Probabilistic estimation and prediction of groundwater recharge in a semi-arid environment

Abstract

Quantifying and characterizing groundwater recharge are critical for water resources management. Unfortunately, low recharge rates are difficult to resolve in dry environments, where groundwater is often most important. Motivated by such concerns, this thesis presents a new probabilistic approach for analyzing diffuse recharge in semiarid environments and demonstrates it for the Southern High Plains (SHP) in Texas. Diffuse recharge in semi-arid and arid regions is likely to be episodic, which could have important implications for groundwater. Our approach makes it possible to assess how episodic recharge can occur and to investigate the control mechanisms behind it. Of the common recharge analysis methods, numerical modeling is best suited for considering control mechanisms and is the only option for predicting future recharge. However, it is overly sensitive to model errors in dry environments. Natural chloride tracer measurements provide more robust indicators of low flux rates, yet traditional chloride-based estimation methods only produce recharge at coarse time scales that mask most control mechanisms. We present a data assimilation approach based on importance sampling that combines modeling and data-based estimation methods in a consistent probabilistic manner. Our estimates of historical recharge time series indicate that at the SHP data sites, deep percolation (potential recharge) is indeed highly episodic and shows significant interannual variability. Conditions that allow major percolation events are high intensity rains, moist antecedent soil conditions, and below-maximum root density. El Niño events can contribute to interannual variability of percolation by bringing wetter winters, which produce modest percolation events and provide wet antecedent conditions that trigger spring episodic recharge.

(cont.) Our data assimilation approach also generates conditional parameter distributions, which are used to examine sensitivity of recharge to potential climate changes. A range of global circulation model predictions are considered, including wetter and drier futures. Relative changes in recharge are generally more pronounced than relative changes in rainfall, demonstrating high susceptibility to climate change impacts. The temporal distribution of rainfall changes is critical for recharge. Our results suggest that increased total precipitation or higher rain intensity during key months could make strong percolation peaks more common.