Estimating changes in river faecal coliform loading using nonparametric multiplicative regression.

Faecal coliform (FC) concentration was monitored weekly in the Tangipahoa River over an eight year period. Available USGS discharge and precipitation data were used to construct a nonparametric multiplicative regression (NPMR) model for both forecasting and backcasting of FC density. NPMR backcasting and forecasting of FC allowed for estimation of concentration for any flow regime. During this study a remediation effort was undertaken to improve disinfection systems of contributing municipal waste water treatment plants in the watershed. Time-series analysis of FC concentrations demonstrated a drop in FC levels coinciding with remediation efforts. The NPMR model suggested the reduction in FC levels was not due to climate variance (i.e. discharge and precipitation changes) alone. Use of the NPMR method circumvented the need for construction of a more complex physical watershed model to estimate FC loading in the river. This method can be used to detect and estimate new discharge impacts, or forecast daily FC estimates.