B31C-0297
Trace Gas Emissions From Tropical North Australian Savanna Fires
We present measurements of atmospheric trace gases within smoke plumes from tropical North Australian savanna fires. The remote sensing measurements are made from Darwin (12.4°S, 130.9°E) using Fourier Transform spectroscopy with the sun as the source. From these infrared spectra column amounts of carbon monoxide (CO), formaldehyde (H2CO), acetylene (C2H2), ethane (C2H6) and hydrogen cyanide (HCN) have been determined. Literature esimates of emission factors for CO are then used to infer emission factors for these other gases.
B31C-0298
Modeling Fire Emissions from Multiple Land Use Transitions in Southern Amazonia
Fires for deforestation and other land cover changes in southern Amazonia are an uncertain but significant source of carbon emissions to the atmosphere. Recent expansion of mechanized cropland in the region has increased the rates, clearing sizes, and combustion completeness of forest and Cerrado conversion compared to previous deforestation for cattle ranching. To more accurately quantify the influence of agricultural intensification on carbon emissions, we developed a high-resolution (250 m) model of DEforestation CArbon Fluxes (DECAF). DECAF estimates variations in forest and Cerrado biomass based on time series of MODIS NDVI and explicitly tracks the duration and combustion completeness of new deforestation as a function of post-clearing vegetation phenology and MODIS-based fire frequency. In our model runs for the Brazilian state of Mato Grosso, we quantified the contribution of fires for deforestation, conversion of pasture and Cerrado to mechanized cropland, and pasture maintenance to total fire emissions under low, middle, and high emissions scenarios. During 2001-2005, carbon losses from all types of deforestation were 48-82 Tg per year (mean = 67 Tg C), representing approximately 74% of annual fire emissions in the study region. Cropland expansion in non-forest areas contributed 19% of estimated fire emissions, while maintenance fires in pasture and Cerrado land cover types averaged 7% of all fire emissions during 2001-2005. Conversion of forest to other land uses often takes more than one year, and part of the biomass that was not burned in the dry season following deforestation burned in consecutive years. This led to a partial decoupling of annual deforestation rates and fire emissions, and lowered interannual variability in fire emissions. In total, DECAF-based emissions for Mato Grosso represent 1/3 of estimated fire emissions for all of southern hemisphere South America during this period. Our results demonstrate how DECAF can be used to model deforestation fire emissions at relatively high spatial and temporal resolutions. Detailed model output is suitable for policy applications concerned with annual emissions estimates distributed among post-clearing land uses. Our results suggest that effective policies to reduce fire emissions in Mato Grosso would promote intensification of already-cleared land for cropland conversion and would target large (>25 ha) rather than small clearings.
B31C-0299
Development of a Wildland Fire Smoke Marker Emissions Map for the Contiguous United States
Several chemical species, including levoglucosan, mannosan, galactosan and K+, are characteristic components of smoke produced from biomass combustion. In the Fire Lab at Missoula Experiment (FLAME), over 30 different wildland fuels were burned, and the smoke produced was analyzed for physical, chemical and optical properties. High volume filter samples were collected and analyzed for sugars and anhydrosugars using high performance anion exchange chromatography with pulsed amperometric detection. Total carbon concentrations were analyzed on a Sunset organic/elemental carbon analyzer. Major ion concentrations were quantified using ion chromatography. Several patterns emerged from statistical analysis of the smoke marker species, particularly that different vegetation classes (e.g. leaves, needles, branches and grasses) produced different marker to carbon ratios. Because vegetation classes are spatially distributed throughout the country, this information can be translated into geographic information. For example, wildfires in the softwood forests of the Western United States (U.S.) might burn mostly branches and needles, while prescribed fires of the plains in the Midwest might burn mostly grasses. Using a wildland fuel type map from the United States Forest Service, several geographic biomes have been identified including softwood forests, hardwood forests, scrub brush, grasslands, and floral shrub lands. Source profiles for smoke markers based on the measurements made during FLAME were assigned to each vegetation class. Using the fuel type map and fuel specific source profiles, a 12 km gridded map of fire emission smoke profiles was created for the U.S. Each cell of the grid has been assigned a percentage of each vegetation class likely to be consumed in an average fire and the source profile for this mix of vegetation consumed. The end product is a map of smoke marker source profiles. Provided the location of the fires impacting a receptor can be estimated, these geographically distributed source profiles can be used in receptor models to better apportion primary particulate carbonaceous material to wild and prescribed fire.
