Difference between revisions of "MEGAN biogenic emissions"

Revision as of 19:03, 14 August 2014

This page describes the implementations of the MEGAN (Model of Emissions of Gases and Aerosols from Nature) biogenic emissions in GEOS-Chem.

Overview

GEOS-Chem has used two versions of the MEGAN biogenic emissions:

MEGAN 2.0

Used in GEOS-Chem v8-02-03 and prior versions

MEGAN 2.1

Used in GEOS-Chem v8-02-04 and later versions

MEGAN prior to GEOS-Chem v8-02-04

GEOS-Chem v8-02-03 was the last version to use the MEGAN v2.0 emissions. Please see the discussion below for the latest information.

The source code for MEGAN v2.0 emissions are in the files:

1. lai_mod.f
2. megan_mod.f

For more information about the data files, please see the following README: GEOS_1x1/MEGAN_200510/README

MEGAN documentation

Dylan Millet wrote:

There was discussion at the 2009 GC User's Meeting about divergent implementations of MEGAN and the need to better document what we're doing in GEOS-Chem. Here is a description (current as of 5/1/2009) for people to use and modify in their publications.

We use the Model of Emissions of Gases and Aerosols from Nature (MEGAN) [Guenther et al., 2006] to derive isoprene emissions for each GEOS-Chem grid square [Palmer et al., 2006; Millet et al., 2008]. MEGAN computes emissions for plant functional types (PFTs) as a function of temperature, solar radiation, leaf area index (LAI), and leaf age. Fractional coverage for each PFT and vegetation-specific emission factors (EFs) are based on the MEGAN landcover data (PFT v2.0, EFs v2.0). The temperature effect on leaf emissions is computed as a function both of the current temperature and the average temperature over the previous 15 days [Guenther et al., 1999]. The implementation also includes a 5-layer canopy model [Guenther et al., 1999] to account for the effects of direct and diffuse PAR on sunlit and shaded leaves. We explicitly consider the effect of leaf age on emissions, with leaf age fractions for new, young, mature, and old leaves estimated for each model grid square from local LAI and temperature changes according to Guenther et al. [2006]. We drive MEGAN in GEOS-Chem with assimilated surface air temperature and photosynthetically active radiation (direct and diffuse) observations from the NASA Global Modeling and Assimilation Office (GMAO). Monthly mean LAI values are based on AVHRR satellite data [Buermann et al., 2002].

See the list of references below for more information.

--Dbm 14:48, 1 May 2009 (EDT)

Updated annual emission factor for Isoprene

In GEOS-Chem v7-04-12 (as of Sep 2007), we use the older annual emission factors (AEF's) for isoprene.

However, Alex Guenther (personal communication) uses the current isoprene AEF's were obsolete and that we should be using updated isoprene AEF's. Dylan Millet has created an updated AEF file for isoprene but it has not been implemented into the standard GEOS-Chem code yet. (Some folks have already used the new AEF's in their research codes.) The new AEF's are currently scheduled to be implemented in GEOS-Chem beta release v8-01-02.

Reference: Guenther, A., T. Karl, P. Harley, C. Wiedinmyer, P.I. Palmer, and C. Geron (2006), Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols from Nature), Atmospheric Chemistry and Physics, 6, 3181-3210.

MEGAN Website: http://bai.acd.ucar.edu/Megan/index.shtml

--Bob Y. 11:57, 2 December 2008 (EST)

Impact on isoprene emissions

Here are some presentations which document the change in the MEGAN biogenic emissions of isoprene caused by switching to the new annual emission factor (AEF) file.

1. HERE is a comparison (by Dylan Millet) of isoprene emissions from GEIA vs. old MEGAN (i.e. v8-01-03 and prior) vs. new MEGAN (v8-01-04 and later, with updated baseline annual emission factors from Alex Guenther). 2001, 2x2.5, v7-04-11, GEOS-4 met.
2. HERE is a comparison (by Rynda Hudman) on the effects of MEGAN on the v7-02-04 benchmark.

NOTE: The newer AEF's for isoprene are now used in GEOS-Chem v8-01-04 and higher.

