Fighting the Amazon forest fires with advanced analytics
A team of SAS employees recently participated in a data-for-good project focusing on forest fires in the Amazon. In conjunction with the Amazon Conservation Association (ACA), the team explored options to collect and analyze publicly available imagery and fire data to better understand the drivers for forest fires as well as to try and predict the future behavior of fires in different parts of the Amazon region.
As a result of this work, a SAS Visual Analytics dashboard has been created to monitor fire activity. The report describes the current state of active fires (current count, YTD) as well as fire hotspots and their locations on a map. The investigator can interact with the dashboard and filter by dates of interest or by specific geographical locations. A KPI indicator at the top also shows the current overall fire risk as determined by our analytical model.
Users are also able to perform ad-hoc clustering and segmentation using the built-in Cluster Analysis tool by selecting desired risk factors and exploring fire locations given their corresponding cluster details on a geographical map. In the example below, the system determined three different clusters given metrics Forest Carbon Flux, Distance to Roads and Elevation. As a result, a cluster such as #1 represents fire locations with low or negative CO2 emissions, a location close to a road, and diverse elevation levels. The right-hand geographical map shows all fire locations colored by their corresponding cluster-ID.
You can learn more about this project and what analysis has been done in the related blog post.
Enjoy!
