4-D Statistical Surface Method for Visual Change Detection in Forest Ecosystem Simulation Time Series

Abstract

Rising uncertainties associated with climate change compel forest management planning to include forest ecosystem simulations. The output of such models is often of high spatio-temporal complexity and difficult to interpret for the user. This contribution describes a novel visualization method called four-dimensional (4-D) statistical surfaces, which aims at improving the visual detection of change in time series. The method visualizes attribute values as surfaces, which are interpolated and animated over time; the interactive attribute surfaces are combined with color-coding and contour lines to support absolute and relative height judgment as well as faster perception and better location of change. A design study and prototypical implementation of the visualization method is described in this contribution. Time-series simulation results of LANDIS-II, a commonly used modeling tool in forest ecology, as well as a temporal vegetation index dataset (NDVI) are visualized using 4-D statistical surfaces. Usability challenges are addressed based on explorative interviews with a small group of users. The method is not limited to ecological model output; it can be used to create three-dimensional (3-D) temporal animations of arbitrary time-series datasets where parameters are supplied in regular raster format.