--Bob Y. 09:39, 3 March 2009 (EST)

MEGAN in GEOS-Chem v8-02-04 and later versions

GEOS-Chem v8-02-04 is the first "standard" code version to include the MEGAN v2.1 emissions. Please see the Overview section below for full details.

The source code for MEGAN v2.1 emissions are in the files:

1. lai_mod.f
2. megan_mod.f
3. meganut_mod.f

For more information about the data files, please see the following README: GEOS_1x1/MEGAN_200909/README

Overview

Mike Barkley has written a document entitled Description of MEGAN biogenic VOC emissions in GEOS-Chem. Please see this document for a full description of the MEGAN v2.1 emissions.

For more information, please follow this discussion about the implementation of MEGAN 2.1:

Mike Barkley wrote:

I've read in the latest GEOS-Chem news letter that your working on terpene emissions from MEGAN. I've got the latest v2.1 emission factors for these species (+isoprene & MBO) already included in my model code (v8-01-04) which I drive with MODIS LAI collection 5 (a supposedly much improved version) & which is divided by the fraction of vegetated area of each grid cell.

Be aware that the current implementation of the emissions in Geos-Chem is 'wrong' as it takes the LAI into account twice: once in the canopy model & once in the seasonality factor (although somehow the global total comes out okay ~500 Tg isoprene year). I've spoken to Alex about this, if the LAI is included in the canopy model, then the factor: 0.49d0 * LAI / SQRT( 1.d0 + 0.2d0 * LAI*LAI ) should not be applied as well.

I now run the emissions with two different algorithms:

1. Using the existing canopy model but with the megan module restructured & with new leaf-age & temperature algorithms.
2. Using the new PCEEA approach (which doesn't use a canopy model) outlined in Gunther 2006 which now takes into account the light dependency of monoterpene emissions.

From these two parameterizations the 10 year (1996-2006) global averages using ND46 at 4x5 & GEOS-4 met data are:

1. = 518 ± 26 Tg isoprene year (1sd)
2. = 403 ± 21 Tg isoprene year (1sd)

Obviously, these are quite large differences which is why I've hesitated about putting them forward to the standard code. I've done some validation at a three tropical sites, at 2 sites the algorithms give pretty much the same and at the other there are about ~30%-ish differences in the emissions. With the PCEEA algorithm the emissions are much closer in total to the estimates of Muller et al (2008) which were ~410 Tg isoprene year. It makes a big difference to the HCHO columns!

What I am going to do is put an optional flag in the input.geos file to turn the PCEEA algorithm on/off as required. I've got to tidy all this code up, including the LAI data sets but once I've done this I can send it you & Harvard, as it needs checking thoroughly.

--Dbm 13:10, 17 June 2009 (EDT)

Yuhang Wang wrote:

Jun Nam in my group has done some updates with isoprene emissions in GEOS-Chem. He's using MEGAN v2.1 with PCEEA (and w/o canopy model). The biggest problem he's got so far is when we switched from GEOS-4 to GEOS-5 met fields. Isoprene emissions are much higher. Apparently, surface temperature in GEOS-5 is too high (for 2006; we are using the current release of GEOS-Chem and GEOS-5 met field). He's sent emails to Steve Pawson but didn't get a response yet. So he used NCEP surface temperature instead and the global isoprene emission rate is 562 Tg C/year.

Can you set up a thread on wiki for this discussion so that Jun can provide more details if anyone wants to know.

--Dbm 11:00, 18 June 2009 (EDT)

Jun Nam wrote:

GEOS-5 has higher temperature than NCEPv2 reanalysis for 2006. Global isoprene emission is 658 Tg C (GEOS-5) vs. 562 Tg C (NCEPv2). For this analysis, I used PCEEA algorithm (Guenther et al., 2006) and MEGAN V2.04 (most recent version). I've checked high HCHO regions in 5 continents (except australia) and the surface temperature difference (in July afternoon) is 2-3 degrees (temporal variation agrees very well). As a result, activity factor due to temperature (gamma_t) is about 30-45% higher using GEOS-5 than NCEPv2.

(...) I'm using version 2.0 of the MEGAN LAIv (LAIv data described in Guenther et al. 2006), which is in netcdf_30min format available in MEGAN download portal (CDP) input folder.

