Understanding Infiltration and Groundwater Flow at an Artificial Recharge Facility using Time-lapse Gravity Data

Groundwater provides a fundamental resource for modern life. Throughout the world, groundwater is managed by storing (recharging) it underground in natural aquifers for future withdrawal and consumptive use. In Arizona, over 4 million people benefit from managed aquifer storage, but little effort is made to track the movement of recharged water through the subsurface. Motivated by current limitations in our ability to monitor percolation and groundwater movement at the scale of a recharge facility, an effort to collect time-lapse gravity data was carried out at the Southern Avra Valley Storage and Recovery Project (SAVSARP) operated by the City of Tucson, Arizona. In addition to collecting water-level data 12 wells, there were three primary gravity experiments: (1) five continuously-recording gravity meters (2 iGrav superconducting gravity meters and 3 gPhone gravity meters) were installed semi-permanently in control buildings adjacent to the recharge basins, (2) absolute gravity measurements were made at nine locations over a 17 month period, and (3) three relative-gravity campaigns were carried out on a network of 70 stations. This large-scale controlled experiment, with known infiltration and pumping rates, resulted in one of the most comprehensive datasets of its kind. Gravity data led to several hydrologic insights, both through direct measurement and modeling. First, the infiltration rate could be estimated accurately based on the initial rate of gravity change during infiltration, regardless of the specific yield. Using two gravity meters, the depth, and therefore speed, of the wetting front beneath a recharge basin was observed, including the time at which the water table was reached. Spatial maps of gravity change from relative gravity surveys show areas where infiltration efficiency is highest, and where groundwater accumulates; storage accumulated preferentially to the west of the recharge basins, away from pumping wells. Such information would be valuable for planning the location of pumping wells at a new facility. Gravity data were useful for calibration of a Modflow-NWT groundwater-flow model using the Unsaturated Zone Flow package to simulate recharge; the reduction in the posterior parameter distribution compared to the a priori estimate was substantial and similar to head data. In contrast to model-simulated head data, model-simulated gravity data were less sensitive to more distant model elements and more effective for calibration of a superposition-type model. Observed head data had a strong regional signal reflecting basin-scale conditions with only minor variation associated with individual recharge basins, and were therefore of limited usefulness for model calibration. Together, the methods developed by the study and interpretations they made possible suggest that gravity data are an effective way to better understand large-scale infiltration and groundwater movement