Injection height for biomass burning Injection height for biomass burning emissions from boreal forest firesemissions from boreal forest fires
Fok-Yan LeungFok-Yan Leung
April 12, 2007.April 12, 2007.Harvard UniversityHarvard University
Special thanks to:Special thanks to: Jennifer Logan, Rokjin Park, and Dominic Spracklen (Harvard)Jennifer Logan, Rokjin Park, and Dominic Spracklen (Harvard)
Edward Hyer and Eric Kasischke (UMD)Edward Hyer and Eric Kasischke (UMD)Leonid YurganovLeonid Yurganov
David Diner, Dominic Mazzoni, David Nelson, and Ralph Kahn (NASA/JPL)David Diner, Dominic Mazzoni, David Nelson, and Ralph Kahn (NASA/JPL)
Funding from the NSF and EPAFunding from the NSF and EPA
On opponents of greenhouse gas abatement: "Your political base will melt away as surely as the polar ice caps... You will become a political penguin on a smaller and smaller ice floe that is drifting out to sea. Goodbye, my little friend! That's what's going to happen." – Arnold Schwarzenegger
We began by looking at emissions estimates We began by looking at emissions estimates for 1998 boreal fires, which vary significantly.for 1998 boreal fires, which vary significantly.
KAS05 emissions2x as large as KAJ02emissions
KAJ02emissions
“Interannual” emissions Derived using TOMS-AI(for 1998)
“Climatological”emissions
Discrepancies between KAS05 and KAJ02 stem primarily from differences in assumptions about belowground burning
Comparison of surface and column data from 1998 Comparison of surface and column data from 1998 with results of GEOS-Chem simulationswith results of GEOS-Chem simulations
KAJ02emissions
baselineemissions
KAJ02emissions data - 1998 data - average
KAS05emissions
During intense boreal fires, intense heat can result in During intense boreal fires, intense heat can result in lofting of emissions well above the boundary layerlofting of emissions well above the boundary layer
In GEOS-Chem, all biomass burning emissions are injected in boundary In GEOS-Chem, all biomass burning emissions are injected in boundary layer by default.layer by default.
Base initial parameterization on assumptions that:Base initial parameterization on assumptions that: The majority of emissions from crown fires are more likely to be lofted The majority of emissions from crown fires are more likely to be lofted
into the free troposphere.into the free troposphere. Crown fires are prevalent during highest burning monthsCrown fires are prevalent during highest burning months
Based on the above, and the work of Kasischke, (2005), we injected:Based on the above, and the work of Kasischke, (2005), we injected: 40% of all emissions in boundary layer40% of all emissions in boundary layer 60% into free troposphere60% into free troposphere
Putting large fraction of emissions in free troposphere reconciled Putting large fraction of emissions in free troposphere reconciled model results with both surface and column data model results with both surface and column data
For both surface and column data (anomaly data): KAS05 seems to For both surface and column data (anomaly data): KAS05 seems to perform better in capturing the CO behavior using parameterizationperform better in capturing the CO behavior using parameterization
KAS05 – 60% of emissions in FT, 40% in BL
Anomaly = 1998 - baseline KAS05 – 100% ofEmissions in BL
Conclusions from study of 1998 study:Conclusions from study of 1998 study:
Injection of biomass burning emissions in the free troposphere are Injection of biomass burning emissions in the free troposphere are necessary to reconcile ground and column datanecessary to reconcile ground and column data
Injection of biomass burning emissions in free troposphere results in Injection of biomass burning emissions in free troposphere results in higher tropospheric ozone throughout the northern hemisphere due to higher tropospheric ozone throughout the northern hemisphere due to longer sequestration of NOlonger sequestration of NOxx by PAN formation. by PAN formation.
