Sensitivity Analysis of the Ohio Phosphorus Risk Index

The Phosphorus (P) Index is a widely used tool for assessing the vulnerability of agricultural fields to P loss. This study is focused on the Ohio P Index, which was developed in the mid-1990s and has yet to be evaluated or revised. The objective of the study was to complete a stochastic sensitivity analysis of the Ohio P Index in order to determine the input variables to which the P Index score is most sensitive and identify variables for which future research and development are needed. Input variable probability distributions were created using the best available data from five agricultural watersheds in Ohio. Monte Carlo simulation was then used to generate 10,000 iterations of the P Index score based on the input variable probability distributions for each watershed. Results showed that three variables (connectivity to water, runoff class, soil-test P) explained 78% to 81% of the variance in the P Index score. Phosphorus application rate, P application method, soil erosion, and filter strip variables each explained <10% of the variability. Findings suggest that the structure of the Ohio P Index may not accurately account for the interrelationship between source and transport variables, and current input variable weightings may not provide any incentive for producers to modify management practices. Differences in input variable sensitivities among watersheds also suggest that a watershed P Index, rather than a statewide P Index, may yield better predictions of a field's risk of P loss by placing an emphasis on variables and practices that are relevant in individual watersheds. To increase the predictive capability of the Ohio P Index, it is recommended that (1) the structure of the P Index be changed from additive to multiplicative, (2) input variable weights be re-evaluated to ensure that implementation of management practices is accurately reflected in the P Index score, and (3) additional input variables, including subsurface drainage, be considered for inclusion in the P Index.