Laserfiche WebLink
<br />. <br /> <br />. <br /> <br />.. <br /> <br />, <br /> <br />. <br /> <br />REGIONAL ANALYSES OF STREAMFLOW CHARACTERISTICS <br /> <br />By H. C. Riggs <br /> <br />Abstract <br /> <br />This manual describes various ways of generalizing <br />str.eamtlow characteristics and evaluates the applica- <br />bility and reliability of each under various hydro- <br />lOgic conditions. Several alternatives to regionaliza- <br />tion are briefly described. <br /> <br />Introduction <br /> <br />Regional analysis is concerned with extend- <br />ing records in space as differentiated from <br />extending them in time. Because streamflow <br />records are collected at only a few of the many <br />sites where information is needed, gaging- <br />station information must be transferred to <br />ungaged sites. A regional analysis provides a <br />tool for doing this. In addition, a regional <br />analysis may produce improved estimates of <br />the flow characteristics at the gaged sites. <br />The specific purposes of a regional analysis, <br />then, are to provide estimates of the charac- <br />teristics of the frequency distributions at un- <br />gaged sites and to improve estimates of the <br />frequency distributions of flow characteris- <br />tics at gaged sites. Consider, for example, a <br />frequency curve of annual floods derived from <br />50 years of record. This frequency curve is an <br />estimate of the population frequency curve; <br />it will differ from the true curve, however, <br />because a 50-year sample of floods is never <br />completely representative. Frequency curves <br />for other streams would also differ from their <br />respective true curves. It these several curves <br />were based on samples from the same popu- <br />lation frequency curve and if the samples <br />were independent of each other, then we would <br />expect that an average of the several curves <br />would be a better estimate of the population <br />curve than anyone of the samples. This aver- <br /> <br />aging of curves can be accomplished by re- <br />gional analysis. <br />No group, or even pair, of stream sites <br />would have the same population frequency <br />distribution of floods. The true distribution <br />at a site depends on a great many factors, the <br />principal ones being basin characteristics <br />such as size, topography, surficial geology, <br />and climate. Thus the variability among a <br />group of flood frequency curves is made up of <br />two components: chance variation due to sam- <br />pling, and variation due to differences in basin <br />characteristics. A regionalization procedure <br />should average the chance variation but <br />should maintain the variation due to basin <br />characteristics. This is a difficult task because <br />the total variation cannot be neatly separated <br />into the two types of variation. The degree of <br />success attained by a given method of region- <br />alization depends on the relative sizes of the <br />variations due to chance and those due to dif- <br />ferences in basin characteristics, the degree <br />of independence of the samples at the various <br />gaging stations, the quality of the relation <br />with basin characteristics, and the general <br />suitability of the method. <br />Following sections describe and illustrate <br />some methods of regional analysis applicable <br />to various flow characteristics. In describing <br />these methods, it is assumed that the fre- <br />quency curves at gaging stations have been <br />prepared by one of the methods described by <br />Riggs (1968b) or the method recommended <br />by Water Resources Council (1967). Back- <br />ground material needed for understanding <br />some of the procedures described in this man- <br />ual is available in book 4, chapter Al of this <br />series (Riggs, I968a). <br /> <br />1 <br />