Evaluating Differences between Measured and Modeled Potential Evapotranspiration in the State of Colorado

Potential evapotranspiration (PET) is a critical variable to measure and constrain accurately in order to model the hydrologic cycle, monitor drought, and track trends in aridity in a changing climate. In Colorado, potential evapotranspiration is monitored using both weather station measurements, and gridded reanalysis datasets. In this study, PET measurements and anomalies from two unlinked data sources are compared for recent years. Measured PET data from 18 Colorado Agricultural Meteorological Network (CoAgMET) weather stations are compared and contrasted with overlaying gridded data from the North American Land Data Assimilation System (NLDAS) Phase II. Of the 18 CoAgMET sites used, six are on fully irrigated lands, six are on partially irrigated lands, and six are on non-irrigated lands. NLDAS data used in this study will not include an irrigation parameterization. Findings from this study explore the role of irrigation, and the role of complex topography in generating differences between measured and modeled seasonal evaporative demand measurements and anomalies. Specifically, this study seeks to answer the following questions about measured vs modeled PET across the state of Colorado:

1. How different are warm season PET accumulations for CoAgMet stations and overlaying NLDAS grid cells?
2. Does NLDAS gridded data more accurately depict warm season evaporative demand over unirrigated landscapes than irrigated landscapes?
3. How large a role does complex topography play in creating apparent discrepancies between modeled and measured PET?
4. How do standardized seasonal PET anomalies differ between CoAgMET and NLDAS? Are there certain types of weather that cause large discrepancies in departure from normal seasonal PET between the two data sources?