Mike Barkley wrote:

I've done a quick comparison between the emissions resulting from GEOS-4 and GEOS-5 met data for the year 2004 only, using the PECCA model approach:

   GEOS-4 ==> 2004  ::  362 Tg C  =  410 Tg isoprene
   GEOS-5 ==> 2004  ::  338 Tg C  =  383 Tg isoprene

which is about ~7% difference. This difference maybe due to variations in LAI as well as in the meteorology.

--Dbm 14:56, 24 June 2009 (EDT)

There seems to be some contradiction here between what Jun is finding for GEOS-5 2006 (658 TgC) and what Mike is finding for GEOS-5 2004 (338 TgC).

Dylan Millet wrote:

Michael Barkley has packaged his updates to MEGAN in GEOS-Chem into a beta/non-standard version of the model (v8-2-1), and is very kindly making it available. There's a description above, but briefly it drives emissions of isoprene, terpenes, and MBO with MODIS LAI collection 5 and uses updated emission factors and emission algorithms (MEGAN v2.1). It also includes an option to use the canopy model currently implemented in GEOS-Chem or an alternative approach (PCEEA) described by Guenther et al. [ACP 2006].

For anyone interested in testing/using this code, you can find it as a gzipped tarball here:

<http://www.atmoschem.umn.edu/pub/index.php?dir=bvoc/>

Once you unpack it you'll find a README detailing changes from standard code.

If you do use it please be sure to get in touch with Mike (and also let me know, so I can see how much of a demand there is) - and if you use it in a paper it would be appropriate to offer Mike co-authorship.

Thanks!

--Dbm 17:54, 10 July 2009 (EDT)

Use MEGAN with MODIS leaf area indices

The discussion about the GEOS-v8-02-04 benchmark simulations has revealed that one should always use the MODIS-derived leaf area index product when using the updated MEGAN v2.1 biogenic emissions in GEOS-Chem. This is because the older AVHRR-derived leaf area index product (used by the previous implementation of MEGAN in GEOS-Chem) is biased high w/r/t MODIS and will produce more isoprene as a result.

Typical settings in the input.geos file should be:

Include biogenic emiss? : T
 => Use MEGAN inventory?: T
 => Use PCEEA model?    : T
 => Use MEGAN for MONO? : T
 ...
Use AVHRR-derived LAI?  : F
Use MODIS-derived LAI?  : T

We have installed a trap into GEOS-Chem v8-02-04 and higher versions that will automatically select the MODIS LAI product when the MEGAN v2.1 biogenic emissions are used.

--Bob Y. 16:11, 23 February 2010 (EST)

Mike Barkley wrote:

One thing about the MODIS LAI, at high northern latitudes (typically > 60ish degrees) during the winter months (DJF) there are missing pixels that haven't been processed owing to cloud cover, high surface reflectance, etc. In such cases here I've assigned a default LAI value of zero. The true LAI could be > 0 but is most likely to be small <1.0. The cold temperatures, low light levels & snow cover would make the bigoenic VOC emissions negligible anyway, so using LAI=0 should be ok in this respect.

--Bob Y. 09:48, 3 March 2010 (EST)

Speciated biogenic emissions

The MEGAN v2.1 inventory contains speciated emissions for several secondary organic aerosols. Please see this post on our Secondary Organic Aerosols wiki page for more information.

--Bob Y. 14:02, 19 March 2010 (EDT)

Met field requirements for MEGAN

The current MEGAN 2.1 emissions code computes a 10-day running average of the surface temperature field. At the start of a GEOS-Chem simulation, the MEGAN emissions code will attempt to read the previous 10 days's worth of surface temperature data. If you do not have the met field data files that contain the surface temperature field on disk for the 10 days prior to the start of your GEOS-Chem simulation, then GEOS-Chem will die with an error..

P.S. The now-obsolete MEGAN 2.0 emissions code used to keep a 15-day running average of surface temperature.

--Bob Y. 10:12, 8 March 2011 (EST)

Technical notes in MEGAN 2.1

AEF_OVOC is not currently used

Jintai Lin wrote:

For global simulations, at approximately line 1863 of megan_mod.F, the file name

         !---------------------------------------------
         ! Read in other VOC Annual Emission Factors
         !---------------------------------------------

         ! 1x1 file name
         FILENAME = TRIM( DATA_DIR_1x1 ) //
        &           'MEGAN_200510/MEGAN_AEF_MTP.geos.1x1' 

should probably be

         ! 1x1 file name
         FILENAME = TRIM( DATA_DIR_1x1 ) //
        &           'MEGAN_200510/MEGAN_AEF_OVC.geos.1x1' 

with respect to emissions of other VOCs.

