Empirical Model for Quantifying Total Phosphorus Reduction by Vegetative Filter Strips

Abstract

Abstract. Vegetative filter strips (VFS) are commonly used conservation measures to protect nearby receiving streams and water bodies from sediment, nutrients, pesticides, and other contaminants applied in upslope contributing areas. Their efficiency for trapping phosphorus (P) has been examined in numerous field studies, but limited research has been performed on developing predictive equations for use in their design and assessment of water quality benefits. Existing predictive equations for P trapping efficiency are based on physical characteristics of the filter or linear regressions with only sediment reduction by the filter. The objective of this research was to develop a simple empirical regression model to quantify total P trapping efficiency by a vegetative filter strip based on the hydrologic (infiltration) and sedimentological response of the filter from ten previous studies reported in the literature. A wide range of trapping efficiencies for total P was reported, with only a few observations of 100% trapping efficiency. Using a traditional multiple linear regression approach, a regression model was derived based on runoff reduction (infiltration) and sediment reduction with a coefficient of determination (R 2 ) of 0.68 and a standard error of 14.2. A linear regression between observed and predicted total P percent reduction resulted in a slope of 0.68 and intercept of 23.5%. The length of the vegetative filter strip was not a statistically important factor, but the runoff reduction (infiltration) was statistically significant in predicting total P reduction. Therefore, P trapping efficiency predictions are improved if infiltration and sedimentation are considered explicitly rather than implicitly via filter strip length. The regression model was derived independent of the initial P concentration attached to the soil, resulting in the model overpredicting P reduction for cases when the observed P reduction was less than 40%. Removing VFS observations of total P reduction in studies with initially high soil P concentrations in the VFS, where sorption may be limited, resulted in a regression model with R 2 = 0.92 and a standard error of 5.1. A linear regression between observed and predicted total P percent reduction resulted in a slope of 0.92 and intercept of 6.3%. Therefore, the proposed regression model can be used in the design or evaluation of early-stage VFS where the sorption of P is not inhibited by saturation of the soil media and/or sediment and its associated P has not accumulated in the VFS. Furthermore, the regression model can be integrated into a numerical filter strip model capable of predicting infiltration and sedimentation without the need for estimates of numerous contaminant transport parameters.