On the Use of Ensemble Kalman Filters to Predict Stream Discharge at Barton Springs, Edwards Aquifer, Texas

Recent drought and population growth in central Texas has increased attention on the sustainability of Barton Springs, located in Austin within the Edwards Aquifer. Degradation of the aquifer is harmful to both residents of the growing city of Austin, who use the aquifer as their primary drinking water source, and for the Barton Springs Salamander, whose only known habitat is the area immediately surrounding the spring. The urban and environmental needs for sustainable water in this region necessitate accurate means to predict stream discharge at Barton Springs. This paper explores the use of an ensemble Kalman filter (EnKF) to make real-time predictions of discharge at Barton Springs. Analysis using EnKF is based on the MODFLOW model developed by Scanlon et al. The EnKF model updates an ensemble of state variables (hydraulic conductivity) as control variables (stream discharge) are acquired sequentially. Updates to the state variable are applied in one of two ways: once per time step or through an iterative approach. EnKF results are then compared to results obtained by Scanlon et al. to assess the ability of EnKF to predict stream discharge.