Dylan Millet wrote:

The other VOC emissions are not currently used. Those routines I think are mainly there as a placeholder for adding things in the future. So it doesn't matter if MEGAN_AEF_MTP is read in to AEF_OVOC, since AEF_OVOC is not used. As far as I know the file MEGAN_AEF_OVC.geos.1x1 doesn't actually exist.

--Bob Y. 13:40, 18 November 2011 (EST)

Nested simulations use wrong AEF's

We have corrected this in GEOS-Chem v9-01-02.

Jintai Lin wrote:

For nested simulations, the subroutine GET_AEF_05x0666 is not properly updated to reflect the megan 2.1 emissions.

Lin Zhang wrote:

I want to bring up an error in the MEGAN biogenic emissions with the nested model in the latest versions (eg. v9-01-01). The MEGAN emissions have been updated from MEGAN 2.0 to MEGAN 2.1 since v8-02-04 in the standard model with 4x5 or 2x2.5 resolution, however, they are not fully implemented in the nested model. The nested model for all regions still uses some old annual emission factors (as we can see in the module megan_mod.f, subroutine GET_AEF vs. GET_AEF_05x0666). This error would cause biogenic isoprene emissions (with MEGAN 2.1) over the US a factor of two higher in the nested NA model than that in the 2x2.5 model. It can also affect the CH and EU nested models.

A quick fix can be implemented by changing megan_mod.f, subroutine INIT_MEGAN from:

   #if   defined ( GRID05x0666)
         CALL GET_AEF_05x0666
   #else
         CALL GET_AEF
   #endif

to

   CALL GET_AEF

In this case, the model would read the annual emission factors at 1x1 resolution and then regrid to the underlying resolution.

--Bob Y. 12:55, 22 November 2011 (EST)

New Update Required in MEGAN algorithm

Chris Chan Miller found that the isoprene emissions in MEGAN do not capture the diurnal variability of isoprene observed in the Michigan state forest. This is because the meteorological variables input to GEOS-Chem (temperature and photosynthetically active radiation) are not interpolated between the 3-h average surface data read in to megan_mod.f. This can be fixed by including a CALL INTERP statement in main.f for I-3 met fields. Chris is working on updating this in his working code and will send necessary updates to Bob to be included in the standard code.

--Emarais 16:32, 4 April 2012 (EDT)

Update to MEGAN2.1 (Guenther et al., 2012)

Please see this file that describes an available update to MEGAN2.1 in GEOS-Chem. The update applies emission algorithm updates according to Guenther et al. (Geosci. Model Dev., 5, 1471–1492, 2012), enables computation of emissions for more compounds, uses CLM4 plant functional type distributions, facilitates computing emissions with different/dynamic land cover schemes, and streamlines the code.

Please contact Dylan Millet to obtain the code or with any questions.

Documentation:

• Readme

• Comparison between previous and updated BVOC emissions for standard tracers (year-2006):
  • 2x2.5
  • 4x5

• Comparison between previous and updated BVOC emissions for standard tracers (nested NA, year-2011):
  • 0.5x0.67

• Updated emissions for all BVOCs, including non-standard tracers:
  • 2x25

• Monthly CO comparison for previous and updated MEGAN implementation (year-2006; 4x5 resolution):
  • 4x5

--Dbm 14:18, 22 January 2013 (EST)

CO2 direct effect on isoprene emissions

This update is slated for inclusion in GEOS-Chem v10-01 or later

Amos Tai developed code updates to include CO2 inhibition of isoprene emissions in MEGAN.

The reference for this work is:

Tai, A.P.K., L.J. Mickley, C.L. Heald, S. Wu, Effect of CO2 inhibition on biogenic isoprene emission: Implications for air quality under 2000-to-2050 changes in climate, vegetation, and land use, Geophys. Res. Let., 40, 3479-3483, 2013. [pdf]

--Melissa Sulprizio 15:45, 23 January 2014 (EST)

Previous issues that have been resolved

Bug in MEGAN emissions when running with MERRA or GEOS-FP

NOTE: This issue was resolved in 1-month benchmark simulation v9-02e (07 Jan 2013).

