Carolina Euan: Threshold autoregressive models for spatiotemporal data

In this talk, I will introduce the use of threshold autoregressive models to describe spatiotemporal dynamics of precipitation data. While threshold autoregressive models have been widely studied in time series with applications in econometrics, few studies have utilized these models for environmental data applications. One key difference is that econometric data typically consists of a single, sometimes long time series, whereas environmental data can have replicates or spatial dependencies. Our model is a precipitation model that is based on regimes. This model builds the regime based on past values of neighboring stations. The threshold autoregressive dynamics in this model reflect the presence of different precipitation systems, such as convective and frontal. Finally, we will discuss the advantages and limitations of using threshold autoregressive models for environmental applications.