Kriging and Kalman filter to estimate dynamic spatio-temporal models

Nowadays there is a growing interest in understanding the dynamics of certain physical and biological processes. Those concepts are being partially observed and are generated on a large space and time scale. The use of spatio-temporal statistical models have been increase in a wide variety of scientific disciplines such as mapping diseases in certain re- gions, interpretation of oil industry seismic traces, robotic sensor networks analysis, and monitoring of stations meteorological among other applications. This methodology is appropriate to describe and predict spatial explicit processes, which evolves over time. A Bayesian methodology that involves the combination of a universal Kriging filter and the Kalman filter were proposed. The spatial prediction surfaces of the model were con- structed using the Kriging algorithm and the Kalman filter algorithm. It will be able to estimate the temporal effects. Kriging provides a successful estimation approach from the point of view of spatial statistics. On the other hand, the Kalman filter facilitates describe a well-established recursive procedure in order to estimate models in the form of space- state. Some measures of goodness of fit were used to validate model predictions. The methodology was illustrated using 30-year time series from three meteorological stations in Ecuador. The unified structure of the model allows predictions about temperature, precipitation and humidity in the 3 states analyzed to obtain good adjustments. Under- standing spatial patterns and trends can help to evaluate policies which contribute to climate change reduction. The root mean square error was used as a measure of goodness of fit to measure the algorithm estimation quality and to get satisfactory results.

• Autore principale: Becerra Loaiza, Inti Israel (author)
• Natura: bachelorThesis
• Lingua: eng
• Pubblicazione: 2020
• Soggetti: Kriging Kalman filter; Universal Kriging filter; Kalman filter; Spatio-temporal models
• Accesso online: http://repositorio.yachaytech.edu.ec/handle/123456789/135