B31C-0300
Carbon and CO Emissions From the 2004 Alaskan Fires
The 2004 fire season was the largest on record for Alaska, affecting some 2.7 million ha. To estimate emissions from these fires, we adapted the wildland fire emissions model developed by the Canadian Forest Service, BORFIRE. This model uses fire weather indices derived from data collected at weather stations to estimate fuel consumption. We modified the BORFIRE model to estimate carbon emissions from the burning of organic soils in black spruce forests and lowland areas based on field measurements collected over the past three years. Vegetation categories and burned area estimates included in the model were mapped using information products derived from Landsat TM/ETM+ data and MODIS data. Seasonal timing of fires was determined from MODIS hotspot data. When Landsat information products were used, carbon and CO emissions were significantly higher than estimated by our previous model (BWEM) and other researchers. Estimates generated using MODIS information products were lower than those estimated using Landsat information products because the area of spruce forest in the MODIS products were lower.
B31C-0301
Monitoring Global Biomass Burning CO emissions: MOPITT, AIRS, and Ground-based Spectrometers.
CO has several natural and human-induced sources. They are comparable in strength, but biomass burning (BB) is the only one that experiences significant interannual and seasonal variations. Thus, estimates of CO emission anomalies from BB using global satellite data are relatively straightforward. The importance of this monitoring is connected with long-term increases in global BB that has been speculated, but not yet proven. A comparison of global data from different orbital instruments in combination with their validation vs ground- based instruments provides a fast and direct way for prompt estimation of BB variations. This report presents analysis of CO global Level 3 measurements retrieved from satellite observations by MOPITT and AIRS through November 2008. Global CO burden anomalies are readily recalculated into anomalies of CO BB emissions assuming stable [OH]. Regional CO burden is a good indicator for BB variation in a region. Pyrogenic CO often is detected by ground-based instruments as well. A zenith-viewing Atmospheric Emission Radiance Interferometer (AERI) measuring mid-IR spectra supplies valuable information about pollution in the lower two km of atmosphere. Examples of AERI data for Oklahoma and Maryland will be presented.
B31C-0302
Use and ground validation of MISR satellite data to constrain injection heights of emissions from forest fires in the northwestern United States
Emissions from forest fires can have a significant effect on the concentration of atmospheric pollutants downwind of the fires. In light of the increased frequency and severity of forest fires as a consequence global climate change, being able to accurately model the transport of smoke from these fires, from the regional to the global scale, is especially important. However, modeling the vertical transport of smoke from forest fires has proved difficult. In order to assess the accuracy of plume rise models, correlative data are needed. For atmospheric models with coarse resolution in which fires are treated as sub-grid scale events, semi-empirical methods of calculating the injection heights of emissions are often necessary. Here, we use observed smoke plume heights from the NASA MISR (Multi-angle Imaging SpectroRadiometer) satellite instrument to assess the accuracy of the predicted injection heights of forest fires emissions used in the AIRPACT regional forecast system for the northwestern United States. These are modeled using Forest Service fire incidence reports and a Briggs type plume model. The MISR heights are retrieved using the MISR Interactive Explorer (MINX). There is no systematic program to verify plume heights from MISR at present. We also present a program, to be implemented during the fire season of 2009, to verify plume heights from the MISR instrument using field portable video cameras with position, time and angular information encoded on the video data stream.
B31C-0303
Quantifying the vertical redistribution of atmospheric mass due to pyroconvection using satellite observations of carbon monoxide
Intense surface heating associated with forest fires generates rapid vertical mixing that can reach as high as the lower stratosphere, so called pyroconvection. The associated vertical transport of trace gases and particles has implications for the atmospheric chemistry and transport of these pollutants. Without accurate knowledge of this vertical mixing, surface flux inversion methods using atmospheric concentration measurements will lead to errors in trace gas budget calculations. Pyroconvection occurs on spatial scales much smaller than currently described by state-of-the-art global transport models, and global modelling the associated fluid dynamics is computational prohibitive, limiting our ability to study its widespread impact on distributions of trace gases. We present the first comparative study of the vertical mass redistribution of trace gases due to boreal and tropical pyroconvection, using a new methodology based on optimal estimation and constrained by satellite observations of carbon monoxide (CO) from NASA Aura MLS and TES instruments. We focus our study over June to October 2006, when fire intensity peaks, and evaluate the approach using NASA Aura OMI measurements of aerosol index.
