obtained using traditional Z-R <br />techniques for rain volume computations. <br />The summer 1981 Bowman data (587 clusters) <br />were used to check the 1980 findings. Frequency <br />distributions of R were computed on a monthly basis <br />using the 25 dBz A TI threshold (Fig. 5). The distri- <br />butions are unimodal. A stronger right skewness <br />occurs in August, generated by a small number of <br />outliers. As visible in Table 1 (1981 data), if the <br />clusters smaller than 1 km2 h A TI are discarded (four <br />clusters), the standard deviation decreases to less than <br />half of its initial value, remaining less than one-half <br />of the mean. The R range decreases from 98.20 to <br />16.44 mm h-I. The frequency distributions (Fig. 5) <br />show only two outliers in August with R exceeding <br />13 mm h -I. These cells were very unusual, being tall <br /> <br />and narrow with tight reflectivity gradients toward <br />the edges, and generating high rain volume estimates <br />for rather small area coverages. The removal of those <br />two clusters would bring the R range to not more <br />than 2.5 times the mean (-13.00 mm h-I). After <br />discarding the very small clusters (A TI up to 1 km2 <br />h), R becomes -4.80 mm h-I (Table 1). A slight <br />increase is visible when clusters with A TI up to 25 <br />km2 h are discarded. Investigating the A TI versus <br />RERV power-law relationship, Doneaud et al. (1984) <br />found a slope slightly greater than one, indicating <br />that the rain volume increased faster than A TI. <br />As mentioned earlier, the 1980 season was dry and <br />R was lower than for the 1981 wet season, suggesting <br />that adjustments for different seasonal conditions <br />may be necessary. The samples analyzed show an <br />increase in R of -20% for the wet weather season. <br />An analysis of the predictive power of the A TI <br />technique to estimate rain volume was made by <br />Doneaud et al. (1984). In that paper, the formula <br />derived from the 1980 NDCMP data was applied to <br />the 1981 echo cluster A TI values to estimate the <br />corresponding rain volumes. These estimates were <br />then compared with the radar-estimated rain volumes <br />computed on the 1981 radar data. Because of the <br />lack of an extensive raingage network in the NOCMP, <br /> <br />NDCMP, SUMMER 1981 BOWMAN DATA. <br /> <br />JUNE <br />NUMBER OF CLUSTERS: 72 <br /> <br /> 60 <br /> (A) <br /> 40 <br /> 20 <br />~ 0 <br />!!- <br />>- (B) <br />0 <br />z <br />w 40 <br />::l <br />a <br />w <br />IIC <br />II. <br />W 20 <br />> <br />i= <br />cc <br />-' <br />w I> <br />IIC <br />.~ (C) <br /> 40 <br /> 20 <br /> <br /> <br /> <br />-, <br />I <br />I <br />, <br />,. <br />I <br />, <br />I r---i <br />\1 r- <br />I' I, <br />, " <br /> <br />MONTH <br />- JULY <br />h__ AUGUST <br /> <br />NUMBER OF CLUSTERS <br />391 <br /> <br />124 <br /> <br />TOTAL <br />NUMBER OF CLUST~RS: 587 <br /> <br />1 <br />, <br />, <br /> <br />o <br /> <br />5 10 15 <br />AVERAGE RAIN RATE (mm hr-') <br /> <br />FIG. 5. Relative frequency distributions of R for 25 dBz ATI thresholds <br />calculated on monthly and seasonal bases. <br /> <br />o <br /> <br /> <br />20 <br />