Laserfiche WebLink
<br />--~-,------- <br /> <br />,I ... <br /> <br />ABSTRACT <br /> <br />A study has been conducted which examines the sampling variance of raingage <br />networks, the most commonly used precipitation estimating system. The study <br />is based on the use of computl~r-s imul ated i sohyetal patterns of known character- <br />istics (absolute ground truth) which were calibrated against measured isohyetal <br />patterns from convective storms that are reported in the literature. The <br />sampling variance of raingage networks is quantitatively related to both the <br />raingage dens ity and the chard.cteri st ics of the i sohyetal pattern. It was <br />found that the sampling variance or coefficient of variation is "inversely <br />proportional to the number of gages per surface raincell and directly propor- <br />tional to the gradient of rainfall. <br /> <br />These results have been used to assess the relative contribution of the <br /> <br />sampl i ng vari ance of ra ingage networks and the natural vari abil ity of rainfall <br />to the estimated experimental unit sample size requirements for E~valuating <br />precipitation augmentation experiments. The convective storm ra'infall data <br /> <br />for the Miles City, Montana area, gathered as part of the Water and Power <br /> <br />Resources Service HIPLEX (High Plains Cooperative Program), formed the basis <br /> <br />of this analysis. For these rainfall characteristics, it was found that <br />sampl i ng vari ance is respons ible for no more than 10 percent of the total <br /> <br />sample size requirement with a gage density of at least an avera~le of four <br />gages per storm (gage density on the order of 80 km2 per gage). The <br /> <br />contribution of network sampling variance to the sample size requirement <br /> <br />. <br />becomes significant for gage densities less than one gage per storm or for <br />s ignificantl y lower natural rclinfall vari abil it ies. <br /> <br />. <br />