Spatiotemporal Helixes for Modeling Environmental Data Kristin Eickhorst, Arie Croitoru, Peggy Agouris, Tony Stefanidis Department of Spatial Information Science and Engineering National Center for Geographic Information and Analysis University of Maine, Orono ME 04469-5711 {snoox, croitoru, peggy, tony}@spatial.maine.edu Abstract Spatiotemporal helixes are a new method for modeling the changes an object experiences over time. They have the potential to be used as a predictive tool for geographical and biological applications. This paper presents the formal foundations of these helixes and includes experiments to demonstrate their usefulness when data collection is not optimal, such as when noise is present or when there is more than one object of interest present in a single video stream. 1. Introduction There are a host of biological and geographical applications that could benefit from the use of spatiotemporal helixes. These helixes represent changes in both an object’s outline and its position over a time period of interest. A motivating example for the development of these helixes has been NASA’s Invasive Species Forecasting System. This program aims to identify geographic areas where invasive plant species are out-competing the native flora of the area. Satellite images of the area are collected and can be analyzed to see how the non-native plants have moved across the landscape. The ultimate aim of the project is to effectively predict future movements of invasive species, to aid in conservation practices. Non-indigenous invasive species are emerging as one of the most profound threats of natural disaster in this century. Non-indigenous plants, animals, and other organisms have been detected in the US at increasing rates in recent decades and are now present in numerous different ecosystems across the continent. A growing number of these species are becoming invasive as they suppress native species, thus changing native ecosystems. In many cases this is reflected by declines in the number and diversity of native species [7]. The invasive species threat is not affecting the United States alone. Globalization has greatly increased the international mobilization of non- indigenous invasive species through travel and agricultural industries. The impact of this phenomenon is profound: the direct cost to the American economy alone is estimated at $100-200 billion per year [7]. Compared to the damages inflicted by all other natural disasters, the impact if invasive species is undoubtedly immense. In the United States, key organizations, such as the USGS and NASA have been collecting large amounts of spatial data and remotely sensed imagery, and are currently developing dedicated infrastructure for delivering scientific information concerning invasive species through out the US. NASA has collected imagery from various sensors, such as QuickSCAT, Landsat 7, SeaWiFS (ocean color imagery) and high-resolution satellite imagery from commercially available sensors, such as IKONOS and QuickBird, and other private-sector satellites [7]. A key to effective and reliable detection and evaluation of invasive species will be the ability to reliably process and track changes in ecosystems over time. This should include key capabilities such as: a) Detection of objects in various imagery data sources, such as optical, multi and hyper-spectral data, SAR, Lidar and other available data sources. b) Tracking individual objects over time and space. c) Supporting multiple object handling, including amalgamation and divergence of objects. d) Fusing various data sources through the matching of detected objects. This includes detecting similar objects in data sources of different view angles, resolutions and temporal coverage. e) Predicting future positions of objects based on existing data. f) Effectively delivering helix-based spatio-temporal data, through metadata structures and standards.