Abstract
The conjunctive use of nonlinear integer programming and principal component analysis is suggested for specification of a constrained optimal gaging program whose objective is to define efficient estimates of the several annual means. The assumptions in the analysis are restrictive; data from thensites are presumed to be available forn1years, andn2years of additional gaging are contemplated. The observations are presumed to derive from a multivariate normal population with no serial correlation. The method specifies which sites from among thencandidates should be continued and which should be estimated by linear regression. Although the discussion is directed toward hydrologie applications, it should be recognized that the technique is generally applicable to sampling surveys.

This publication has 3 references indexed in Scilit: