Description of the Spatial and Temporal Dependence in Surface and Groundwater
Water is in many of its occurrences highly mobile in space and time. Unfortunately observations are limited, thus there is a need for estimating water quality and quantities at unsampled times and locations. For this purpose the dependence structure has to be investigated.
The purpose of this presentation is to present different geostatistical methods to describe the specific features of the fields under study. Specifically properties related to symmetry are investigated. Examples show that many observed fields show non symmetrical behaviour. The reasons for this will be discussed using examples. The consequence of these investigations is that Gaussian processes are not suitable to describe these fields. This also means that well known Kriging methods are not optimal for interpolation. Possible alternatives using copulas are presented. The problem of higher order dependence and tail dependence are also discussed.
Examples of measured rainfall and groundwater quality parameters illustrate the methodologies.