modeling and observational constraints of nh3 emissions ... · modeling and observational...

Modeling and Observational Constraints of NH3 Emissions and Sources of Nitrogen Deposition

Daven K. Henze University of Colorado, Boulder

Hyungmin Lee, Juliet Zhu, Jana Milford (CU Boulder) Aika Davis, Ted Russell (GIT), Gill-Ran Jeong (KIAPS) Fabien Paulot, Daniel Jacob, Katie Travis (Harvard) Jesse Bash, Robert Pinder, Riche Scheffe, James Kelly (US EPA) Bret Schichtel, John Vimont (NPS), Linda Pardo (USFS)

Environmental impacts of NH3

Estimated N deposition from NHx, Dentener et al. (2006)

Areas where color approaches dark red --> deposited levels are hazardous to ecosystem. NH3 emissions: - increased by factor of 2 – 5 since preindustrial era. - to double by 2050 (IPCC, Denman et al., 2007; Moss et al., 2010). - contribute to 46 Tg gap in global N budget (Schlesinger, 2009)?

mg(N

)/m

2/y

5

10

15

20

Gg

(NH

3)

da

y−1

J F M A M J J A S O N D

OptimizedAlternate

Gilliland (2006)Henze (2009)Zhang (2012)Pinder (2006)Park (2004)Pinder+Cooter

Uncertainties in NH3 emissions

Why so uncertain? - lack of direct source measurements (hard, expensive) - difficulty in relating associated species to NH3 sources - constraints from observations of [NH4

+] or [NHx] complicated by model/measurement error, precipitation - observations of [NH3] less prevalent

- Global inventories also uncertain (e.g., Beuson et al., 2008)

- Substantial variability in estimates of total US NH3 emissions. - Large uncertainties at regional scales (e.g., Novak et al., 2012; Walker et al., 2012)

Gg(N

H3)/

day

Paulot et al., 2014

Uncertainties in NH3 emissions: Implications for air quality and environment

• contribute to errors in assessing PM2.5

• undermine regulatory capabilities for secondary standards on SOx, NOx to control Nr dep (e.g., Koo et al., 2012)

• uncertainties in projections of aerosol direct radiative forcing impacts (Henze et al., 2012)

(also Liao et al., 2007; Henze et al., 2009; Zhang et al., 2012)

Ex: GEOS-Chem overestimates nitrate at IMPROVE / CASTNET (July)

Zhu et al., 2013 Heald et al., 2012 Walker et al., 2012

measured [µg/m3] measured [µg/m3] measured [µg/m3]

GEO

S-C

hem

[µg/m

3]

GEO

S-C

hem

[µg/m

3]

GEO

S-C

hem

[µg/m

3]

US

CA

Constraints on NHx deposition from inverse modeling

Observations: wet NHx = aerosol NH4+ + gas NH3

Method: adjust (w/Kalman Filter) monthly nation-wide scale factors Results:

Gilliland et al., 2003; Gilliland et al., 2006

2003 2006

EPA NEI NH3 emission adjustment factors

Assumptions: - uniform seasonality throughout broad regions of US

Many US air quality models get NHx deposition correct via assimilation.

Top-down constraints based on NHx

Zhang et al., 2012: Seasonality of NH3 sources adjusted so that Modeled matched RPO and SEARCH NHx measurements

- Resulting annual NHx and NO3 deposition unbiased. - Enforces a spatially uniform seasonality / correction factor across the US.

Mendoza-Dominguez and Russell, 2001: constraints on NH3 sources in the SE

Spatial heterogeneity in source-receptor relationships for NH3

Spatial correlations of ∆emiss with:

∆[NH3] ∆ wet dep [NHx] 0.83 0.54 0.17 -0.06

Kg NH3/ha/month

April

July

Spatially heterogeneous impacts of NH3 emissions – can be accounted for using 4D-Var / adjoint inversions

Jeong et al., submitted

Consider emissions perturbation, ∆emiss:

Sensitivity of all model concentrations to one model source or sector

t0 Perturbation at source region

Forward

tn

Changes of concentration

Forward Model (source-oriented)

Source attribution techniques



Forward

tn



US Anthropogenic 5.0 Tg N / yr

Foreign Anthropogenic 0.42 Tg N / yr

Natural 1.0 Tg N / yr

Zhang et al., 2012




Forward

tn



Sensitivity of model concentration in specific location to many model sources and sectors

Adjoint Model (receptor-oriented)

t0 area of possible origin

adjoint

tn

Concentration at the receptor




Zhang et al., 2012




Forward

tn



Sensitivity of model concentration in specific location to many model sources and sectors

Adjoint Model (receptor-oriented)

tn

Concentration at the receptor




Zhang et al., 2012


Using receptor = sum of squared model error, these relationships can be used for high resolution inverse modeling

[unitless]

Constraints from NHx deposition, and an alternate bottom up inventory

Paulot et al., 2014 - GEOS-Chem 4D-Var

(Henze et al., 2007) - Global 2x2.5 - Assimilate NTN, EMEP, …


New bottom-up inventory

No support for homogeneous seasonality in the US. New bottom-up inventory (MASSAGE) can reproduce optimized emissions in some areas.

