Laserfiche WebLink
<br />elevations. Methods were developed to correct the recording <br />gage data to the non-recording gage basis, and to fill in miss- <br />ing recording gage data by using multiple regressions on <br />neighboring stations. In addition, a procedure was developed <br />for deriving improved estimates of areal mean precipitation <br />over terrain of high relief from point measurements at a modest <br />number of gages. <br />The development, testing and application of these data <br />interpretation methods required considerable effort, and conse- <br />quently the statistical analysis of the project results so far <br />carried out is of a preliminary nature only; in particular, it <br />is based solely on a comparison of periods randomly assigned to <br />be seeded ("seed" periods) with periods randomly assigned not <br />to be seeded ("no seed" periods), rather than on a comparison <br />of periods which were actually seeded with periods which were <br />not seeded. Because unfavorable meteorological conditions for <br />seeding sometimes occurred during "seed" periods, and because of <br />the need to observe safety restrictions, seeding actually <br />occurred for only about half of the total hours assigned to be <br />seeded. Thus, the effect of seeding can appear in this prelim- <br />inary analysis only in a somewhat diluted form. <br />In the preliminary statistical analysis comparing the <br />precipitation received in the analysis area during "seed" and <br />"no seed II periods, the varying duration of these periods \'Jas <br />taken into account by "assuming a Gamma distribution of precipi- <br />tation amounts, in which. the shape parameter was assumed to be a <br />linear function of duration. The effect of seeding was reflected <br />by a change in the scale parameter, corresponding to a multipli- <br />cative effect. The estimates of areal precipitation were made <br />in a number of different ways, all of which gave essentially the <br />same result. <br />By' ~eans of a forecast, storms were classified as heavy, <br />moderate, light and marginal (labeled A, 3, C and D respectively). <br />Statistical significance was not achieved, and the evicence for <br /> <br />A-2 <br />