The application of Kriging technique to mathematical modelling of estuarine water quality

Abstract

It is essential that estuarine modelling and surveying are carried out simultaneously because not only does the latter provide data required by the former but also the former is verified with data from the latter. This study integrates both research subjects from the panoramic point of view, aiming at improving modelling accuracy and reliability and increasing survey efficiency. Partially stratified estuaries are the most difficult types of estuaries to be modelled, in particular, the velocity field in such an estuary. A review of two commonly used methods to determine the velocity field, i. e., theoretical method and empirical method, revealed their inadequacies in real applications. Thus, a new approach using Kriging technique was originated and was tested on a finite element model of water quality. The model was formulated using a Galerkinfinite element method and was programmed in Fortran. Comparison between the simulation results and the field measurements for a salinity intrusion showed a high simulation accuracy. It is believed that the model in combination with the new approach would be a useful tool for estuarine modelling. The generalized Kriging method ensured that the new approach would be appropriate in practice. It was also applied to the investigation of sampling stations in the partially mixed estuary of the River Tees. It is essential to know how many sampling stations should be used and how they should be positioned. Two procedures were designed for solving the survey problems. They were the procedure of overall variance and the procedure of re-estimation. These procedures were capable of quantifing the relative significance of each sampling station and detecting redundant sampling stations. The 1975 survey was investigated, and useful conclusions were obtained.