Kriging the Water Table with Predicted Water Levels
Abstract
Salinity management relies heavily upon the mapping of the watertable to predict regions at risk of salinity. Current mapping techniques often ignore the influence of topographic and interpolate using only the reduced water level (RWL) at each bore. This paper builds on Peterson et al [2003] which addresses the above deficiencies using an advanced geostatistical method called Kriging with External Drift (KED). It presents methods for expanding the dataset with historic and qualitative information while incorporating the uncertainty in such data. The two sources of uncertainty addressed are 1) that due to rapidly fluctuating water levels and 2) predictions of current water level derived from bores decommissioned up to 10 years ago. The later allowed estimates of the current water level to be calculated from historic bore hydrographs using a multiple regression model. These modifications produced a 19% increase in the amount of input data and reduced the errors as quantified by cross validation. Most significantly, cross validation indicated that the multiple regression model is a valid and unbiased method of predicting future water levels independent of the years of extrapolation. Thus the future water table can be mapped without the influence of climatic fluctuations and making use of local knowledge and historic water level data.
Authors
T. Peterson & B. Barnett