Wildland Fire Emissions Information System for retrospective fire emissions mapping for North America

Transcription

1 Wildland Fire Emissions Information System for retrospective fire emissions mapping for North America Nancy HF French*, Don McKenzie, Tyler Erickson *Presented at the 2012 ACCENT-IGAC-GEIA Conference, June Toulouse, France

2 Fire & Carbon GFED: The global amount of carbon combusted through fires is about 2.0 Pg C year -1, or about 22% of global fossil fuel emissions. van der Werf et al Atmos. Chem. Phys. Discuss.,10: , CBM-CFS3: From 1990 to 2006 Canada s managed forest was a Carbon Sink on average, but it was a Carbon Source in years with large burned area. Kurz, W.A. et al PNAS 105:

3 Fire Emissions Project Purpose: Improve access to fire emissions model inputs and results for targeted users Provide best estimates of total carbon emissions and some emission components for retrospective fire emissions mapping Geospatially at 1km resolution At daily to annual temporal resolution Progress: Improved data & model for retrospective fire emissions estimation Development of a prototype web-based system (WFEIS) for US Regional-scale retrospective estimates of US fire emissions based on two burn area data sets - Supported by a 3-year grant from NASA Carbon Cycle Science Program & Applied Sciences Program - Additional funding under another NASA Applied Sciences - Actively seeking additional funding to continue development

4 WFEIS Project Team Nancy French, Ben Koziol, Tyler Erickson, Mike Billmire, Jessica McCarty Don McKenzie Roger Ottmar Ernesto Alvarado Bill de Groot Pacific Northwest Research Station, USDA Forest Service Great Lakes Forestry Centre, Canadian Forest Service Eric Kasischke Department of Geography, University of Maryland Research Associates & Interns: Liza Jenkins, Eric Keefauver, Arthur Endsley, Reid Sawtell, Susan Pritchard, Ron Kempker, Jef Cieslinski, Marlene Tyner, Christina Nolte, Peter Gamberg, Naomi Hamermesh 4

5 Other fire emissions models: -- CONSUME 3.0 (USFS) -- FOFEM 5.7 (USFS) -- CanFIRE -- Canadian FBP -- GFED (NASA) -- FINN (NOAA) Non-spatial, plot-level (Consume 3.0, FOFEM 5.7) Semi- or Pseudo-spatial (Canadian FBP, CanFIRE) Global, coarse-scale (GFED) Fire Emissions Modeling Area Burned - A Fuel Consumption Emissions Active fire estimates of burn area (FINN) Fuel Loading - B Combustion Factors - ß Many other efforts in place for emissions inventory/air quality applications (e.g. US-EPA) WFEIS uses moderate resolution data of burn area and biomass to make regional-scale fire emissions estimates Emission Factors - EF

6 WFEIS Estimation of Fire Emissions Location and day of fire Daily weather Fuel load (biomass) and fire behavior Consumption & Emissions model (Consume) Output: Spatial representation of emissions Fuel (vegetation) type (FCCS)

7 Burn Area Datasets Perimeters from Monitoring Trends in Burn Severity (MTBS) MODIS-derived Direct Broadcast Burn Area Product (MCD64A1) see Giglio et. al m spatial resolution Burn cells tagged by approximate burn date North America-wide for 2001 to present 7

8 Fuel Consumption and Emissions: CONSUME CONSUME estimates fuel consumption and emissions for prescribed and wildland fire. It imports fuelbed data directly from the FCCS, and can be used for all forest, shrub, and grassland types in North America. Low-intensity prescribed fire and high-intensity crown fire consume different proportions of each stratum in each combustion phase. Flash Fuels 1-hr Estimates combustible biomass of woody fuels in each of the three stages of combustion. Litter Pyrolysis zone Flame-Available Fuel Smolder-available Residual-available 100-hr Log Predicts fuel consumption, pollutant emissions, and heat release based on: fuel loadings fuel moisture and other environmental factors 8

9 Fuel Loading: Fuel Characteristic Classification System (FCCS) FCCS provides an comprehensive description of fuel layers. Compiled from scientific literature, fuels photo series, fuels inventories, and expert opinion, and represent fuel conditions at multiple scales, from single plots to 1-km cells or larger. 9

10 Fuels Mapping: Mapping standard FCCS fuelbeds Fuelbed Map Fuelbeds are mapped via crosswalks to satellite-derived vegetation and land cover, at scales from < 25 m (landscape applications) to >36 km (continental and global applications). USGS Landfire project has developed a crosswalk from 30-m scale existing vegetation maps For WFEIS, forest & rangeland fuels are mapped to 1-km includes fuel loadings by type

11 30-m resolution 234 fuelbeds 5 lost in aggregation 1-km resolution

12 Fuels Mapping: Filling in the Gaps with the Cropland Data Layer Problem: Original FCCS map does not include all land types with vegetation fuels; Missing types primarily are areas that are urban, bare, or agricultural; When fires occur in these area, CONSUME calculates their emissions as 0. Fires do often occur in agricultural areas. Solution: The Cropland Data Layer (CDL) contains detailed information about agricultural fuelbeds in many areas within the FCCS gaps. Using ERDAS Imagine Model Maker, we merge the CDL with the FCCS to create an integrated fuelbed map. Create new agricultural fuelbeds, reclassify CDL merge giving priority to FCCS in areas where the CDL is vague (e.g. Deciduous Forest) and giving priority to the CDL in all other locations.

