A Data-Driven Approach to Dynamic Spatio-Temporal Clustering
Climate data sets available currently, whether these are paleo-reconstructions, long-term weather monitoring records, or remote sensing data, contain a whelm of space-time information that needs to be analyzed under the pressure of computational and data storage requirements. This has led to a spark of interest in dynamic space-time clustering algorithms that are particularly suitable in the analysis of data streams. The trend-based clustering algorithm TRUST allows for space-time clustering in real time. However, this method requires the user to set a number of tuning parameters by hand. Here we propose a data-driven approach to automatically select the tuning parameters based on a penalized loss function. We focus on the two most important parameters of the TRUST algorithm: the current likeness of observations across the slide level and the temporal persistence within an analyzed window. We demonstrate the performance of the enhanced clustering procedure using simulated time series and illustrate its applicability using long-term records of water temperature in the Chesapeake Bay.
Schaeffer, E.*, J. Testa, Y. Gel, and V. Lyubchich. 2016. On information criteria for dynamic spatio-temporal clustering. In A. Banerjee et al. (Eds.) Proceedings of the 6th International Workshop on Climate Informatics: CI 2016. NCAR Technical Note NCAR/TN-529+PROC, p. 5-8. DOI: 10.5065/D6K072N6.
Schaeffer, E.*, J. Testa, Y. Gel, and V. Lyubchich. 2017. Enhanced data analysis using modern dynamic spatio-temporal clustering. ASLO Aquatic Sciences Meeting, Honolulu, Hawaii.
Schaeffer, E.*, J. Testa, Y. Gel, and V. Lyubchich. 2016. On information criteria for dynamic spatio-temporal clustering. The 6th International Workshop on Climate Informatics, Boulder, Colorado.