Sensitivity analysis of a phosphorus index for Québec

摘要

The presence of phosphorus (P) in surface water is a major cause for water quality degradation in agricultural areas. Available tools for estimating the relative P contribution from fields include P-index (PI) models. We performed sensitivity analyseson a PI model, adapted for the province of Québec, to identify the site characteristics and input variables impacting most on the calculated PI value. The probability density functions of the 17 input variables and ten site characteristics were determined on a 10 301 km~2 study area south of Québec City. Monte Carlo simulations were performed with the @RISK software and stepwise regressions were used to determine which variables and characteristics the PI was sensitive to. The PI was mainly affectedby the weights, determined by experts, related to each site characteristic. When ignoring these weights, 58.7% of the PI variation was caused by the organic P budget and 11.7% by the subsurface drain spacing, Results for the input variables showed that the crop type impacted the most PI by explaining 46.4% of the PI variation, followed by subsurface drain spacing (12.0%), and amount of organic P applied to the field (10.9%). This information can be used by field staff to obtain accurate PI values without wasting human and financial resources on input variables that contribute little to the final P index value.