Fire activity across the United States has increased dramatically across the western United States in recent decades. For example, California experienced a fivefold increase in annual burned area from 1972-2018. Drivers of these trends include warming temperatures and drought, as well as decades of fire suppression that have allowed fuel to accumulate, leading to a “fire deficit.” Whatever the drivers, the scientific consensus is that anthropogenic climate change will bring warmer and drier conditions to the West, providing more fuel for fires to consume and further enhancing fire activity. These trends are concerning in part because emerging evidence shows that smoke from fires, like other airborne particles, has a deleterious effect on human health. Improved monitoring of the magnitude of the smoke exposure currently experienced by populations across the western United States would help policymakers and stakeholders plan for present-day wildfires and pave the way toward strategies for future wildfires.
Here we propose to improve understanding and update the monitoring of smoke hazards resulting from wildfire activity in the western United States. By combining long-term climate records, observations from satellites, new fire emissions inventories, and models of land cover and atmospheric chemistry, we will address the following questions:
1. Can we quantify the impact of anthropogenic climate change on current smoke exposure in the West?
2. Which regions are especially vulnerable to the long-term fire deficit and would benefit the most from prudent land management?
3. Which fire-prone regions, in addition to those identified in #2, have the greatest potential to expose large populations to smoke pollution?
4. Using a machine learning approach, can we update the monitoring of smoke plumes in GOES satellite data?
Our proposed research would lead to two monitoring products. First, we would construct a smoke risk index, identifying those regions where potential fires could lead to the greatest smoke exposure among populations downwind, allowing government agencies to more wisely deploy scarce resources. Second, we would devise a machine-learning algorithm to streamline the process by which smoke plumes are detected in satellite data. The current method to detect such plumes involves human analysts, but machine learning promises to make that method more efficient, accurate, and reliable.
The project targets the NOAA MAPP call for New Climate Monitoring Approaches and Products for Areas of Climate Risk. It promises to develop climate modeling capabilities and applications relevant to decision-makers based on climate analyses, predictions, and projections. Throughout our project, we will work closely with NOAA scientists Heath Hockenberry (NIFC) and John Simko and Wilfrid Schroeder (NESDIS). We will also engage Christine Wiedinmyer (CIRES), who is lead developer of the Fire Inventory from NCAR (FINNv2).