Paulot et al., 2014


Paulot et al., 2014

Comparison to surface NH3 measurements (Puchalski et al., 2011) before and after assimilation:


Paulot et al., 2014

Comparison to surface NH3 measurements (Puchalski et al., 2011) before and after assimilation:

Closure for NHx deposition does not necessarily imply better model NH3

Potential for making new inroads on this problem: ambient measurements of NH3

EPA’s AMoN sites (>2007) (Puchalski et al., 2011)

Also LADCO, SEARCH, CSU, ANARChE

TES: - 5 km x 8 km footprint - sensitive to BL - detection limit of ~ 1 ppb - bias of +0.5 ppb more precise & sparse than IASI

(Beer et al., 2008; Clarisse et al., 2009; Clarisse et al., 2010; Mark Shephard et al., 2011)

TES NH3 sensitivity

Remote sensing with TES and IASI:

Passive surface measurements:

Validating TES NH3 with surface observations

Overlap surface obs with TES Transects for 2009:

NH3 Emission Density

[kg NH3 / km 2 ]

< 100

1000

>10000TES Transect

CAMNet Monitoring Site

TES reflects real-world spatial gradients and seasonal trends

Pinder et al., 2011

Constraining emissions of NH3 in GEOS-Chem using 4D-Var technique (Zhu et al., 2013)

X - 32 ZHU ET AL.: INVERSE MODELING RESULTS OF NH3 EMISSIONS

BALES ET AL.: SHORT TITLE X - 3

(a) (b)

F igur e 1. Global figure capt ion (a) describes the first subfigure; (b) describes the second subfigure;

m = 1.22

r2 = 0.94

2 4 6 8 10 12 14 16

x 106

2

4

6

8

10

12

14

16

x 106

True

Modeled

r2 =1

m =1

r2 =0.935

m =1.22

(a)(b) (c)


3. A ct ual appl icat ion

For an applicat ion with real data, we will use TES

observat ions throughout 2009 and compare these to

model est imates from the GEOS-Chem chemical t rans-

port model in a global 2◦ ⇥2.5◦ simulat ion.

A ck now ledgm ent s. (Text here)

R efer ences

Dubovik, O., T . Lapyonok, Y . J. K aufman, M . Chin, P. Ginoux,R. A . K ahn, and A. Sinyuk (2008), Ret rieving global aerosolsources from satellit es using inverse modeling, Atmos. Chem.Phys., 8 (2), 209–250.

Elbern, H., H. Schmidt , O. Talagrand, and A. Ebel (2000), 4D-varat ional data assimilat ion wit h an adjoint air quality modelfor emission analysis, Environ. Model l. Softw., 15, 539–548.

(a)


(a) (b)


(a)

2 4 6 8 10 12 14 16

x 106

2

4

6

8

10

12

14

16

x 106

True

Modeled

r2 =1

m =1

r2 =0.865

m =1.21

m = 1.21

r2 = 0.87

(b)(c)

F igur e 2. Global figure capt ion (a) describes the first subfigure; (b) describes the second subfigure; (c) describes thesecond subfigure;







R efer ences

Dubovik, O., T . Lapyonok, Y . J. K aufman, M . Chin, P. Ginoux,

R. A . K ahn, and A. Sinyuk (2008), Ret rieving global aerosol

sources from satellit es using inverse modeling, Atmos. Chem.

Phys., 8 (2), 209–250.

(b)


(a) (b)


(a)(b)

2 4 6 8 10 12 14 16

x 106

2

4

6

8

10

12

14

16

x 106

True

Modeled

r2 =1

m =1

r2 =0.992

m =1.03

m = 1.03

r2 = 0.99

(c)








R efer ences

Dubovik, O., T . Lapyonok, Y . J. K aufman, M . Chin, P. Ginoux,R. A . K ahn, and A. Sinyuk (2008), Ret rieving global aerosolsources from satellit es using inverse modeling, Atmos. Chem.Phys., 8 (2), 209–250.

Elbern, H., H. Schmidt , O. Talagrand, and A. Ebel (2000), 4D-varat ional data assimilat ion wit h an adjoint air quality modelfor emission analysis, Environ. Model l. Softw., 15, 539–548.

