Stream Scouts

Stream Scouts is a smart e-client system that will facilitate ad-hoc classification and validation of hydraulic features on very recent, high-resolution aerial photography of streams and rivers. It will be geared toward aspects relevant to the protection of both human uses (i.e., drinking water quality, hydropower, flood protection) and ecological status. Specifically, we will extend our present setup of small hand held Pocket PC devices used to take field measurements to form ad-hoc wireless networks and communicate with Tablet PC servers located nearby on a boat or on shore. A heavyweight desktop server in a distant location will host the river-habitat database, run complex aquatic habitat simulation models and management applications over the internet, and facilitate data exchange with the Tablet PC. This system will support semi real-time simulations of aquatic habitats and creation of a self-learning database, enabling the development of algorithms for classification of critical habitat features from aerial photographs. We see this project as a natural extension of many issues encountered in the TerraServer project, extending methods from computational geography to the domain of computational hydrology and computational ecology, through enabling the annotations necessary for simulating the dynamics of a watershed. Such a system is in high demand among environmental scientists and resource managers.

Present Status: We have made substantial steps towards achieving the goals of the Stream Scout project. The data pipeline has been completed and is in the process of being tested. The process begins with aerial photography. We have improved the processing of images using an AIMS sensor, which has a mulispectral camera pulse laser, which allows for more accurate and specific geospatial processing. To increase the speed of data flow, we developed a script for ArcPAD that allows Pocket PC's to send data to the Tablet PC in the field. Using omni-directional antennaes we have increased the Ad-hoc network range to around 200m. We are currently researching ways of further increasing the communication range. The Tablet PC now runs image classfications in real-time as soon as it recieves data from the Pocket PC's. The Geo-database has been populated with data. The Geo-database has also been restructured for ease of use, and has been made more secure. Habitat Analysis is no longer hampered by processor speed and is more automated.

This project is sponsored by Microsoft Research.