Evapotranspiration over a rainfed maize field in northeast China: how are relationships between the environment and terrestrial evapotranspiration mediated by leaf area?

As a central process in the hydrological system and the climate system, terrestrial evapotranspiration is a key factor furthering our understanding of the climate change processes. Knowledge of factors controlling the variability in evapotranspiration is crucial for the prediction of the fate of terrestrial ecosystems under environmental changes. Based on long-term (2005-2014) eddy covariance flux data observed at a rainfed maize site in northeast China, the purpose of this study was to clarify the environmental regulation of actual evapotranspiration (ET) and the extent to which the regulatory effects on ET are directly or indirectly mediated by changes in biotic factors, using the structural equation modeling (SEM) method. The results showed that annual total ET was 397 ± 35 mm for the rainfed maize site in comparison with 575 ± 169 mm of precipitation (Prec), with an ET/Prec ratio ranging from 0.43 (2012) to 1.14 (2014). It was revealed that net radiation (Rn) was the primary controlling factor of the maize ET, followed by leaf area index (LAI), vapor pressure deficit (VPD), air temperature (Ta), and soil water content (SWC). The adjusted SEM models explained 71%, 67%, and 67% of the variation in daily ET of the maize growing season (ET_gs) for dry, normal, and moist years, respectively. Rn and VPD dominated ET_gs in an increasing order of dry, normal, and moist years. Conversely, the effects of LAI and Ta on ET_gs followed the opposite trend. This indicated that drought may increase the sensitivity of maize ET to temperature changes, and decrease the sensitivity of maize ET to radiation changes. In SEM analysis, LAI played an important mediating role in the relationship among climate, soil variables, and ET_gs. Rn, VPD, Ta, and SWC all had significant indirect effects on ET_gs mediated through LAI. At the annual scale, it was identified that most active days could be a robust predictor of annual ET.