Development of a reference potential evapotranspiration climatological dataset for the Great Lakes region

English Abstract:

This study describes the development of a gridded historical climatological dataset of reference potential evapotranspiration (ET0) for the Great Lakes region using the Penman-Monteith methodology following Allen et al. 1998. The rate of ET0 is dependent on meteorological conditions including the intensity of solar radiation, air temperature, humidity and wind speed. Following international convention, we calculate ET0 assuming a flat, unshaded, 12 cm-tall grass-covered surface with no soil water limitations. Although ET0 is a primary input variable for most irrigation scheduling systems, little is known regarding its temporal and spatial variability.

Given the limitations of objectively analyzing single site station data (e.g., incomplete station records, heterogeneous station density), we used the North American Land Data Assimilation System (NLDAS) gridded analysis as a source of weather information for computing ET0. Before calculation, adjustments were made to the NLDAS solar radiation field to reduce bias and correct for a tendency of NLDAS to be too conservative in the tails of the distribution. NLDAS fields were then used as input to generate hourly ET0 estimates at each grid point in a domain covering the entire Great Lakes watershed over the 30-year period from 1983-2012.

Preliminary analysis of the NLDAS-derived dataset shows a pronounced influence of the lakes on ET0 spatial and seasonal patterns, distinct links with drought years in 1988 and 2012, and a general increasing trend of ET0 with time over the 1993-2012 period of record. Future efforts planned include coupling the ET0 dataset to a precipitation dataset, evaluating the sensitivity of ET0 to its individual components, developing future climate projections, and linkage to an operational irrigation scheduling decision support system.