We propose a forecasting method called spatio-temporally explicit model averaging (STEMA) to combine spatial and temporal information through model averaging. We examine the performance of STEMA against two popular forecasting models and a modern spatial prediction model: the autoregressive integrated moving averages (ARIMA) model, the Bayesian hierarchical model, and the varying coefficient model. We focus on applying the methods to four species of Alaskan groundfish for which only catch data are available. Our method reduces forecasting errors significantly for most of the tested models when compared to ARIMA, Bayesian, and varying coefficient methods. The STEMA method is capable of accounting for spatial information in forecasting and can be applied to various data because of its flexible varying coefficient model structure. It is therefore a suitable forecasting method for many applications in areas including ecology, epidemiology, and climatology.