B31C-0304
Comparing Measures of Subpixel Fire Sizes and Temperatures From ASTER and MODIS
Some of the most widely-used datasets for monitoring fires and their effects come from the Moderate- Resolution Imaging Spectroradiometer (MODIS) sensors aboard NASA's Terra and Aqua satellites, which can cover the entire Earth multiple times each day and provide data on active fires at nominal resolutions of 1 km, during daytime and nighttime. The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), which is aboard Terra but not Aqua, has a shortwave infrared subsystem that can also acquire images of active fires during daytime and nighttime with global coverage, providing data on fire properties at a nominal resolution of 30 meters. While ASTER can offer a finer spatial resolution than MODIS can, MODIS fire data products are more useful in many applications because MODIS has a much wider swath, and thus images a greater number of fires than ASTER can. However, fire pixels from ASTER and MODIS are actually mixed pixels that can contain flaming, smoldering, and non-burning components, and existing data products from these sensors provide little information about the sizes or temperatures of their subpixel components. This study uses multiple endmember spectral mixture analysis (MESMA) to estimate subpixel fire sizes and temperatures from ASTER and MODIS images of active fires, demonstrating new methods that can provide information not available from other sources. This study also demonstrates how ASTER data can be used to validate MESMA estimates of subpixel fire properties from MODIS, and how these MESMA estimates can overcome some limitations of existing methods for characterizing fire intensities from remotely sensed data, such as estimating the fire radiative power (FRP). Prior to this work, few studies, if any, had used MESMA for estimating fire sizes and temperatures from ASTER or MODIS, or compared MESMA estimates of fire properties to higher-resolution data or other measures of fire properties like FRP. Because a fire's size and its temperature exert strong influences on its trace gas and aerosol emissions, ecological effects, and spreading rates, these MESMA estimates from ASTER and MODIS could contribute useful new information towards monitoring, forecasting, and understanding the behavior and impacts of many fires worldwide.
B31C-0305
Estimating Fire Radiative Energy From MODIS
An alternative approach to biomass burning emission estimates has emerged from remote sensing science in which the fire intensity measured during combustion serves as a proxy for emissions released. The rate of energy emitted is referred to as the fire radiative power, or FRP. Integrating FRP over the lifespan of the fire event provides the total fire radiative energy (FRE) released, which in turn is directly proportional to the total fire emissions. We present an approach to estimate FRE from the Moderate Resolution Imaging Spectrometer (MODIS) 8-day, 0.5° climate modeling grid (CMG) FRP observations. An important characteristic of the radiative energy emitted from fires is the temporal trajectory which describes how the fire intensity behaves over the span of seconds, minutes, hours, days, and even seasons. To characterize the fire diurnal cycle we first examined FRP retrievals from Africa using the European Space Agency's geostationary Meteosat SEVIRI sensor. SEVIRI's temporal sampling resolution (15-minute) provides the capability to integrate FRP observations and calculate FRE. We also included diurnal cycle probability density functions (PDFs) from the Tropical Rainfall Measuring Mission (TRMM) sensor to expand our investigation area beyond Africa. A modified Gaussian function was shown to provide a simple and accurate representation of the observed diurnal cycles. Aqua and Terra MODIS 30-day gridded FRP observations were then used to parameterize the Gaussian function based on the ratio between the morning (Terra) and afternoon (Aqua) MODIS FRP values. Our contention is that the variation in the Terra/Aqua ratio can serve as a proxy for the diurnal cycle trajectory. The relationship between the Terra/Aqua ratio and Gaussian function parameters was established for multiple regions within the tropics. In addition, we supplemented the diurnal cycle parameterization with MODIS FRP retrievals at high latitudes. High latitude FRP retrievals were chosen because of Terra and Aqua's polar orbit and therefore greater frequency of observations made by each sensor. The greater frequency of FRP retrievals allowed us to characterize the temporal trajectory for higher latitude fires, which tend to have a distinctly different diurnal burning cycle than tropical fires. A comparison of estimated FRE with observed FRE from SEVIRI demonstrates the accuracy of our approach. For example, within a southern African region (1900 km2), which emitted a total FRE of 4.84e+06 MJ in July 2004, results from 49 2.5°x2.5° test cells showed that our calculation of FRE was only slightly overestimated (bias = 7%) with little error (RMSE = 3.05e+04 MJ). Comparable results were found in northern African and within other tropical regions when compared with TRMM PDFs.
