Removal of Barometric Pressure Effects and Earth Tides from Observed Water Levels

Abstract

The effects of barometric pressure and earth tide changes are often observed in ground water level measurements. These disturbances can make aquifer test interpretation difficult by masking the small changes induced by aquifer testing at late times and great distances. A computer utility is now available that automatically removes the effects of barometric pressure and earth tides from water level observations using regression deconvolution. This procedure has been shown to remove more noise then traditional constant barometric efficiency techniques in both confined and unconfined aquifers. Instead of a single, instantaneous barometric efficiency, the procedure more correctly accounts for the lagged responses caused by barometric pressure and earth tide changes. Simultaneous measurements of water levels (or total heads) and nearby barometric pressures are required. As an additional option, the effects of earth tides can also be removed using theoretical earth tides. The program is demonstrated for two data sets collected at the Waste Isolation Pilot Plant, Carlsbad, New Mexico. The program is available free by request at http://www.hydrology.uga.edu/tools.html.

Understanding diagnostic plots for well-test interpretation

“…Therefore, the original data must have been preprocessed in order to remove the head variations independent of pumping (both natural or induced by nearby pumping operations). For example, tools are available to remove barometric or earth tide effects (Toll and Rasmussen 2007). This is a standard procedure (Dawson and Istok 1991) that needs to be applied whether one uses diagnostic plots or not, therefore this aspect will not be discussed further here.…”

Guide to using Multiple Regression in Excel (MRCX v.1.1) for Removal of River Stage Effects from Well Water Levels

“…In MRCX, users can choose to run the multiple regression using either a) the original data, or b) the first differences of the original data (change in water level between successive time steps). Rasmussen and Crawford (1997) suggested using the original time series data in the multiple regression correction; however, subsequent workers have chosen to use differenced data (change in water level) in correcting barometric (Spane, 1999 and2002;Toll and Rasmussen, 2007) and river effects (Vermeul et al, 2009;Spane and Mackley, 2010).…”

“…Rasmussen and Crawford (1997) suggest increasing the maximum lag term (n) to a value sufficiently high to incorporate long-term responses. Others (Spane, 1999 and2002;Toll and Rasmussen, 2007) recommend increasing the number of lags until the response function stabilizes. In practice, this is observed as an asymptotic approach in the cumulative response to some maximum response value.…”

“…Rasmussen and Crawford (1997) reported better success with using original data in the regression deconvolution correction of barometric effects. However, others have chosen to use the first differences instead for barometric (Spane 1999 and2002;Toll and Rasmussen 2007) and river correction (Vermeul et al, 2009;Spane and Mackley, 2010).…”

“…As discussed in Section 1.2.1, the maximum number of lags should be increased until either the RRF asymptotically approaches a maximum cumulative response value or there is no major improvement in the model results (Rasmussen and Crawford, 1997;Toll and Rasmussen, 2007;Spane and Mackley, 2010). In practice, maximum lags of several hundred hours or more may be necessary to adequately capture the full river-stage effects in water levels in wells located several hundred meters from the river (e.g., Vermeul et al, 2008;Spane and Mackley, 2010).…”