Analysis of root-zone soil moisture control on evapotranspiration in two agriculture fields in Australia

Abstract

© International Congress on Modelling and Simulation, MODSIM 2013.All right reserved.Soil moisture content in the root-zone is an important variable in modeling hydrological and biophysical processes and agricultural applications. For example, partitioning of surface energy exchange into the latent and the sensible heat fluxes is affected much by soil water content. Typically surface moisture content has been mapped using the microwave remote sensing techniques at large scales. In addition to the microwave remote sensing, there is a growing interest in mapping high-resolution soil moisture content using remotely sensed thermal signals or evapotranspiration (ET). This new technique is based on the assumption that the evaporative fraction (EF) or the ratio of actual ET to potential ET (AET: PET) is controlled by soil water content available for bare soil evaporation or vegetation transpiration. However, there are a number of environmental and biological factors contributing to EF and AET: PET such as the net radiation, vapor pressure deficit, vegetation biomass and growth stage, etc. This work investigates effectiveness of the EF and AET: PET based root-zone soil moisture estimation in the agricultural landscapes. In order to analyze the sensitivity of EF and AET/PET to soil moisture, meteorological and biophysical variables, we use continuous observations of surface reflectance, eddy-covariance ET, profile soil moisture and other meteorological variables collected by tower-based Spectro-Eddy-Covariance (SEC) systems installed in experimental farm sites in Victoria, Australia. Two SEC systems are installed in two rain-fed agricultural fields; wheat, pasture covered with lucerne. Statistical analyses were conducted to identify key contributors to EF and AET: PET, which can be ultimately used to constrain ET-based soil moisture estimation scheme. The correlation between EF and AET/PET was good throughout the vegetation period with R2 value 0.83. Therefore, we showed that EF or AET: PET exhibits consistent correlations with soil water available for vegetation in both study sites. The exponential model appears to provide reasonable estimates of EF using soil moisture with R2 value of 0.78. The model performance was further decreased yielding an R2 value of 0.52 with inclusion of bare soil conditions. EF vs. soil moisture at surface and root-zone is sensitive to vegetation biomass. For soil moisture 0-8 cm, correlation is higher during bare soil condition while for 0-30 cm, correlation is significantly higher when there is enough biomass. The relation between the EF and soil moisture is strong when ET is water limited and correlation become weak when ET is energy limited. Implications are that, when estimating root-zone soil moisture using EF, we have to consider net radiation and NDVI greater at critical threshold levels and soil depth should be specified properly. Analysis indicates that the net radiation and NDVI are the most important factors, other than soil moisture, that influences the ET versus soil moisture relationship, and the influence varies with season.