Modified spatio-temporal interpolation algorithm using a combination of Kriging method and Kalman filter

Abstract

Spatio-temporal geostatistical modeling constitutes a challenge within diverse scientific and engineering disciplines. This dissertation proposes a novel modification of spatial kriging called spatio-temporal dual kriging (ST-DK), incorporating trend functions into three coefficient types: fixed coefficient, adaptive coefficient, and adaptive coefficient with Kalman filter. The adaptive coefficients are estimated using a Kalman filter, enabling the model to capture complex spatio-temporal dynamics. Furthermore, in order to illustrate the efficacy of the proposed technique, ST-DK is compared with the classical spatio-temporal regression kriging (ST-RK) method for temperature and air pressure data estimation across Thailand in year 2017. The results reveal that both ST-DK and ST-RK employed adaptive coefficient and adaptive coefficient with Kalman filter outperform these two methods using fixed coefficient counterparts for air pressure data. Additionally, the ST-DK model consistently exhibits superior performance comparing to the ST-RK model in air pressure estimation.