Salinity of Rivers: Transfer Function‐Noise Approach

Abstract

The theory of transfer function‐noise (TFN) models is used in this work to study the flow‐salinity relationships at four gaging stations located on the Ebro river basin, in Spain. This technique, which relates an input to an output series by considering the dynamic relations between them, is used to generate salinity data series from flow series that preserve not only the salinity‐flow relationship by design, but also the overall statistical characterisitcs of the salinity series. TFN models represent an improvement over time‐series univariate models in terms of a higher explained variance, and they present the advantage of having a related flow series. They also represent great improvement over classical regression models for cases in which the relation between salinity and flow is not instantaneous.