(c)

Figure 4. Tests for the possible impacts of inversion error, retrieval bias and measure-

ment error: (a) ret rieval algorithm with a polluted profile as an init ial guess; (b) modified

retrieval algorithm with a moderate profile as the init ial guess; (c) model profiles from

the true model were ascribed error of the same size as the measurement error.

April

July

October

Initial Optimized ln(Optimized/Initial)

0 2.33 4.67 7.00 [10 6 kg] -2.00 -0.67 0.67 2.00 [ unitless]

Figure 5. NH3 emissions from GEOS-Chem before and after the assimilat ion

D R A F T Jul y 23, 2012, 11: 33am D R A F T

NH3 emissions in GEOS-Chem

+80%

+57%

+33%

Constraints from TES improve estimates of NH3 at AMoN sites in April and October. Contradicting in July.

AMoN surface obs (ppb)

GEO

S-C

hem

NH

3 (ppb)

Apri

l Ju

ly

Octo

ber Agrees with constraints using

NHx deposition & new bottom up inventory from Paulot in April (+/- 20%) but not in July

Diurnal variability of NH3: case study in Warsaw, NC, with CMAQ regional model

CMAQ*

CMAQ* modified diurnal NH3 emissions

Observations downwind of livestock facility

(Walker et al., 2006)

Time of day

NH

3 [

µg/m

3]

* Using NEI05 emissions, simulated year not same as observations

Improved diurnal variability (Bash) can help resolve discrepancies between in situ and satellite obs (Jeong et al., submitted)

Impacts of bidirectional exchange in GEOS-Chem

AMoN (ppb)

Bidi applied to optimized emissions

Optimized (Zhu et al 2013)

Improved (mechanistic) representation of NH3 fluxes may help resolve inconsistencies between NH3 and [NHx]dep constraints.

Other considerations in remote-sensing constraints: - temporal sampling bias - spatial sampling bias

Zhu

Base M=1.13

Bidi M=1.23

GC [

g/(

yr

m2)]

NTN NHx [g/(yr m2)]

NASA AQAST Tiger Team

Overview: • multi-model assessment of current

and future sources of reactive nitrogen deposition in Class I and at-risk ecosystems in the US

Members: • Daven Henze, Jana Milford (CUB) • Fabien Paulot, Daniel Jacob (Harvard) • Aika Yano,Ted Russell (Georgia Tech) • Bret Schichtel, John Vimont (NPS) • Rich Scheffe, James Kelly (US EPA) • Linda Pardo (USFS)

Tools / Observations: • NH3 remote sensing, in situ observations (RMNP,…) • GEOS-Chem and CMAQ models • Source attribution techniques: sector perturbations, DDM, adjoint

Ellis et al., 2013)

What are the sources contributing to exceedences in Federal Class I Areas?

model configurations and domains

GEOS-Chem: - 0.5° x 0.667° - 2010 - NEI 2008 - GFEDv3

CMAQ v5: - 36km CONUS - 4km over NPs - 2010 - NEI 2005 scaled to 2010 - Bidirectional NH3 exchange - CB05 with Pleim-Xiu LSM - WRF v3

Footprints of reactive Nitrogen deposition

NH3

Area (livestock) Fert

NOx

Area (mobile) Point Nonpt Lightng Soil

Lee, Paulot, Davis

Rocky Mountain NP: - CMAQ: 0.85 kgN/ha/a from livestock NH3, 1.55 kgN/ha / from mobile NOx

- Gebhart et al. (2011): 50% of NH3 inputs from out-of-state - Benedict et al. (2013):

Footprints of reactive Nitrogen deposition

Lee, Paulot

NH3

Area (livestock) Fert

NOx

Area (mobile) Point Nonpt Lightng Soil

Effectiveness of NH3 vs NOx emission controls for approaching deposition Critical Loads

3

3.5

4

4.5

5

kgN

ha

−1 a

−1

Voyageurs NP

4

6

8

10

12Great Smoky Mountains NP

0 5 10

x 105

4

6

8

10

12

kg

N h

a−

1 a

−1

Area (km2)

Shenandoah NP

0 5 10

x 105

3

4

5

6

7

8

Area (km2)

Adirondack

NH3 (an − 20%) NO

x (an − 20%) N (an − 20%) CL (Ellis) CL (EPA)

3

3.5

4

4.5

5

kgN

ha

−1 a

−1

Voyageurs NP

4

6

8

10


0 5 10

x 105

4

6

8

10

12

kg

N h

a−

1 a

−1

Area (km2)

