Laserfiche WebLink
<br />radar, visibility below cloud base, and <br />a need to develop a statistical base by <br />sampling clouds rather uniformly <br />distributed along the size range from <br />cumulus congestus to moderate <br />thundershowers. <br /> <br />Numerous studies have shown that <br />small convective clouds greatly <br />outnumber large ones. For example, <br />Dennis and Fernald (1963) found that the <br />radii of radar echoes from showers in <br />many parts of the world follow an <br />exponential distribution. Miller et al. <br />(1975) fitted exponential distributions <br />to the diameters of echoes in the North <br />Dakota pilot Project radar data from <br />1972 to determine the parameter a in <br /> <br />N = No exp(-aD) <br /> <br />(3) <br /> <br />where N is the number of shower echoes <br />with diameters between D and (D+dD), No <br />is a constant reflecting the number of <br />showers in the space-time domain <br />studied, and the parameter a <br />characterizes the size distribution. <br />They found a for the entire data set to <br />be about 0.35 km1, in good agreement <br />with Dennis and Fernald (1963). As the <br />height of a convective cloud tends to <br />approximate its diameter (e.g., Miller <br />et al., 1975), o/e assume that the <br />exponential relationship with a similar <br />value for a holds for CDP. <br /> <br />Equation (3) was derived from <br />hourly snapshot views. As there is a <br />positive correlation between echo size <br />and echo lifetime, a large shower has a <br />greater chance of showing up in such a <br />census than does a smaller one. Miller <br />et a1. (1975) studied a supplementary <br />sample of 479 echoes, from which echoes <br />that were obviously multi-cellular were <br />excluded; they found a roughly linear <br />relationship between echo diameter and <br />echo lifetime (I' = 0.65). Therefore, in <br />order to make table 1 reflect more <br />closely the size distribution of all <br />echoes that form, exist, and dissipate, <br />as opposed to the size distribution <br />observed at a given moment, we have <br />modified (3) as follows: <br /> <br />N' = No D-1 exp(-aD) <br /> <br />(4) <br /> <br />There is a further complication in that <br />the snapshot views must catch some <br />undefined fraction of the echoes in <br />their growing or dissipating stages, <br />thereby leading to underestimates of <br />their maximum size. No satisfactory <br />method of adjustment for this second- <br />order complication has been devised. <br />Therefore no allowance is made for it in <br />the calculations that follow. It is <br />considered a minor factor compared to <br /> <br />I ~ <br /> <br />the other um:ertainties in the analysis. <br />All results should be interpreted as <br />indicating only general trends, rather <br />than as exac1: calculations of the <br />possible rainfall increases. <br /> <br />Column 2 of table 1, which shows <br />the relative frequency of clouds of <br />different sizes, was obtainE!d by <br />applying (4) with the paramE!ter a set at <br />0.35 km1 and normalizing so that the <br />total number of clouds consi.dered came <br />out to 100. This is a trunc:ated <br />distribution. Ignoring clouds with CDP <br />exceeding 12.5 km is of litt:le <br />consequence as far as number of clouds <br />is concerned, although such clouds may <br />be significant in terms of t.otal <br />rainfall. On the other hand, there are <br />many clouds with CDP less than 2 km. <br />Dennis and Fernald (1963) quoted work by <br />earlier authors showing that the <br />exponential distribution they observed <br />for radar echoes from isolated showers <br />extends down-scale to the visual <br />dimensions of small cumulus clouds. <br />However, we can ignore such clouds in <br />the present case because they do not <br />produce any precipitation whether seeded <br />or unseeded. <br /> <br />The numbers in column 4 of table 1 <br />arl::! the products of the corresponding <br />numbers in columns 2 and 3. They show <br />the relative contributions of clouds of <br />different sizes to the total natural <br />convective rainfall over the Black <br />Hills, and peak for CDP around 10 km. <br />As the bases of summertime convective <br />clouds near the Black Hills are, on <br />average, very close to 3 km above sea <br />level, the 10 km value corresponds to <br />large storms with cloud tops about 13 km <br />above sea level. This number agrees <br />with the common perception that much <br />Black Hills slimmer rain falls from large <br />thunderstorms. <br /> <br />The entri.es in column 6 of table 1 <br />for all values of CDP up to 10 km were <br />obtained by multiplying the <br />corresponding numbers in columns 2 and <br />5, and represent the expected rainfall <br />from silver iodide seeded showers <br />according to (2) and (4). On the <br />assumption that seeding would not be <br />done on clouds more than 10 km deep or, <br />if it were done, would have no effect, <br />the entries of column 4 for such clouds <br />are repeated in column 6., <br /> <br />A comparison of columns 4 and 6 <br />ignores the possibility of increases in <br />CDP but suggests, nevertheless, that <br />silver iodide seeding can increase the <br />total rainfall from isolated summertime <br />convective clouds by a factor of roughly <br />1.7. It is instructive to consider <br /> <br />5 <br />