Western U.S. Disasters and Health & Air Quality: California Flames and Oregon Plains

The Concern
During the months of July and August 2012, the western United States experienced extreme drought conditions, gusty winds, and severe thunderstorms. These extreme weather conditions led to several severe wildfires, particularly in southern Oregon. Ignited by a lightning strike in Malheur County, Oregon, the Long Draw Fire charred through short grass and sagebrush as it spread. This fire became Oregon’s largest fire in 147 years, as it destroyed over 557,648 acres of land between July 8 and July 15, 2012 [1]. The Barry Point Fire in Lake County burned 93,071 acres of timber (grass and understory) after ignition (also by lightning strike) on August 6th, 2012 until containment on August 31, 2012 [1]. In total, wildfires during 2012 in Oregon burned 1.26 million acres of land [2].

As a result, the region suffered extensive damage to its socioeconomic and environmental systems, and the wildfires significantly degraded air quality and resulted in a haze for hundreds of miles. The Klamath Basin National Wildlife Refuge Complex of the U.S. Fish and Wildlife Service, due to its low elevation and close proximity to the wildfires, was particularly prone to the resulting air pollution.

Many states, including Oregon, utilize different products from NASA sensors and related missions to detect, analyze, and understand wildfire risk and mapping. Through discussions with the Oregon Department of Forestry the team learned that current fire management practices included Gradient Nearest Neighbor (GNN) modeling, a statistical framework that utilizes data from NASA’s Landsat Thematic Mapper (TM) and the Forest Inventory and Analysis program.

It became evident during discussions with the Oregon Department of Forestry and the Klamath Basin Wildlife Refuge Complex that there were needs for (a) assessment of risk prior to events and (b) monitoring of impacts as fire events unfold. To respond to these needs, partners were interested in incorporating of NASA’s Earth observations alongside existing techniques related to wildfire management. Therefore, in fall 2012, a team of four DEVELOP interns on the Western US Disasters team at NASA Langley Research Center began to implement techniques to address Oregon end user disaster planning and response requirements with satellite-based inputs.

The Science

In consultation with their partners, the team decided to focus their pre-event study areas on Malheur County, assessing risk prior to Long Draw Fire, and Deschutes County, which was known by a partner to be at high risk area for wildfires. The team chose the Long Draw Fire in Malheur County, and the Barry Point Fire in Lake County to demonstrate monitoring of impacts for ongoing wildfire events. Analysis of these fires included data from before the ignition, mid-fire, and after the extinguishing of the fires to show the full lifecycle of the wildfires. Except where noted otherwise, the DEVELOP intern team conducted their analyses using ArcGIS 10.0.

Pre-Fire Risk Assessment

For pre-event risk assessment, the DEVELOP team prepared fire risk maps, created using a multi-criteria evaluation (MCE) method incorporating multiple land cover components to determine areas of high fire risk (Figure 1). For this study three variables were chosen to include in the fire risk map: proximity to roads, slope of the region, and fuel type. The MCE utilized the fuzzy logic method to describe geographic components that are inherently continuous in nature. This method also tries to correct the inherent error added when crisp lines are formed to describe a seamless geographic phenomenon. This method gives the end user the capability to tailor the fire risk map accordingly to better suit what is biophysically reasonable and more easily change the map output to a fire manager’s discretion.

Figure 1: Malheur County Fire Risk Map (top) showing highest risk in dark red, the Long Draw Fire perimeter is shown outlined in black. Deschutes County Fire Risk Map (bottom) showing areas at highest risk in dark red.

Monitoring Fire Impacts

To demonstrate ongoing wildfire event monitoring, the DEVELOP team produced vegetation health, burn severity and air quality products. Burn severity and vegetation maps were created for Long Draw Fire and the Barry Point Fire.

The team employed the use of the new Suomi National Polar-orbiting Partnership (NPP) sensor, Visible Infrared Imaging Radiometer Suite (VIIRS), which has similar capabilities to the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor, to analyze vegetation health for drought severity. The imagery bands of the VIIRS sensor have 400 meter resolution, giving a daily overall analysis of the western region of the United States in one swath. A Normalized Difference Vegetation Index (NDVI) was used to measure the amount of greenness on the land, by detecting the presence of chlorophyll and leaf pigments. An NDVI for Malheur County and Lake County was produced using VIIRS imagery for July and September 2012 to show the change in the land pre- and post-fire. The imagery from VIIRS indicated a considerable change over the time period, demonstrating the effectiveness of the new sensor to analyze drought and landscape changes (Figure 2).

Figure 2: VIIRS NDVI images derived from imagery bands I1 and I2 show the impact of the Barry Point fire on vegetation cover in Lake County. The top image is from before the fire on 7/11/2012, and the bottom image is from after the fire on 9/5/2012. The Barry Point fire perimeter is outlined in black.

Using Landsat 7 Enhanced Thematic Mapper Plus (ETM+), the more specific burn dynamics of the individual study areas were analyzed. Landsat 7 gives a much more detailed analysis than VIIRS, with 30 meter resolution and coverage every 16 days. Band combinations (burn scars) and burn ratios (dNBR, RdNBR) were used to show the impact of the burn severity at a more detailed level.

