Vision
Shifting field science to continuous monitoring to get results faster.
The bottleneck has moved
Getting geospatial data has traditionally been slow. Often it involved fieldwork, which is difficult and time-consuming. Even with data analysis there were bottlenecks: cleaning data, accessing computing, and finding capable models. Field work will likely always be challenging, but the Fossett Lab strives to make it easier by streamlining data collection, analysis, and visualization. Satellite imagery, computing, and trained models are now easily accessible. What remains difficult is the slow work of putting them all together to answer scientific questions.
The Fossett Lab works on this bottleneck for scientific inquiry.
What we're building toward
We are working toward continuous field science instead of in one-off field campaigns.
- Analysis that doesn't stop
- Satellites continually deliver data and fly over field sites every few days. GPS base stations, seismographs, and water gages log data continuously. Drone flights can happen on short notice and repeatedly. We need to build analysis pipelines that keep up with the volume of continuous data, instead of rebuilding them after each data collection.
See Argus, our live Earth-observation monitor → - A queryable record
- We carefully curate data and metadata so that every piece of data can be compared against the corpus we already have. This way researchers are always working with the complete context of previous results, compounding our understanding.
Browse the data catalog → - AI as a collaborator
- Automated models and agents can flag observations worth a closer look and suggest where to look next, handing these over to researchers to focus on scientific interpretation.
- Reproducibility by default
- We use reproducible code and track data provenance, so that every result can be replicated. Data processing steps are both logged and verifiable, providing a complete chain of custody for scientific integrity.
How the pipelines work → - Rapid iteration
- We are working to improve the feedback loop between the field and the lab. We are developing tools to stream 3D data from the field to XR goggles worn in the lab, so that scientists in the lab with access to computation can rapidly iterate with scientists making observations in the field.
Visualization and XR → - Discovery as the criterion of success
- The only benchmark is whether we can more quickly arrive at rigorous and reproducible scientific results.
What it looks like in practice
In practice, this means continuous monitoring of areas of interest, drawing researchers' attention when something interesting happens. It means rapid response to events like fires, floods, and earthquakes. We integrate data from satellites, drones, and other sources, and try to seamlessly provide as much context from the field back to the lab using visualization technologies. The hope is to more quickly understand a river channel that has shifted, a hillslope that has eroded, or vegetation that has grown back after a fire. The infrastructure behind this is substantial, but the goal is to have it be nearly invisible so researchers can focus on science.
Grounded in real research today
The lab's field, processing, analysis, and data services keep this vision grounded in research happening now, and they are where new collaborations begin.