Multivariate dependence among extremes, abrupt change and anomalies in space and time for climate applications
RIS ID
109016
Abstract
This paper discusses multivariate spatio-temporal dependence between extremes or abrupt change and unusual values or anomalies in the context of climate dynamics and climate change. In climate, as in many other applications, anomalies (or extremes) in one variable like sea surface temperature may be a precursor for extremes (or abrupt change) in another variable like regional precipitation. In addition, this multivariate dependence may be spatially or temporally lagged, owing to climate "teleconnections". However, the anomalies may not be easily detectable and their dependence with extremes and rapid change may be difficult to quantify. This paper provides a brief review of the literature, which is followed by a description of critical gaps, both in the data or computational sciences as well as in the climate sciences. The quantification and visualization of multivariate dependence among extreme values and anomalies in highly nonlinear or stochastic systems is an emerging research area in theoretical statistics, with limited development in application areas and/or for massive or disparate space-time data. Further development is needed in these areas for multiple domains ranging from climate sciences and geography to sensor networks and national security.
Publication Details
Ganguly, A., Hsing, T., Katz, R., Erickson, D. J., Ostrouchov, G., Wilbanks, T. J. & Cressie, N. (2005). Multivariate dependence among extremes, abrupt change and anomalies in space and time for climate applications. KDD-2005 Workshop on Data Mining Methods for Anomaly Detection: Workshop Notes (pp. 25-26).