Partitioning evapotranspiration in semiarid grassland and shrubland ecosystems using time series of soil surface temperature

Abstract

Information about the ratio of transpiration ( T) to total evapotranspiration ( T/ET) is related to critical global change concerns, including shrub encroachment and non-native species invasion. In this study, a new approach was developed to partition measurements of ET into daily evaporation ( E D) and daily transpiration ( T D) in a semiarid watershed based on the low-cost addition of an infrared thermometer and soil moisture sensors to existing eddy covariance and Bowen ratio systems. The difference between the mid-afternoon and pre-dawn soil surface temperature (Δ t) was used to identify days when E D approached a seasonal minimum ( E Dmin) and thus, T D ≈ ET D − E Dmin. For other days, an empirical approach was used to partition ET D into E D and T D based on volumetric soil moisture. The method was tested using Bowen ratio estimates of ET and continuous measurements of surface temperature with an infrared thermometer (IRT) at a grassland and shrubland site within the Walnut Gulch Experimental Watershed in southeast Arizona USA in years 2004–2006. Validation was based on a second dataset of Bowen ratio, IRT and shrub sap-flow measurements in 2003. Results showed that reasonable estimates of T D were obtained for a multi-year period with ease of operation and minimal cost. Estimates of T D and E D were summed over the study period when plants were actively transpiring for years 2004, 2005 and 2006 to estimate totals over the study period, T S and E S, respectively. Preliminary analysis suggests that the accuracy of T S estimates was 7% of the total measured sum and the precision of T S estimates was about 4%. For this study period, T S was related strongly to ET S, with a slope of 0.79 for the grass-dominated site and 0.64 for the shrub-dominated site for the 3 years. Thus, for these sites during the study period in these years, the T S/ET S was higher for the grass-dominated site than for the shrub-dominated site, and did not vary systematically with variation in amounts and timing of precipitation. The Δ t-based partitioning method has potential for international application in other well-instrumented ecosystems but will need to be tested for application when evaporation is limited by energy rather than water.