Implementation of a probabilistic curve number method in the PRZM runoff model

Runoff from agricultural fields, which is a major factor influencing ecological and human exposure assessments, is often estimated by models that use the curve number method. One such model, PRZM, adjusts curve number on a daily basis according to calculated soil moisture in a manner that does not usually capture the full range of variability in rainfall–runoff relationships. Under the assumption that soil moisture alone cannot account for the large observed variability in event-to-event curve number, we modified PRZM by decoupling the curve number from any soil moisture relationship. Instead, we varied the daily curve number by randomly selecting the day's initial abstraction from a distribution. Using field data, we showed that the modified PRZM better characterized the variability in the rainfall–runoff relationship than the method based on soil moisture.