Laserfiche WebLink
<br />- 67 - <br /> <br />Suppose we wish to know the difference in sampling requirements <br />for warm and cold season precipitation for a network of this size. It is <br />a simple matter to substitute the appropriate intercept and regression <br />constant$- into equation (1) and assumed values for E, P and T and then <br />calculate G, the gage density. As an example, if we let E = 0.05 in <br />(1.30 mm), P = 1.00 in (25 mm) and T= 6 hr, then G is 100 mi 2/gage <br />(270 km2/gage) and 306 mi2/gage (826 km2/gage) for the warm and cold sea- <br />sons, respectively. This factor of three relaxation of the sampling re- <br />quirements for cold season precipitation relative to warm season precipi- <br />tation holds over the range of area-m'ean rainfalls. It is explained by <br />the nature of Illinois rainfall which is much more convective and more <br />difficult to measure in the warm season. <br /> <br />Errors as a Function of Storm Mean Rainfall, Gage Density, <br />Duration and Size of Sampling Area <br /> <br />Because Huff found no definite trend in the regression con- <br />stants with increasing sampling area, he was able to combine all the warm <br />season network data into a single expression having a form similar to <br />equation (1) <br /> <br />logE = - 1. 5069 + 0.65 10gP ~. 0.82 10gG - 0.22 10gT - 0.45 10gA (2) <br /> <br />where A is sampling area and the other terms are as in (1). From this <br />expression one can calculate the gage density required to maintain a con- <br />stant-sampling error over sampling areas of varying size. Assume that <br />E is to remain 0.05 in (1.3 mm), P = 1.0 in (25 mm), T = 6 hr, then the <br />calculated values of G for various values of A are as shown in Table 2. <br /> <br />Note that the samp11ng requ1remenLs are re.laxea ror .larger <br />sampling areas. It may not be valid to extrapolate equation (2) to areas <br />larger than 550 mi2 (1485 km2) because this is the largest area for which <br />there were raingage data. As He shall see later, the sampling require- <br />ments are relaxed further for even larger areas. <br /> <br />An illustration of the relative importance of storm mean rain- <br />fall, storm duration and raingage density in determining sampling error <br />is shown in Figure 1 for Illinois warm season precipitation over an area <br />of 400 mi2. Note that raingage density is the most important factor in <br />determining sampling errornthe lesser the density the greater the sam- <br />ling error. Sampling error also increases as storm duration decreases <br />and storm amount increases. Although sampling error increases with in- <br />creasing storm rainfall, it actually decreases in terms of percentage <br />error. <br /> <br />A similar sampling error presentation obtained by Woodley et al <br />(1975) for Florida convective showers is presented in Figure 2 with the <br />