article
A nonhomogeneous Poisson process geostatistical model
Registro en:
10.1007/s00477-016-1275-x
Autor
Morales, Fidel Ernesto Castro
Vicini, Lorena
Hotta, Luiz K.
Achcar, Jorge A.
Resumen
This paper introduces a new geostatistical model
for counting data under a space-time approach using nonhomogeneous
Poisson processes, where the random intensity
process has an additive formulation with two
components: a Gaussian spatial component and a component
accounting for the temporal effect. Inferences of
interest for the proposed model are obtained under the
Bayesian paradigm. To illustrate the usefulness of the
proposed model, we first develop a simulation study to test
the efficacy of the Markov Chain Monte Carlo (MCMC)
method to generate samples for the joint posterior distribution
of the model’s parameters. This study shows that the
convergence of the MCMC algorithm used to simulate
samples for the joint posterior distribution of interest is
easily obtained for different scenarios. As a second illustration,
the proposed model is applied to a real data set
related to ozone air pollution collected in 22 monitoring
stations in Mexico City in the 2010 year. The proposed
geostatistical model has good performance in the data
analysis, in terms of fit to the data and in the identification
of the regions with the highest pollution levels, that is, the
southwest, the central and the northwest regions of Mexico
City.