Tzung-May Fu wrote:

I wanted to report a small but important bug in the MEGAN emissions when running with MERRA and GEOS-FP. In subroutine READ_MET_FIELDS (in GeosCore/main.F, the MEGAN subroutine UPDATE_T_DAY was never called for MERRA or GEOS-FP, except when reading the 10-day period preceeding the simulation start date. The call line was commented out in MERRA, and missing in GEOS-FP. This can lead to a 10% error in the biogenic emissions (particular isoprene) when a simulation is run a month at a time. However, say you start your simulation in January, you would still be using the same temperature in July! This can lead to a 80% underestimate of the biogenic emissions. This bug is present in v911, v912, v913 and the v92 code that Melissa sent me.

GeosCore/main.F, in routine READ_MET_FIELDS:

        !%%% NEED TO UPDATE FOR MERRA %%%
        ! Update daily mean temperature archive for MEGAN biogenics
! (tmf, 1/4/2012) This should be turned on!
        IF ( LMEGAN ) CALL UPDATE_T_DAY

--Melissa Payer 11:07, 4 January 2013 (EST)
--Bob Y. 13:49, 20 August 2013 (EDT)

Unresolved issues

None at this time.

References

1. Barkley, M., Description of MEGAN biogenic VOC emissions in GEOS-Chem, 2010. PDF
2. Buermann, W., Wang, Y.J., Dong, J.R., Zhou, L.M., Zeng, X.B., Dickinson, R.E., Potter, C.S., and Myneni, R.B.: Analysis of a multiyear global vegetation leaf area index data set, J. Geophys. Res., 107, 4646, doi:10.1029/2001JD000975, 2002.
3. Guenther, A., Baugh, B., Brasseur, G., Greenberg, J., Harley, P., Klinger, L., Serca, D., and Vierling, L.: Isoprene emission estimates and uncertainties for the Central African EXPRESSO study domain, J. Geophys. Res., 104, 30625-30639, 1999.
4. Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P.I., and Geron, C.: Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols from Nature), Atmos. Chem. Phys., 6, 3181-3210, 2006.
5. Guenther, A., and C. Wiedinmyer, User's guide to the Model of Emissions of Gases and Aerosols from Nature (MEGAN), Version 2.01, 2007.
6. Millet, D.B., Jacob, D.J., Boersma, K.F., Fu, T.M., Kurosu, T.P., Chance, K., Heald, C.L., and Guenther, A.: Spatial distribution of isoprene emissions from North America derived from formaldehyde column measurements by the OMI satellite sensor, J. Geophys. Res., 113, D02307, doi:10.1029/2007JD008950, 2008. PDF
7. Mueller, J.-F., et al. Global isoprene emissions estimated using MEGAN, ECMWF analyses and a detailed canopy environment model, Atmos. Chem. Phys., 8, 1329-1341, 2008.
8. Myneni, R. B., et al., Large seasonal swings in leaf area of Amazon rainforests, Proceedings of the National Academy of Sciences, 104(12), 4820{4823, doi:10.1073/pnas.0611338104, 2007.
9. Palmer, P.I., Abbot, D.S., Fu, T.M., Jacob, D.J., Chance, K., Kurosu, T.P., Guenther, A., Wiedinmyer, C., Stanton, J.C., Pilling, M.J., Pressley, S.N., Lamb, B., and Sumner, A.L.: Quantifying the seasonal and interannual variability of North American isoprene emissions using satellite observations of the formaldehyde column, J. Geophys. Res., 111, D12315, doi:10.1029/2005JD006689, 2006. PDF
10. Sakulyanontvittaya, T., T. Duhl, C. Wiedinmyer, D. Helmig, S. Matsunaga, M. Potosnak, J. Milford, and A. Guenther, Monoterpene and Sesquiterpene Emission Estimates for the United States, Environ. Sci. Technol., 42(5), 1623{1629, doi:10.1021/es702274e, 2008.

--Bob Y. 09:39, 3 March 2010 (EST)