Short-term forecasts of air-pollution levels in big cities are now reported in news papers and other media outlets. Studies indicate that even short-term exposure to high levels of an air pollutant called atmospheric particulate matter (PM) can lead to long-term health effects. Data are typically observed in fixed monitoring stations throughout a study region of interest at different time points. Statistical spatio-temporal models are appropriate for modeling these data. In this article we consider short term forecasting of these spatio-temporal processes using a Bayesian Kriged-Kalman filtering model. The spatial prediction surface of the model is built using the well known method of Kriging for optimum spatial prediction and the temporal effects are analyzed using the models underlying the Kalman filtering method. The full Bayesian model is implemented using MCMC techniques which enable us to obtain the optimal Bayesian forecasts in time and space. A new cross-validation method based on the Mahalanobis distance between the forecasts and observed data is also developed to asses the forecasting performance of the implemented model.