Is ET often oversimplified in hydrologic models? Using long records to elucidate unaccounted for controls on ET

Christa Kelleher, Stephen B. Shaw

Research output: Contribution to journalArticle

3 Scopus citations


Recent research has found that hydrologic modeling over decadal time periods often requires time variant model parameters. Most prior work has focused on assessing time variance in model parameters conceptualizing watershed features and functions. In this paper, we assess whether adding a time variant scalar to potential evapotranspiration (PET) can be used in place of time variant parameters. Using the HBV hydrologic model and four different simple but common PET methods (Hamon, Priestly-Taylor, Oudin, and Hargreaves), we simulated 60+ years of daily discharge on four rivers in New York state. Allowing all ten model parameters to vary in time achieved good model fits in terms of daily NSE and long-term water balance. However, allowing single model parameters to vary in time – including a scalar on PET – achieved nearly equivalent model fits across PET methods. Overall, varying a PET scalar in time is likely more physically consistent with known biophysical controls on PET as compared to varying parameters conceptualizing innate watershed properties related to soil properties such as wilting point and field capacity. This work suggests that the seeming need for time variance in innate watershed parameters may be due to overly simple evapotranspiration formulations that do not account for all factors controlling evapotranspiration over long time periods.

Original languageEnglish (US)
Pages (from-to)160-172
Number of pages13
JournalJournal of Hydrology
StatePublished - Feb 1 2018



  • Evapotranspiration
  • Potential evapotranspiration
  • Streamflow
  • Watershed modeling

ASJC Scopus subject areas

  • Water Science and Technology

Cite this