Discovery of Climate Indices using Clustering Michael Steinbach

Discovery of Climate Indices using Clustering Michael Steinbach Steven Klooster Christopher Potter Rohit Bhingare, School of Informatics University of Edinburgh

Overview • Aim: Applying Clustering to the task of finding interesting patterns in earth science data. • Key interests and research goals • Climate Indices • Using SVD analysis to find Spatial/Temporal Patterns • Using Clustering for discovery of indices • Conclusion and Future Work

Key Interest l l Find global climate patterns of interest to Earth Scientists Finding connection between the ocean/atmosphere and land. Average Monthly Temperature NINO 1+2 Index

The El Nino Climate Phenomenon • El Nino is the anomalous warming of the eastern tropical region of the Pacific. Normal Year: Trade winds push warm ocean water west, cool water rises in its place El Nino Year: Trade winds ease, switch direction, warmest water moves east.

Climate Indices • A climate index is a time series of temperature or pressure – Connecting the Ocean/Atmosphere and the Land – Commonly based on Sea Surface Temperature (SST) or Sea Level Pressure (SLP) • Why climate indices? – They extract climate variability at a regional or global scale into a single time series. – They are well-accepted by Earth scientists. – They are related to well-known climate phenomena such as El Nino.

Finding Patterns using SVD and Clustering • SVD Analysis: – Impressive for finding the strongest patterns falling into independent subspaces. – All discovered signals must be orthogonal (difficult to attach physical interpretation) – Weaker signals may be masked by stronger signals. • Use of Clustering: – The centroids of clusters summarize the behaviour of the ocean/atmosphere in those regions.

Clustering Based Methodology • The SNN Procedure: – Apply the SNN clustering on the SST (or SLP) data over a specific time period. – Eliminate all the clusters with poor areaweighted correlation. – The cluster centroids of remaining clusters are potential climate indices : <G 0, G 1, G 2, G 3>

Clusters with correlation to known indices G 0 G 2 G 1 G 3

Conclusion • Clustering plays a useful role in the discovery of interesting ecosystem patterns. • Clustering is used to discover previously unknown relationships between regions of the land sea.

Future Work • Can all climate indices be represented using clusters? • Extending the research to land ocean variables - Many more opportunities for data mining/data analysis in Earth Science data. Earth Observing System: Detecting patterns such as finding relationships between fire frequency and elevation as well as topographic position