Spatio-temporal Modeling of Zero-Inflated Count: Efficient Posterior Computation and Application to Monitoring Capelin Distribution in the Barents Sea

Abstract

We consider a model for predicting the spatio-temporal distribution of a marine species based on zero-inflated count observation data that vary continuously over a specified survey region. The model is a mixture of two components; a one-point distribution at the origin and a Poisson distribution with spatio-temporal intensity, where both intensity and mixing proportions are related to some auxiliary information. We develop an efficient posterior computational algorithm for the model using a data augmentation strategy. An attractive feature of the modeling framework is that it accommodates scenarios where the auxiliary information is partially observed, or when the response variable is observed at spatially varying locations over non-uniform time intervals. We present results to show how utilizing the posterior distribution of the auxiliary information facilitates the successful prediction of future spatio-temporal distributions of an example marine species.