Preliminary studies suggest that model results are not particularly Preliminary studies suggest that model results are not particularly sensitive to the exact fractional split of emissions (Turquety et al., [2007], sensitive to the exact fractional split of emissions (Turquety et al., [2007], and unpublished work)and unpublished work)
We were motivated to move beyond the “sensitivity analysis” level, and to We were motivated to move beyond the “sensitivity analysis” level, and to our ongoing study of plume heights using the MISR instrumentour ongoing study of plume heights using the MISR instrument
Using the MISR InstrumentUsing the MISR Instrument Satellite instrument aboard Satellite instrument aboard
TERRA platformTERRA platform 4-5 days repeat time at high 4-5 days repeat time at high
latitudes latitudes Visual and infrared cameras at 9 Visual and infrared cameras at 9
different angles allows heights of different angles allows heights of clouds, smoke plumes, terrain, clouds, smoke plumes, terrain, etc... to be calculated at etc... to be calculated at 0.5 km 0.5 km vertical resolutionvertical resolution
can distinguish smoke from clouds can distinguish smoke from clouds or other aerosolsor other aerosols
Left: August 17, 2002NW corner =(73 ˚N,130 ˚E) SW corner =(60 ˚N,130 ˚E) 0 5 10
km
We look at discrete plumes from fires. Algorithm of Mazzoni and Nelson:We look at discrete plumes from fires. Algorithm of Mazzoni and Nelson: Detects plumes by trained plume shape recognition algorithmDetects plumes by trained plume shape recognition algorithm Uses MODIS hotspots to narrow down number of plumesUses MODIS hotspots to narrow down number of plumes Determines the maximum plume height Determines the maximum plume height
Using the algorithm, 66 discrete plumes were found in Alaska and Northern Canada Using the algorithm, 66 discrete plumes were found in Alaska and Northern Canada during summer of 2002during summer of 2002
Example of algorithm at work…Example of algorithm at work…
Kahn et al, [2006] observed clear relationship between atmospheric Kahn et al, [2006] observed clear relationship between atmospheric stability and observed plume heights.stability and observed plume heights.
We compare the stability profiles calculated using the coarser GEOS4 We compare the stability profiles calculated using the coarser GEOS4 data with the finer resolution BRAMS data, at the 66 sitesdata with the finer resolution BRAMS data, at the 66 sites
GEOS4GEOS4 11°°x1.25x1.25°° horizontal resolution horizontal resolution 30 vertical levels in 30 vertical levels in
tropospheretroposphere Reanalysis product.Reanalysis product. Pressure is instantaneous Pressure is instantaneous
pressurepressure Temperature is 6 hour average Temperature is 6 hour average
BRAMS (courtesy Marcos Longo)BRAMS (courtesy Marcos Longo) 45 km horizontal resolution45 km horizontal resolution 150 km vertical resolution in 150 km vertical resolution in
tropospheretroposphere Boundary and initial conditions Boundary and initial conditions
use GFS analysisuse GFS analysis Mesoscale model “nudged” by Mesoscale model “nudged” by
GFSGFS Pressure and temperature are Pressure and temperature are
instantaneousinstantaneous
S ddz
Stability profiles: BRAMS (courtesy Marcos Longo) Stability profiles: BRAMS (courtesy Marcos Longo) and GEOS4: “Neutral” profiles and profiles with and GEOS4: “Neutral” profiles and profiles with
regions of high stability regions of high stability
In general, same vertical structural characteristics in In general, same vertical structural characteristics in BRAMS and GEOS4 stability profilesBRAMS and GEOS4 stability profiles
Vertical “Offset” between GEOS4 and BRAMS profilesVertical “Offset” between GEOS4 and BRAMS profiles
“Neutral” “High stability”
Example of plume in trapped in a layer of high Example of plume in trapped in a layer of high stabilitystability
From data courtesy David Nelson, 2007
Example of plume distributed in the free Example of plume distributed in the free tropospheretroposphere
Neutral Stable layer present
Above 11 (17%) of plumes are distributed throughout the column
18 (27% of 66 cases)
In 35 (53%)
Below 2 (3%) If there is a layer of If there is a layer of high stability, plumes to high stability, plumes to tend become trapped in tend become trapped in itit
If plumes in a “neutral” If plumes in a “neutral” atmosphere, they tend atmosphere, they tend to be disperseto be disperse
Directions: Moving forwardDirections: Moving forward Preliminary results suggests that Preliminary results suggests that
stability profiles may provide a stability profiles may provide a way to parameterize injection way to parameterize injection heightsheights
First pass analysis of coarser First pass analysis of coarser grid, 2grid, 2°°x2.5x2.5°° data show very data show very similar stability profiles to similar stability profiles to those calculated using data those calculated using data from 1from 1°°x1.25x1.25°° grid. grid.
