The abundance of information available from satellite, government reporting, and on-the-ground observation requires complex and challenging data integration approaches to unlock patterns and relationships between the physical drivers of wildfire, societal impacts, and response. In this talk we present three new data products being developed by the University of Colorado’s Earth Lab. The FIRED product transforms the MODIS burned area product for the United States into fire perimeters and corresponding statistics. The ICS-209-PLUS dataset transforms government incident status reports into a research grade dataset capturing daily snapshots and incident level summaries for large incidents. Finally, the Earth Lab Global Social Sensor (GSS) systematically collects and transforms public social media into measures of societal response. We will provide an overview of the datasets for each of these products and architecture for real-time curation.
Each of these products offer new opportunities for discovery, but in combination much greater insights are possible. We present initial findings from within our team looking at trends across these combined sources. We also provide an overview of our data curation strategy for large fire and how this strategy could be extended to other hazards such as extreme weather events in the future.
Results/Conclusions
The FIRED dataset contains perimeters for 51,871 fire events from January 2001 to May 2019 for the coterminous United States with a corresponding database of derived metrics that can be used to better understand U.S. fire regimes. The ICS-209-PLUS-WF dataset contains 120,825 Daily Situation Reports and 24,608 Incident Summaries for large fires from 1999 to 2014. These reports provide valuable information about fire characteristics, contributing factors, accruing costs, damages, response resources, levels of social disruption, and values at risk. Finally, we have social media data repository containing tweets for 57 large-scale catastrophic fires from 2010 to the present and a state-of-the-art classifier for distilling metrics of social response across events. Highlights using these unique datasets in combination will be presented in our talk.
Each of these products will be part of a cloud-based data lake with new data added in real-time. The algorithm used in the FIRED product can be applied to any remote sensor with finer temporal and spatial resolution to generate real-time estimates. The ICS-209 reports are available daily and will drive the GSS to capture societal response to large fire at scale. We will discuss the value this has for both researchers and real-time emergency response.