Cross-Correlation Analysis in Evaluating Spatio-Temporal Data Dependence of Climate Variables Through the GSTAR Model

Abstract

Climate change analysis requires an approach which is capable to accommodate the dynamics of relationships between climatological variables in space and time dimensions. Temperature, humidity, and rainfall vary temporally and exhibit spatial dependence across locations. This study applies the Generalized Space-Time Autoregressive (GSTAR) model to analyze the spatial and temporal dependence patterns of climate variables in Tasikmalaya. The novelty of this study lies in the cross-correlation analysis of climate variables using actual data and model estimation results. This analysis can be used to assess how well the GSTAR model maintains the spatio-temporal dependence pattern. GSTAR modeling is performed by applying the Three-Stage Iterative Box-Jenkins method to space-time data. The results indicate that GSTAR(3;1,1,1) is the best-fitting model. Furthermore, this model consistently captures historical data patterns and accommodates the dynamic dependencies between climate variables at the two observation locations. The results show that the GSTAR(3;1,1,1) model consistently represents historical data patterns and accommodates the dynamics of relationships between meteorological variables at the observed locations. This finding confirms that the GSTAR model is an effective approach for capturing the spatio-temporal dependence of climate data, particularly in preserving the natural relationship patterns in actual data.

Keywords

• Climate variables
• cross-correlation
• dependence
• estimation
• space-time