Kalman filter updating of possibilistic hydraulic conductivity

Preferential sampling regimes too often leave hydrogeological site characterization in need of hydraulic conductivity data. Furthermore, faith in a kriged hydraulic conductivity field (K-field) wanes when viable data are few. A straightforward solution to this problem is the application of expert knowledge to the identification of hydraulic conductivity at specified locations. However, the nature of this supplementary data suggests that its use should be limited to the initial estimation of a K-field. Available hydraulic conductivity measurements are used to update the initial estimate of the K-field in a fuzzy Kalman filtering procedure. The results are fuzzy and can be defuzzified into traditional crisp estimates. An application of these protocols to the CIBA-Geigy site in New Jersey is provided to illustrate the ease of the procedure and the impact of viable expert input.