For the burn scar maps, pre- and post-fire images were acquired for both study areas, and three bands were stacked together. This combination demonstrates the health of the vegetation and damaged landscape, resulting in a burn scar assessment of the land. To show the burn scar using VIIRS, the RGB band combination was set to M11, I2, M4. For Landsat 7, the RGB band combination was set to 7, 4, 2.

Landsat 7 ETM+ was used to create a more detailed and reliable picture of the impact of the fire. The Normalized Burn Ratio (NBR) was computed for the individual counties of Malheur and Lake at three points in the fires, pre-, mid-, and post-fire. Two Difference Normalized Burn Ratios (dNBR) per fire were calculated, using pre-fire and mid-fire NBRs and then pre-fire and post-fire NBRs for both fires. This shows the difference in the burn of the land at the beginning of the fire and throughout the fire by displaying a quantitative measure of environmental change.

A more accurate way to show burn severity is with the Relative difference Normalized Burn Ratio (RdNBR) for both pre and post-fire dates. The RdNBR allows for better estimates of burn severity by removing the correlation of dNBR and pre-fire biomass. The dNBRs and RdNBRs also qualitatively display areas of potential enhanced regrowth, which is helpful to see which areas are undergoing succession after the fire has been extinguished. However, the RdNBRs were found to be more accurate and sensitive for the study areas because of the high density of vegetation present. The RdNBRs showed more variation throughout the land as it was more sensitive to changes in vegetation, whereas the dNBR mostly classified the study area as highly severe burn (Figure 3) [3].

Figure 3. Mid-fire RdNBR for the Barry Point Fire in Lake County (left) and post-fire RdNBR showing the severity of the fire (right) giving more accurate results by removing the correlation between the dNBR and the pre-fire mass.

In addition to land cover and burn analyses, negative impacts on air quality as a result of the wildfires were studied. The Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) sensor on the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) satellite measures vertical profiles of aerosols and clouds. This data was downloaded and the swaths were viewed to measure the smoke plume extent, then was converted to a KMZ file and inputted into Google Earth with the NOAA Hybrid Single Particle Langrangian Integrated Trajectory (HYSPLIT) model to give a vertical and horizontal depiction of the smoke plume direction (Figure 4).

Figure 4: August 23, 2012 NAM12 meteorological data in NOAA’s HYSPLIT model, overlaid with a CALIPSO curtain, and viewed simultaneously in Google Earth.

NAM12 Meteorological data, due to its high spatial resolution, was utilized for the NOAA HYSPLIT model. Forward trajectories were run for August 15, 16, 17 and 23 because these days had the highest burn severity and average of acres burned. The Oregon Department of Environmental Quality identified August 27 as the date with the worst air quality during the month of August 2012. Therefore, the team downloaded NAM12 data for August 27 using the latitude and longitude of an air quality monitoring station at Peterson Elementary School in Klamath Falls, Oregon and ran backward trajectories from that date.

The Benefit

The fire risk maps and imagery were used to classify the effects of fires on the ecosystems by assessing the magnitude of ecological damage caused. Through study of the associated health and air quality impacts of the wildfires, an analysis of smoke plume and particle distribution was provided to the partners. These findings improve the understanding of the impact of wildfires in the South Central Region of Oregon, and assist project partners in assessing areas of drought and high risk for future wildfires. Detailed tutorials were created and transferred to the Oregon Department of Forestry and the Klamath Basin National Wildlife Refuge Complex of the US Fish and Wildlife Service, allowing for replication of the methodologies in the future.

This project employed and enhanced the methodologies first described by the NC Disasters Summer 2012 DEVELOP Team. The partnership between DEVELOP and the Oregon Department of Forestry has continued into a second project, with a team currently studying more recent fires in Oregon and their impacts on the surrounding land and air quality.

References

[1] InciWeb: Incident Information Website. (2012). Oregon. Retrieved from <>.

[2] Wildfire, Oregon Department of Forestry. (2012, October 23). Wildfire season ends statewide. Retrieved from http://wildfireoregondeptofforestry.blogspot.com/2012/10/wildfire-season-ends-statewide-on-oct-22.html

[3] Key, Carl H. and Nathan C. Benson. (2006). Landscape Assessment (LA): Sampling and Analysis Methods (USDA Forest Service Gen. Tech. Rep. RMRS-GTR-164-CD). Retrieved from http://www.fs.fed.us/rm/pubs/rmrs_gtr164/rmrs_gtr164_13_land_assess.pdf.

Acknowledgements:

Christina Welch, Co-Lead Western US Disasters Fall 2012 DEVELOP Team
Jessie Larson, Co-Lead Western US Disasters Fall 2012 DEVELOP Team
NC Disasters Summer 2012 DEVELOP Team

Editor’s Note: The DEVELOP National Program is a capacity building internship sponsored by NASA’s Applied Sciences Program that provides interns the opportunity to learn about NASA Earth Science and the practical applications of Earth observations.