Vector Autoregressive Models with Spatially Structured Coefficients for Time Series on a Spatial Grid

Abstract

We propose a parsimonious spatiotemporal model for time series data on a spatial grid. Our model is capable of dealing with high-dimensional time series data that may be collected at hundreds of locations and capturing the spatial non-stationarity. In essence, our model is a vector autoregressive model that utilizes the spatial structure to achieve parsimony of autoregressive matrices at two levels. The first level ensures the sparsity of the autoregressive matrices using a lagged-neighborhood scheme. The second level performs a spatial clustering of the non-zero autoregressive coefficients such that nearby locations share similar coefficients. This model is interpretable and can be used to identify geographical subregions, within each of which, the time series share similar dynamical behavior with homogeneous autoregressive coefficients. The model parameters are obtained using the penalized maximum likelihood with an adaptive fused Lasso penalty. The estimation procedure is easy to implement and can be tailored to the need of a modeler. We illustrate the performance of the proposed estimation algorithm in a simulation study. We apply our model to a wind speed time series dataset generated from a climate model over Saudi Arabia to illustrate its usefulness. Limitations and possible extensions of our method are also discussed.