Ultimately interested in Ultimately interested in relationship between plume relationship between plume heights and height of diffuse heights and height of diffuse smoke smoke
We are moving towards a We are moving towards a parameterization for injection parameterization for injection heights of emissions from boreal heights of emissions from boreal forest fires in GEOS-Chemforest fires in GEOS-Chem
S1: Modeling fire plumes is actually S1: Modeling fire plumes is actually a quite well defined problema quite well defined problem
Essentially plume rise is governed by the characteristics Essentially plume rise is governed by the characteristics of the fire itself (rate fuel consumption determines of the fire itself (rate fuel consumption determines buoyant energy) and on local meteorology (wind buoyant energy) and on local meteorology (wind direction, convection, stability of the atmosphere)direction, convection, stability of the atmosphere)
However, challenge is in parameterizing plume injection However, challenge is in parameterizing plume injection height on a coarse gridheight on a coarse grid
In a coarse gridIn a coarse grid Meteorology is averaged over a large geographical area.Meteorology is averaged over a large geographical area. Other factors are highly uncertain at best (e.g. fuel loading)Other factors are highly uncertain at best (e.g. fuel loading)
Need a statistical method, preferably one that can be Need a statistical method, preferably one that can be done online during simulations done online during simulations
S2: Effect of PAN on ozone S2: Effect of PAN on ozone chemistrychemistry
Implications for ozone chemistry – the effect Implications for ozone chemistry – the effect of PAN carried aloft.of PAN carried aloft.
The Ox anomaly (primarily ozone) in September 1998 for simulation KAS05.D2 at the surface (left) and at ~500 hPa (right) in ppb.
S3: Comparing 66 plume histograms to S3: Comparing 66 plume histograms to stability profiles derived from GEOS4 data:stability profiles derived from GEOS4 data:
“Neutral profile”i.e. increasing or
nearly constant stability throughout troposphere
Level of lower stability (<3)
Levels of higher stability (>5)
Plume mostly IN stability level
0 6 (9%) 35 (52%)
Plume mostly OUT of stability level
8 (12%) 5 (7%) 13 (19%)
S4: Comparing 66 plume histograms to S4: Comparing 66 plume histograms to stability profiles derived from BRAMS data:stability profiles derived from BRAMS data:
“Neutral profile”i.e. increasing or
nearly constant stability throughout troposphere
Level of lower stability (<3)
Levels of higher stability (>5)
Plume mostly IN stability level
0 7 (10%) 33 (49%)
Plume mostly OUT of stability level
9 (13%) 6 (9%) 12 (18%)
S5:Comparing 66 plume histograms to S5:Comparing 66 plume histograms to stability profiles derived from BRAMS and stability profiles derived from BRAMS and
GEOS4 data: BRAMS/GEOS4GEOS4 data: BRAMS/GEOS4“Neutral profile”i.e. increasing or
nearly constant stability throughout troposphere
Level of lower stability (<3)
Levels of higher stability (>5)
Plume mostly IN stability level
0 7 (10%)/6 (9%) 33 (49%)/35 (52%)
Plume mostly OUT of stability level
9 (13%)/8 (12%) 6 (9%)/5 (7%) 12 (18%)/13 (19%)
S6: GEOS4 vs. BRAMS stability S6: GEOS4 vs. BRAMS stability profilesprofiles
Generally, similar vertical structures Generally, similar vertical structures More levels of high stability at lower latitudes in GEOS4More levels of high stability at lower latitudes in GEOS4 Tropopause tends to be higher in BRAMs dataTropopause tends to be higher in BRAMs data Both stability profiles calculated by simple forward method – however, Both stability profiles calculated by simple forward method – however,
BRAMS has higher vertical resolution (150m) and horizontal resolution BRAMS has higher vertical resolution (150m) and horizontal resolution (40km)(40km)
Assumption of standard US atmosphere in calculation of GEOS4 dataAssumption of standard US atmosphere in calculation of GEOS4 data Difference in terrain levels not sufficient to account for “vertical shift”Difference in terrain levels not sufficient to account for “vertical shift”