Authors: Zhiying Li*, Ohio State University, Steven M. Quiring, The Ohio State University
Topics: Climatology and Meteorology, Global Change
Keywords: Potential Evapotranspiration, Temperature, Trends, Climate Change
Session Type: Virtual Paper
Start / End Time: 9:35 AM / 10:50 AM
Room: Virtual 40
Presentation File: No File Uploaded
Potential evapotranspiration (PET) is an important input variable to many hydro-climatological applications such as drought monitoring, hydrological modeling, and climate change detection. Estimation methods of the PET and the input datasets such as gridded temperature are two important sources of uncertainties for the applications. Despite excellent previous work on local and global comparisons of PET estimations, relatively few studies have compared the uncertainties sourced from the input datasets and their relevance to the trend of PET with a warming climate over a long period at a large spatial scale. This study compares three PET estimations including Hargreaves and Samani (HS), Thornthwaite (TH), and Penman-Monteith (PM) methods to detect the trend in PET over six decades in 1950-2009 in a total of 1079 watersheds across the continental U.S. Two daily gridded temperature datasets, Livneh and TopoWx, are used to calculate HS-based PET. Results show that the Livneh dataset has a cold bias, especially in the minimum temperature in mountainous regions before 1970. At the mean annual timescale, the cold bias leads to a higher value of HS-based PET values than other PET datasets. Temporally, the large cold bias in early periods leads to statistically significant decreasing trends in diurnal temperature range and further in HS-based PET estimation. The HS-based PET using the TopoWx dataset as input has the smallest biases compared with PM-based PET. This study demonstrates that the TopoWx dataset is better suited for trend tests of PET.