Shenandoah NP

0 5 10

x 105

3

4

5

6

7

8

Area (km2)

Adirondack

NH3 (an − 20%) NO


3

3.5

4

4.5

5

kgN

ha

−1 a

−1

Voyageurs NP

4

6

8

10


0 5 10

x 105

4

6

8

10

12

kg

N h

a−

1 a

−1

Area (km2)

Shenandoah NP

0 5 10

x 105

3

4

5

6

7

8

Area (km2)

Adirondack

NH3 (an − 20%) NO


3

3.5

4

4.5

5

kgN

ha

−1 a

−1

Voyageurs NP

4

6

8

10


0 5 10

x 105

4

6

8

10

12

kg

N h

a−

1 a

−1

Area (km2)

Shenandoah NP

0 5 10

x 105

3

4

5

6

7

8

Area (km2)

Adirondack

NH3 (an − 20%) NO


What is the impact of reducing anthropogenic emissions by 20% as a function of distance (area) away from the park?

3

3.5

4

4.5

5

kgN

ha

−1 a

−1

Voyageurs NP

4

6

8

10


0 5 10

x 105

4

6

8

10

12

kg

N h

a−

1 a

−1

Area (km2)

Shenandoah NP

0 5 10

x 105

3

4

5

6

7

8

Area (km2)

Adirondack

NH3 (an − 20%) NO


NH3

NOx

Both EPA CL Ellis CL

Voyager Great Smokey Shenandoa

-local NOx -distant NH3

- NOx - same

All regions are far from attaining CL values with small reductions to emissions over a wide 106 km2 area (size of France!) Paulot

What is the nitrogen deposition efficiency?

Grand Teton

NH3 NOx SO2

Joshua Tree

(kg N dep / ha / yr) / (mol emission / yr)

Implications for impacts of new sources Lee

Projections of Nr deposition

Projections of the evolving roles of NH3 and NOx on Nr deposition following emission projections from IPCC AR5 (Moss et al., 2010)

While Nr may be decreasing, role of NH3 increasing

Paulot et al., 2012; also Ellis et al. 2013

Final comments

• Constraints from multiple sources (remote sensing, deposition, in situ measurements) helping reduce uncertainty in NH3 emissions.

• 4D-Var techniques allow inversion process to consider spatially heterogeneous biases in emissions inventories.

• It’s an iterative procedure, and we’re learning more about process-level emissions (diurnal variability, bi-directional fluxes).

• NH3 and NOx sources can contribute significantly to reactive nitrogen deposition several states away.

• Substantial controls required to approach critical loads, particularly given projected increases in NH3 emissions.

Remote sensing of NH3: IASI

Van Damme et al., ACPD, 2013

16

Model evaluation: GEOS-Chem vs observed (NTN) N deposition

0

1

2

3

GrandTeton JoshuaTree RockyMountain Sequoia

kgN/ha/season

MAM(meas) MAM(model)

JJA(meas) JJA(model)

SON(meas) SON(model)

DJF(meas) DFJ(model)

2010

TES NH3 visualization

TES NH3 constraints in GEOS-Chem: spatial sampling / retrieval bias

Consider all 12 x 12 km2 CMAQ grid cells Of these, in which did we have a successful TES retrieval? => TES constraints may be ~30% high


Paulot in prep

a priori

optimized

Alternative bottom-up

Top-down constraints agree with recent bottom up inventories: Huang (2012) and Alternate.

Annual NH3 emissions in GEOS-Chem

Monthly SE Asia NH3 emissions

Seasonality in SE China from TES NH3

observations (Shephard et al., 2011)

Optimized

Constraints from CASTNet NH4+? n(NH4

+) : 2n(SO42-) + n(NO3

-)

CASTNet, all sites, 2005-2006 (R. Pinder)

Field campaigns (Sorooshian et al.)

SJ Valley

Houston

Issues with evaporation

January

April

Base

Diurnal - Base

NO3-*0.67 - Base

(Heald 2012)

NO3- (surface) NH3 (2 km)

Mechanistic NH3 emissions an important future direction for global models. Other factors: - BL heights (Dalhousie, following Lin and McElroy, 2010) - excessive N2O5 (Zhang et al., 2012; Paulot et al., submitted)

NH3 (surface)

Conundrum of nitrate (too high) and ammonia (too high at surface, too low higher up) in July in GEOS-Chem

NH3: CMAQbidi - CMAQbase

April

July

October

Decreased deposition in July leads to enhanced NH3 lifetime throughout the US.

Jeong et al., submitted

Impacts of bidirectional exchange in GEOS-Chem

modeling and observational constraints of nh3 emissions ... · modeling and observational...

Documents