Groundwater head sampling based on stochastic analysis

Abstract

The problem of determining a uniform sampling time interval for monitoring serially correlated groundwater heads is the main subject discussed here. The problem is approached by stochastic time series analysis and modeling. An autoregressive integrated moving average model is assumed to fit the underlying series. Given that groundwater head data can be sampled at different time intervals and that the same stochastic model must represent the time series regardless of the sampling timescale, the parameters of the underlying model for the series sampled at a given arbitrary time interval h are obtained as a function of h and as a function of the model parameters for the series sampled at a unit time interval. This is accomplished by linking the derived variances and autocovariances at the two sampling scales. The derived equations and the sampling design procedure are tested and illustrated using the groundwater head data of Collier County, Florida.