13 Fuels Mapping: Filling in the Gaps with the Cropland Data Layer Agricultural areas missing data FCCS (30m resolution) CDL (30m resolution) Merged FCCS/CDL Layer (30m resolution)

14 Merged FCCS/CDL layer (30m resolution) 242 Total fuelbeds 212 original FCCS fuelbeds 15% of total CONUS area filled in with croplands New fuelbeds from CDL publically available; fuel loadings vary by crop type

15 15

16 WFEIS Components User Input Geospatial Emissions Modeling System Output Fire Extent & Timing FCCS Fuelbed Daily Weather/ Fuel Moisture GeoTIFF Select By: AOI/lat,long Fire Name Place Name Time Select Fire Perimeter: Landsat (MTBS) MODIS (MCD64A1) CONSUME KML SHP NetCDF Modify Pre-set Inputs (optional) Canopy Consumption Shrub Consumption WFEIS Back End TXT 16

17 WFEIS Web-accessible Framework The WFEIS website allows for two approaches for making fuel consumption and emissions estimates Method #1: WFEIS Emissions Calculator: Method #2: WFEIS responds to queries submitted via properly encoded URL requests: it implements a RESTful Web API.

18 WFEIS Framework Web Browser Access oregion=western/1000hr_fm=none/duff_fm=none/canopyperconsume =None/PercentBlack=50/combustion_stage=total/stratum=total/output_u nits=tonnes/emistype=carbon/map.kml?drng= , &ROI=StateProvince,usa-OR User Interface URL-formatted QUERY RESTful Interface HTTP Access CONSUME Backend Python Scripts Django Web Framework Spatial Database

19 WFEIS Output

20 WFEIS Output Formats KML Google Earth format ESRI Shapefile suitable for desktop GIS NetCDF suitable for atmospheric scientists GeoTIFF a georeferenced image Text Report a summary report

21 Aggregating Output WFEIS can be used to consider annual fire emissions across the Contiguous US batch API queries allow for computing a variety of scenarios This example shows results using the MODIS MCD64A1 burned area product Year Burned area (km 2 ) Total consumption (Tg) CONUS total area is approx 8,000,000 km 2 Consumption normalized (kg/m 2 ) Total CO 2 (Tg)

22 Aggregating Output Consumption & emissions for ecoregions of the Contiguous US example is for 2003 CONUS Ecoregion Northwestern Forested Mountains Burned area (km 2 ) Total fuel consumption (Tg) Fuel consumption normalized (kg/m 2 ) Total CO 2 (Tg) Mediterranean California Great Plains North American Deserts Tropical Wet Forests Temperate Sierras Eastern Temperate Forests Southern Semi-Arid Highlands Northern Forests Marine West Coast Forest

23 Comparison of Models Five fire events: Biscuit fire in southern Oregon Montreal Lake central Saskatchewan Boundary fire in interior Alaska -- San Diego County, California 2003 and 2007 Area Normalized Carbon Emissions (kg m -2 ) Fuel Loading (kg m Results: -2 ) -- Models generally agree (within 25% of each other) but vary due to model assumptions -- Vegetation fuel density, structure, and condition (fuel moisture) are important drivers of emissions variability -- Global-scale GFED modeled emissions are consistent with landscape/regional-scale estimates (SanDiego 07) (SanDiego 03) (Montreal Lake) (Boundary) French, N.H.F. et al. (2011), "Model comparisons for estimating carbon emissions from North American wildland fire," Journal of Geophysical Research, 116, G00K (Biscuit)

24 Effect of Burned Area on Carbon Emissions For the Biscuit fire case, the WFEIS model was run with two burn area maps. The burn area and total carbon emitted is higher with the Landsatderived map by about 16% Area normalized emissions were very similar, but vary due to use of daily weather information and the type of vegetation in the burn Landsat: 200,400 ha 3.10 kgc/m TgC MODIS: 170,000 ha 3.07 kgc/m TgC

25 Data Access Fuels maps available from USFS FERA lab Accessible soon from the NASA-DAAC at ORNL: Questions?