B31C-0306
Using MODIS FRP Values to Estimate Forest Fire PM 2.5 Emissions
Accurate estimates of PM 2.5 (particulate matter less than 2.5 micrometers in diameter) emissions from wild fires are important for determining regional radiative budgets, investigation of regional atmospheric chemistry, and assessment of health risks. Several different modeling systems are already in use to determine these emissions. The BlueSky system was developed by the U.S. Forest Service to estimate wildfire emissions for dispersion modeling purposes. Recently, this system has been upgraded with a module called SMARTFIRE to incorporate both ground-based fire reports and MODIS (Moderate Resolution Imaging Spectroradiometer) satellite data. This combination results in an improved estimate of the fire location and size as a basis for estimating emissions within a new BLUESKY framework. MODIS sensor data, collected onboard two satellites, provide data that can be used to estimate FRP (Fire Radiative Power) which is a measure of fire intensity. In this project, MODIS FRP data are used with empirical relationships between FRP and PM2.5 emissions to estimate PM2.5 emissions for selected fires in the Pacific Northwest. These estimates are compared to BLUESKY/SMARTFIRE emissions. This study describes how applying FRP/PM2.5 relationships can provide an independent evaluation of current wildfire emission estimates and to assess the feasibility of the FRP approach for providing emissions for regional air quality modeling.
B31C-0307
The Global Fire Emissions Database (GFED3) Global Burned Area Data Set
We discuss major enhancements to the burned area component of the Global Fire Emissions Database (GFED3) over previous versions (GFED1 and GFED2), which now provides global, monthly burned area estimates at 0.5-degree spatial resolution for the time period 1997-2008. Estimates are produced by calibrating Terra MODIS active fire data with 500-m MODIS burned area maps via geographically weighted regression. Cross-calibration with fire observations from the Tropical Rainfall Measuring Mission Visible and Infrared Scanner (VIRS) and the Along-Track Scanning Radiometer (ATSR) allows the data set to be extended further back in time. We then discuss the spatially-explicit uncertainty estimates accompanying our data set, and the use of these estimates within atmospheric and biogeochemical models. We compare our GFED3 burned area estimates with other recent global burned area data sets, including GFED2, L3JRC, and GLOBCARBON. We quantify areas and time periods in which the different products diverge, and conclude with explanations for some of the discrepancies.
B31C-0308
Validation of Remotely Sensed Fire Detections Using Ground and Aircraft Reports
A daily fire analysis for North America is prepared by NOAA/NESDIS utilizing seven NOAA and NASA
geostationary and polar orbiting satellites. The analysis incorporates automated fire detections into an
analyst quality control procedure. Limited validation on the analysis has been performed to date. One effort
utilized high resolution ASTER sensor data on the NASA Terra spacecraft and another used ground reports
from Montana and Idaho. Owing to inherent limitations in both approaches further validation has been
performed. The current study expands on the ground report method. Daily ground reports (authorizations)
have been obtained from the Florida Division of Forestry. Additional data from Montana/Idaho has been
obtained as well as a small set of ground reports from Washington state. Aircraft data was also obtained for
agricultural burns in Manitoba. The use of these additional data sets has expanded the validation to include a
greater variety of land types/uses as well as geographic locations that have varying geostationary sensor
viewing angles. The results are consistent with the previous ground based validation that only included
Montana/Idaho and indicate a probability of detection (POD) of 20-25%. This low POD is at least partly due
to cloud cover obscuring reported fires during active burning and also due to the small size of many fires.
The analyst quality controlled product for each data set had a higher POD than for the automated detections
only. The nature of the data sets precludes the determination of commission errors. POD information
combined with fire size estimates are important considerations for emission modeling. These results suggest
that emission estimates based on remotely sensed fire detections may be too low.
http://www.ssd.noaa.gov/PS/FIRE/hms.html