Evapotranspiration flux partitioning using an Iso-SPAC model in a temperate grassland ecosystem

Thursday, 18 December 2014
Pei Wang, BNU Beijing Normal University, Beijing, China
To partition evapotranspiration (ET) into soil evaporation and vegetation transpiration (T), a new numerical Iso-SPAC (coupled heat, water with isotopic tracer in Soil-Plant-Atmosphere-Continuum) model was developed and applied to a temperate-grassland ecosystem in central Japan. Several models of varying complexity have been tested with the aim of obtaining the close to true value for the isotope composition of leaf water and transpiration flux. The agreement between the model predictions and observations demonstrates that the Iso-SPAC model with a steady-state assumption for transpiration flux can reproduce seasonal variations of all the surface energy balance components´╝îleaf and ground surface temperature as well as isotope data (canopy foliage and ET flux). This good performance was confirmed not only at diurnal timescale but also at seasonal timescale. Thus, although the non-steady-state behavior of isotope budget in a leaf and isotopic diffusion between leaf and stem or root is exactly important, the steady-state assumption is practically acceptable for seasonal timescale as a first order approximation. Sensitivity analysis both in physical flux part and isotope part suggested that T/ET is relatively insensitive to uncertainties/errors in assigned model parameters and measured input variables, which illustrated the partitioning validity. Estimated transpiration fractions using isotope composition in ET flux by Iso-SPAC model and Keeling plot are generally in good agreement, further proving validity of the both approaches. However, Keeling plot approach tended to overestimate the fraction during an early stage of glowing season and a period just after clear cutting. This overestimation is probably due to insufficient fetch and influence of transpiration from upwind forest. Consequently, Iso-SPAC model is more reliable than Keeling plot approach in most cases.The T/ET increased with grass growth, and the sharp reduction caused by clear cutting was well reflected. The transpiration fraction ranges from 0.02 to 0.99 during growing seasons, and the mean value was 0.75 with a standard deviation of 0.24.