Potsdam Research Cluster for Georisk Analysis, Environmental Change and Sustainability
The research goal of
is a better understanding of natural hazards and their impact. In this interdisciplinary cluster of university and non-university institutions in the Potsdam-Berlin region, we develop methods for an advanced data analysis of massive spatio-temporal data from simulation models, geoarchives and measurement instruments.
Project 1: Extraction of characteristic spatial states from long geospatial time series
We develop a method of combined automatic and visual analysis that allows users to identify characteristic spatial states in geospatial time series, stemming from, e.g., computer simulations. In contrast to strictly automatic methods, the number of relevant spatial states does not have to be set a priori. Rather, users can identify this number in an interactive analysis process. To this end, we combine an automated analysis step with interactive visual data exploration: We cluster all time steps of the time series according to their spatial similarity and aggregate them into a hierarchy. A visual interface allows users to explore this hierarchy to identify characteristic spatial states and to analyze their occurrence over time.
Köthur, P.; Sips, M.; Kuhlmann, J.; Dransch, D. (2012): Visualization of geospatial time series from environmental modeling output. In: Proc. Eurographics Conf. on Visualization (EuroVis) 2012 - Short Papers. Goslar, Germany: Eurographics Association; 2012. p. 115–119.
Projekt 2: Visual Analytics Approaches for the Multiscale Exploration of Time Series
Geo-scientific time series often involve a large amount of measurements over a long period and are often irregularly sampled along time axis which makes it difficult to analyze these time series with standard analytical approaches (in our research project, for example, a time series covers almost 600,000 years, and has a variable sampling rate varying from one month to several decades). In collaboration with scientists at the GFZ and the Potsdam Institute for Climate Impact Research, we develop interactive visualizations of geo-scientific time series at different levels of abstraction, allowing researchers to identify patterns at different time scales.
Sips, M.; Köthur, P.; Unger, A.; Hege, H.-C.; Dransch, D. (2012): A Visual Analytics Approach to Multiscale Exploration of Environmental Time Series. IEEE Transactions on Visualization and Computer Graphics, 18, 12, 2899-2907.
Prof. Dr. Doris Dransch
Dr. Mike Sips