H13A-1472
State-wide space-time water table mapping: cautionary tales, tribulations and resolution
Monday, 14 December 2015
Poster Hall (Moscone South)
Tim J Peterson1, Xiang Cheng2, Elizabetta Carrara3, Andrew William Western1, Justin Francis Costelloe4, Andrew J Frost3 and Chris V McAuley5, (1)University of Melbourne, Parkville, Australia, (2)Agriculture Research and Department of Economic Development, Jobs, Transport and Resourcesch and Development, Agriculture Research and Development, Melbourne, Australia, (3)Bureau of Meteorology, Melbourne, Australia, (4)Melbourne University, Melbourne, Australia, (5)Department of Environment, Land, Water & Planning, Water and Catchments, Melbourne, Australia
Abstract:
Historically, insufficient quantitative value has been derived from state groundwater monitoring networks. Water level data are occasionally used for calibrating local scale groundwater models and for graphical analysis, but very rarely are they used to identify regional groundwater processes and quantify changes in groundwater dynamics over time. Potentiometric maps have occasionally been derived to assist understanding of regional processes but generally they are derived for one point in time, often simply using an average water level over a year or season. Consequently, dynamics of regional groundwater over time has been compromised.
Kriging with external drift (KED) has been a widely adopted approach for regional scale potentiometric mapping in recent years. However, it has a number of unacknowledged fundamental weaknesses - specifically, excessive noise in the head, sensitivity to observation errors and questionable estimation in upland regions and in coastal regions dominated by radial flow. These weaknesses are illustrated and then a multivariate localised colocated cokriging approach is proposed that locally reduces the excessive noise from KED and incorporates the coast line and streams into the estimation. Combined with the temporal interpolation of groundwater head (Peterson & Western, 2014), the approach allows regional scale mapping for a single point in time. To illustrate the approach, the monthly water table level was mapped across Victoria, Australia, from 1985 to 2014. Using the maps, the location and the nature/magnitude of major changes in groundwater dynamics were identified and the surface-groundwater connectivity of major rivers was estimated over time. While geological knowledge can be incorporated, this approach allows data-driven insights to be derived from groundwater monitoring networks without the usual assumptions required for numerical groundwater modeling.
Peterson, T. J., and A. W. Western (2014), Nonlinear time-series modeling of unconfined groundwater head, Water Resour. Res., 50, 8330–8355, doi:10.1002/2013WR014800.
