Identifying and Removing Barometric Pressure Effects in Confined and Unconfined Aquifers

Abstract

Failing to account for barometric pressure effects in water level measurements can introduce errors by misestimating the total head and by adding noise to water level measurements. For determining the total head in an aquifer, we assert that the air pressure head at the water surface in the well must be added to measured water levels (equivalent to using an absolute pressure transducer) even though the resulting values may have larger temporal and spatial variability than the original water level measurements. At the Savannah River Site in South Carolina, the average barometric pressure variation is 6 to 7 cm, with a range of over 30 cm. Failure to account for barometric pressure variability could result in misestimation of the direction and magnitude of the hydraulic gradient at the site. We also demonstrate procedures for removing barometric effects, such as to reduce noise during an aquifer pumping test, and to identify mechanisms by which barometric pressure affects water levels. Three mechanisms are summarized including: an instantaneous response for confined aquifers; a delayed response due to borehole storage in confined and unconfined aquifers; and a delayed response in unconfined aquifers due to the passage of barometric pressure changes through the unsaturated zone. Using data from the Savannah River Site, barometric efficiencies are estimated using linear regression and a modification of Clark's Method. Delayed responses are estimated using regression deconvolution. The type of barometric effect provides diagnostic information about whether the aquifer is confined or not, the presence of borehole storage or skin effects, and the air diffusivity coefficient within the unsaturated zone. We also show how removal of barometric pressure effects improves the ability to observe otherwise unnoticeable effects.