Improved Biogenic Emission Inventories across the West project

Alex Guenther

NCAR Earth System Laboratory, Boulder CO

Greg Yarwood and Tan SakulyanontvittayaENVIRON, Novato CA

Outline

• Biogenic emissions impact on ozone• NCAR/NSF: Ongoing efforts to improve biogenic emission estimates• ENVIRON: WRAP biogenic emissions project

NCAR is sponsored by the National Science Foundation

Biogenic contributions to ozone

• Ozone formation requires both VOC and NOx• VOC

– In many regions, biogenics provide a ubiquitous background of reactive VOCs that are available to form ozone with any NOx

– Much lower emissions in winter than summer

• NOx– Soil NOx emissions are generally low

compared to anthropogenic NOx, but become more significant when fertilizer is applied

Biogenic VOC emissions are much lower in winter and are highly variable across different landscapes

1 mg/m2 = 1 kg/km2

Ozone attributed solely to biogenics

• CAMx APCA source apportionment analysis

• Ozone formed when biogenic NOx andVOC interact– Contribution is

determined by the biogenic NOx

• Biogenic VOCs also interact with NOxfrom other sources (anthro, fires) – See next slide

Highest hourly ozone contribution (ppb) during June/July 2006 from Denver SIP 12 km grid

Colorado

Ozone attributed to other sources

Ozone formed by these categories depends upon availability of biogenic VOC

Fires

Other Anthropogenic

Colorado Oil and Gas

Mobile (on and off-road)

Leaf & Plant Chamber ExperimentsBiosphere 2, Aspen FACE

Forest Canopy Studies

Forest Landscape Studies

Global Studies

PROPHET, Boardman, La Selva, modeling

PROPHET, Boardman, La Selva, modeling

modeling

How can we improve biogenic emission inventories?

• OZIE

GLOBAL STUDIES

REGIONAL STUDIES

CANOPY STUDIES

LEAF/PLANT STUDIES

1. Improve our basic understanding of the processes controlling biogenic emissions

2. Improve our driving variable datasets

-Weather-Temperature-Solar Radiation-Soil Moisture

-Landcover-Leaf Area-Plant Functional Types-Plant species composition-Emission factors

-Disturbance/Stress-Insects-Temperature extremes-Drought-Air Pollution

Greenhouse and growth

chamber

Leaf/Plant Chamber Studies

Field

Process studies to develop numerical algorithms for quantitative modeling

Physiologal: response of emissions to light, temperature, moisture, etc.

Biochemical:Differential gene expression from control/stress trees. Pink arrow= heat stress tree Blue arrow = control tree

1.5

2

2.5

3

30 35 40 45Temperature (oC)

Isop

rene

em

issi

on

activ

ity After cool/cloudy weather

After hot/sunny weather

Whole Canopy Studies

Results reveal complex patterns of emission, loss and transformation in vegetation canopies

Michigan

Duke Forest

Isoprene Flux

Costa Rica

Amazon

Niwot Ridge

1D model constrained by observations to develop/ test parameterizations for 3D models

Radical Production

Airborne Regional Studies

INPE Beinderante

NCAR EC-130Q

Purdue Duchess

Airborne measurements of BVOC fluxes and vertical profiles of oxidation products

June 2011 CABERNET study • Eddy covariance measurements are the only direct evaluation of BVOC fluxes• Vertical profiles of BVOC oxidation products allow us to better constrain emissions and chemistry

CIRPAS Tw in Otter

Global BVOC emission studies

We can estimate regional to global BVOC emissions from satellite observations of formaldehyde, glyoxal and methanol using inverse modeling.

The inverse modeling is limited by our lack of understanding of BVOC chemistry but is useful for identifying missing processes

New (MEGAN 2.1) algorithms and emission factors agree surprisingly well with satellite observations but there are areas of disagreement (e.g., desert shrublands)

MEGAN 2.1 bottom -up

IASI satellite CH3OH columns

Jacob et al. 2005 bottom -up

Adjusted IASI satellite CH3OH

Field study characterizing emission factors and landcover distributions demonstrated that biogenic VOC emissions were overestimated in Phoenix and underestimated outside of Phoenix. This has an important impact on ozone model results

Biogenic emission models can run globally but accurate results require accurate driving variables

with an appropriate spatial resolutionBEFORE Field Study AFTER Field Study

< 100 kg/day

100-200 kg/day

200-300 kg/day

< 100 kg/day

100-300 kg/day

This includes weather!weather data is associated with isoprene emission uncertainties of • 20 to 40% in the Pearl river delta of China (Wang et al. Tellus, 2010)• 15 to 35% in the U.S. Ozarks (Carlton and Baker, ES&T 2011)

Ozone and particles are controlled by emissions, chemistry and deposition

Emissions-known (but uncertain)-unknown

Chemistry-reaction rates-oxidation pathways-product yields

Deposition-dry -wet

Models underpredict BVOC oxidation product deposition by > 30% (Karl et al. Science 2010).

