Soil-water content characterisation in a modified Jarvis-Stewart model: A case study of a conifer forest on a shallow unconfined aquifer

Abstract

Groundwater-vegetation-atmosphere fluxes were monitored for a subtropical coastal conifer forest in South-East Queensland, Australia. Observations were used to quantify seasonal changes in transpiration rates with respect to temporal fluctuations of the local water table depth. The applicability of a Modified Jarvis-Stewart transpiration model (MJS), which requires soil-water content data, was assessed for this system. The influence of single depth values compared to use of vertically averaged soil-water content data on MJS-modelled transpiration was assessed over both a wet and a dry season, where the water table depth varied from the surface to a depth of 1.4m below the surface.Data for tree transpiration rates relative to water table depth showed that trees transpire when the water table was above a threshold depth of 0.8m below the ground surface (water availability is non-limiting). When the water table reached the ground surface (i.e., surface flooding) transpiration was found to be limited. When the water table is below this threshold depth, a linear relationship between water table depth and the transpiration rate was observed. MJS modelling results show that the influence of different choices for soil-water content on transpiration predictions was insignificant in the wet season. However, during the dry season, inclusion of deeper soil-water content data improved the model performance (except for days after isolated rainfall events, here a shallower soil-water representation was better). This study demonstrated that, to improve MJS simulation results, appropriate selection of soil water measurement depths based on the dynamic behaviour of soil water profiles through the root zone was required in a shallow unconfined aquifer system.