|MILLER, DOUGLAS - Pennsylvania State University|
|KNIGHT, PAUL - Pennsylvania State University|
|DROHAN, PATRICK - Pennsylvania State University|
Submitted to: Journal of Soil and Water Conservation
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/19/2012
Publication Date: 5/1/2013
Citation: Buda, A.R., Kleinman, P.J., Feyereisen, G.W., Miller, D.A., Knight, P.G., Drohan, P.J., Bryant, R.B. 2013. Forecasting runoff from Pennsylvania landscapes. Journal of Soil and Water Conservation. 68(3): 183-196. DOI: 10.2489/jswc.68.3.185.
Interpretive Summary: Due to concerns over nutrient runoff and improved use of nutrients by crops, knowing when and where to apply fertilizers and manures is a priority of nutrient management. We developed simple models to predict the likelihood of runoff occurrence on agricultural soils using soil maps and weather forecast data. Validation of the models with monitoring data confirmed that they were as accurate as weather forecasts and have the potential to support nutrient management decisions by farmers requiring daily input on when and where to apply nutrients.
Technical Abstract: Identifying sites prone to surface runoff has been a cornerstone of conservation and nutrient management programs, relying upon site assessment tools that support strategic, as opposed to operational, decision making. We sought to develop simple, empirical models to represent two highly different mechanisms of surface runoff generation, saturation excess runoff and infiltration excess runoff, using variables available from short-term weather forecasts. Logistic regression models were developed from runoff monitoring studies in Pennsylvania, fitting saturation excess runoff potential to rainfall depth, rainfall intensity and soil moisture, and infiltration excess runoff potential to rainfall depth and intensity. Testing of the models in daily hindcasting mode over periods of time and at sites separate from where they were developed confirmed a high degree of skill, with Brier Skill Scores ranging from 0.61 to 0.65 and Gilbert Skill Scores ranging from 0.39 to 0.59. These skill scores are as good as models used in weather forecasting. Results point to the potential to forecast site-specific surface runoff potential for diverse soil conditions, with advances in weather forecasting likely to further improve the predictive ability of runoff models of this type.