Simulating water flow in Hydrus 1D coupled with volumetric water content from electromagnetic induction-based model

Abstract

Understanding groundwater flow dynamics is vital for a number of applications which includes water budget modeling, crop modeling and to understand long term soil water interactions. One of the challenges in simulating water flow in HYDRUS 1D model is the initial condition at which the model is analyzed. This thesis proposes a coupled HYDRUS 1D and Electromagnetic Induction (EMI) based Volumetric Water Content (VWC) prediction model to simulate accurate water flow. The main objective of this thesis is to calibrate, evaluate, and forecast the movement of water in Hydrus-1D coupled with a multifrequency EMI sensor. To achieve the intended objectives, an EMI based prediction model was developed to predict VWC from the Apparent Electrical Conductivity (ECa) of an EMI sensor. ECa surveys were conducted and VWC was obtained using a Time domain Reflectometer (TDR) sensor in a field located south of Manhattan, Kansas. The TDR sensors were calibrated to the site and correlation between ECa and VWC was studied. Five different Regression models which include linear, logarithmic, power, exponential and 2nd order Quadratic models were developed, and their performance analyzed. Regression models for relative and absolute changes were also studied along with a nonlinear Waxman-Smits model. Mean Absolute Error (MAE) was used to select the best model amongst the 17 models. The logarithmic model was found to be the best model to predict VWC from EMI sensors as it had the least MAE of 1.46% amongst all the models. The developed EMI based VWC prediction model was used to predict VWC and was used as an initial condition in HYDRUS 1D. Atmospheric conditions and free drainage condition was used as boundary conditions to predict VWC for later dates which were then compared with the VWC obtained from the EMI based model. The HYDRUS 1D model had a good level of correlation with the EMI based model and the MAE was found to be 1.49% indicating the potential to use a coupled HYDRUS 1D and EMI based model to predict future groundwater flow in near surface soil.