Time-series augmentation of semantic kriging for the prediction of meteorological parameters

Spatio-temporal pattern analysis of meteorological parameters has been studied extensively in the field of remote sensing (RS) and geographic information system (GIS). It is an important data mining strategy for modeling the temporal dynamics of these parameters and forecasting them in future time instances. The meteorological parameters, measured near the earth surface, are eminently influenced by the surrounding land-cover distribution of the terrain. For the time-series prediction of these parameters, the knowledge of spatial land-covers can be contemplated within the space-time trade-off between the sample points. This work presents a new kriging based spatio-temporal interpolation method, namely times-series semantic kriging (SemK_ts) which deals with land-cover dynamics and incorporates this knowledge for better prediction accuracy. It is a time-series extension of our earlier work on semantic kriging (SemK) for spatial interpolation [1] [2]. Experimentation has been carried out by considering real land surface temperature data in the spatial region Kolkata, India. It has been observed that the semantically enhanced space-time trade-off analysis by SemK_ts yields more accurate result than most of the popular methods for prediction.