Estimating Fluvial Mass Transport to Lakes and Reservoirs: Avoiding Spurious Self-correlations
Title:
Estimating Fluvial Mass Transport to Lakes and Reservoirs: Avoiding Spurious Self-correlations
Author:
Galat, David L.
Appeared in:
Lake and reservoir management
Paging:
Volume 6 (1990) nr. 2 pages 153-163
Year:
1990-10-01
Contents:
Three approaches to estimating fluvial mass transport to receiving waters are reviewed: concentration-discharge regressions, load-discharge regressions, and concentration discharge product methods. Concentration-discharge regressions are applied when constituent concentration (C) depends on river discharge. Several dilution, concentration and mixed-effects linear regression models are illustrated. The load (L)-discharge (Q) regression form: log L = a + b (log Q), where L = CQ, is widely used to estimate mass transport because it yields high coefficients of determination (R2) when concentration-discharge regressions do not. This is illustrated for Truckee River, Nevada, using eight years of monthly total nitrogen concentration ([TN]) and discharge data. However, the observed high R2 (0.92) results from product spurious self correlation, where the discharge term appears on both sides of the regression equation, and is not due to a dependency of [TN] on discharge. When concentrations do not depend on discharge, constituent loadings should be estimated as the product of concentration and discharge. Several concentration-discharge product methods are given for fixed-time-interval or flow-interval sampling of concentration.
Journal description