Partitioning Evapotranspiration over a Vineyard in California's Central Valley

The increasing demand for limited water resources due to the ongoing California drought hampers crop production and damages the state's economy. In order to ameliorate the negative consequences of drought and ensure the sustainability of California agriculture, policymakers, resource managers, and agricultural producers must maximize the effective use of the available water. In turn, achieving this goal is predicated on accurate information regarding crop water productivity, the fraction of the total evapotranspiration (ET) that contributes to crop yield expressed in terms of transpiration. However, while a number of approaches, such as isotope analysis and microlysimeter systems, have been developed to partition ET between soil evaporation (E) and transpiration (T), these approaches can be both costly and labor-intensive. Collecting reliable continuous measurements at field scales remains problematic. This study presents the application of a recently developed correlation-based technique that overcomes these difficulties by leveraging high frequency data measured via eddy covariance. Specifically, this scheme combines wavelet decomposition and the theoretical relationship between stomatal and non-stomatal moisture and carbon fluxes to separate E and T. The technique was evaluated over a drip-irrigated vineyard located in California's Central Valley using data collected during the 2015 growing season as a part of the GRAPEX (Grape Remote sensing and Atmospheric Profile Experiment) field campaign. The results indicate a clear diurnal pattern in the fraction of ET due to T with a mid-day peak averaging 80% during the growing season. Similarly, there is a strong seasonal trend with the fraction of ET due T increasing in proportion to the increasing vine biomass during the growing season; at its maximum T accounts for approximately 90% of the total moisture flux. These results are in agreement with those from microlysimeter and sapflow measurements collected at the site. Overall, the results reaffirm the utility of the correlation-based approach.