Active Projects

PEST, automatic calibration software

Researchers:  Sang Min Kim, Brian L. Benham, Kevin M. Brannan, and Rebecca W. Zeckoski,  Biological Systems Engineering Department, Virginia Polytechnic Institute and State University, Virginia, and John Doherty, School of Engineering, University of Queensland, Brisbane, Queensland, Australia.

This research project explores the use of the automated Parameter Estimation software (PEST) as a possible alternative to manual calibration aided by HSPEXP. The Hydrological Simulation Program-Fortran (HSPF) is a comprehensive watershed model that is in wide use, but requires calibration.  Currently, HSPF is one of the primary models used to develop Total Maximum Daily Loads (TMDLs). The TMDL program, which is mandated by the Clean Water Act (33 U.S.C. §§ 1251-1387), is a watershed management process that integrates watershed planning with water quality assessment and protection. The US Environmental Protection Agency (USEPA) estimates that public and private costs associated with TMDL development over the next 15 years will be in excess of $1 billion dollars. Given this large investment of both public and private funds, research is needed into methods to improve the use and application of HSPF to ensure these funds are invested wisely and result in measurable water quality improvement.

Because HSPF requires calibration, improving the efficiency and accuracy of the calibration process has the potential to reduce TMDL development costs and increase the usefulness of the model. PEST has been widely used in the field of groundwater modeling, but there have been very few applications of PEST to surface water models. Recently, PEST has been applied to HSPF calibration, using daily flow, monthly flow, and exceedence time as sub-components of the objective function along with Nash’s coefficient to evaluate goodness-of-fit. This research builds on the work of Doherty and Johnson, 2003 (Methodologies for calibration and predictive analysis of a watershed model, J. Am. Water Resour. Assoc., 39, 251–265) by creating multiple objective functions based on the HSPEXP model performance criteria and then assessing the adequacy of PEST as an alternative to HSPEXP for calibrating HSPF.