Regression analysis of river profiles.

Abstract

The least-squares regression comparison of linear, exponential, logarithmic, and power functions for 13 diverse profiles demonstrated that: 1) a standard method of plotting profiles from source to mouth should be followed; 2) the exponential function is preferred for profiles ending downstream in aggradational areas; 3) the logarithmic function is preferred for profiles that terminate upstream at scarps or steep divides; 4) the power function is inadequate for most longitudinal profiles; 5) a composite function is recommended for complete profiles; and 6) a power regression of slope against distance provides the most useful method of describing and comparing curvature and rugosity, two basic properties of profiles. -Author