Models underpredict BVOC loss rates by > 50% (Lelieveld et al. Nature 2008).

Model of Emissions of Gases and Aerosols from Nature (MEGAN)

• Global model with ~1 km2 and hourly resolution

• Designed for regional and global emission modeling

• >150 registered users from > 20 countries on 6 continents

• bai.acd.ucar.edu Guenther et al. (2006) Atmos. Chem. Physics

Weather and

landcoverdriving

variables

• Improved soil NO algorithm: precipitation “pulsing”, nitrogen fertilizer.

• Higher resolution (temporal and spatial) landcover

• Full canopy environment model

• New emission factors/algorithms for methanol, acetaldehyde and ethanol

• Revised coefficients and emission factors for other VOC

• Isoprene response to CO2

MEGAN 2.1: to be released in September 2011

• Improved soil/litter emission (VOC, NH3, NO)

• Biological particles (pollen and fungal spores)

• Improved isoprene and monoterpene emission factors and algorithms

MEGAN 2.2: ???

MEGAN 2.04: released in 2007

WRAP project deliverables1. Code improvements (MEGAN 2.1) (July 29, 2011)

• Leaf Area (LAI): 8-day average values instead of monthly. • Improved soil NO emissions (precip. pulsing and N fertilizer)

2. Driving variable improvements (July 29, 2011)• Values representative of a specific year (2008) • Leaf Area (LAI) 8-day average values with revised MODIS data • Plant functional types (PFTs) based on 30-m data for non-urban, 1-m data for urban correction• Emission factors based on more detailed and accurate ecoregion averaging and new emission factors for western U.S. landscapes

3. Emission estimates• County level VOC, NO, CO summary (August 29, 2011)• Photochemical grid model-ready data (September 26, 2011)

4. Analysis and Report (October 24, 2011)• Analysis of code and driving variable changes (July 29, 2011)• Comparison with other models (August 29, 2011)• Assessment of accuracy and potential future improvements

Plant Functional Type (PFT) distributions: • 1000 m resolution for homogeneous landscapes (MODIS)

• 30 or 56 m resolution for most non-urban landscapes (LANDSAT-TM /AWiFS)

• 1 m resolution works for all landscapes (Photo imagery)

Needleleaf trees: 0%

Broad-leaf trees: 35%

Shrub: 25%

Barren or water: 25%

Crop: 0%

Grass and other: 15%

1 km

High resolution (<=1 m) imagery finds more trees

Duhl et al. 2011

MODIS 1-km

NLCD 30-m

PI 1-m

Nowak et al. divided the U.S. into 66 regions and estimated an average difference between satellite NLCD (30-m) and photo imagery PI (<= 1m) tree cover estimates. We will use these data to “adjust” NLCD derived distributions of urban tree cover

8-day average North America LAI for specific years

We will analyze-all days in 2008 - June and August days in 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011

Difference in LAI for third week of June in 2010 and 2004 (purple is lower, green is higher, white is about the same)

USDA/NASS CDL 56-m landcover (2008, 2009, 2010)

We will combine CDL with Ecoregion/Urban-region averaged USFS/FIA tree statistics (>600000 trees) and NRCS shrub/grass data.

~100 crop types (e.g. corn, citrus, squash, asparagus, dbl lettuce/barley)~4000 wildland species composition types (4 PFTs in 968 ecoregions)

~250 urban types (4 classes in 66 regions)

1 km

• Biogenic emissions can impact ozone- but only in combination with anthropogenic

• More processes and measured parameters.• More accurate/representative landcover. • Model-ready biogenic emissions on

September 26, 2011• Airborne/satellite measurements can

evaluate/improve biogenic emissions and their impact on regional air quality

Summary

Improved Biogenic Emission Inventories across the West project