Abstract

Vadose zone modeling has broad applications in modern theoretical and applied hydrology for various purposes. However, application of vadose zone models is usually restricted by a number of theoretical and practical problems. Some of the problems are investigated in this research project as a part of the Grassland Destabilization EXperiment (GDEX) conducted in a semi-arid region of Nebraska, U.S.A. We first present an analytical technique, which is based on the traveling wave solution, to generate an exact solution of boundary-value problems for Richards' equation. The technique is illustrated by numerical experiments of infiltration of water into soils with hydraulic properties described by Brooks-Corey and van Genuchten models. Examples of verification are given for HYDRUS-1D, a popular numerical code for solving Richards' equation.

A unique 3D dataset is then presented on the distribution of in-situ soil saturated hydraulic conductivity (K_fs) collected at the GDEX site. The results show that K_fs increases with depth up to 2m and is generally higher in highlands. There exists a significant correlation between K_fs and absolute elevation. The effect of short-term vegetation disturbance on K_fs is found to be minimal. The dataset also presents field evidence of a negative relationship between soil organic matter content and K_fs, which is contradictory to the traditional notion of a positive correlation between those two factors. Moreover, this negative relationship is highly non-linear and strongest at low levels of soil carbon (%C<0.1%). The most likely explanation for this phenomenon is that the water repellency induced by soil organic matter that inhibits water flow exceeds the positive influence of soil organic matter on soil aggregation and bulk density reduction.

Accuracy of using pedotransfer functions for estimating soil hydraulic characteristics in vadose zone models is investigated for evaluating groundwater recharge (GR). Fundamental components of vadose zone models (e.g., soil texture, boundary condition, and vertical variability in soil texture) are discussed. Under unit gradient flow condition, GR largely depends on n, a parameter measuring pore size distribution in the van Genuchten model. Under constant head condition, the distribution of GR depends on groundwater depth and soil texture. If vertical variability in soil textures is considered, especially for regions covered with eolian deposits, underlying coarser soil layers act as hydraulic barriers that reduce GR. The estimated GR at the GDEX site ranges from 2.48cm/year to 33.36cm/year with an average of 18.51cm/year and one standard deviation of 9.38cm/year.

Validity of Bouchet's complementary relationship (BCR) between actual and potential evapotranspiration is tested using a process-based vadose zone model for the first time. The model is applied to Nebraska with an area of about 200,000 km², where a significant spatial precipitation gradient exists. The simulation results strongly support the BCR. The study also shows that soil properties are unimportant for the validity of BCR. The results have important implications for evaluating impacts of climate changes on hydrological components, such as groundwater recharge.