GEOS-Chem Newsletter Fall 2011

Dear GEOS-Chem Users,

We apologize for not getting a newsletter out sooner. We have therefore combined several news items into a Fall 2011 newsletter.

Sincerely,

The GEOS-Chem Support Team
geos-chem-support@as.harvard.edu

GEOS-Chem Support Team

Here is some news regarding the GEOS-Chem Support Team:

Overview

1. Bob Yantosca recently began working on the code to process GEOS-5.7.2 data. He also oversees general GC development and the development of the Grid-independent GEOS-Chem project.
2. Michael Long is primarily working on the Grid-independent GEOS-Chem project.
3. Melissa Payer has completed merging Havala Pye's updates for SOA + nonvolatile POA into GEOS-Chem, which we will add to version v9-01-03. She has also been working on general GC development tasks (i.e. code merging, generating 1-month and 1-year benchmark simulations, user support, documentation, etc.).
4. Matt Cooper (Dalhousie) has been working on regridding the MERRA met data to 2 x 2.5 degree resolution (see below). He has also begun replacing the existing regridding routines in GEOS-Chem with the MAP_A2A regridding package (which is what is currently use for regridding the GMAO met fields).

New Support Team position

GEOS-Chem Support Team/ Harvard University. A position is available for a support scientist, scientific programmer, or research assistant to join the GEOS-Chem Support Team in the Atmospheric Chemistry Modeling Group at Harvard (http://acmg.seas.harvard.edu and http://geos-chem.org). GEOS-Chem is a global 3-D model of atmospheric composition used by over 60 research groups worldwide for a very wide range of applications. Tasks will include model development and documentation, targeted research projects, and user support. The position may be filled at the Ph.D., M.S., or B.S. level. A degree in atmospheric science, applied mathematics, or engineering is required. Experience in scientific programming including knowledge of FORTRAN and UNIX is a must. Please email a CV, contact information for three references, and 1-paragraph statement of interest to (Prof. Daniel J. Jacob), School of Engineering and Applied Science, Harvard University. Harvard University is an equal opportunity / affirmative action employer.

Release of GEOS-Chem v9-01-02

We are happy to announce the official release ofo GEOS-Chem v9-01-02. This version is the first GEOS-Chem version to be rigorously tested with our new benchmarking protocol.

What's new in GEOS_Chem v9-01-02?

GEOS-Chem v9-01-02 contains some important new updates:

1. GFED3 biomass burning emissions
   • NOTE: at this time, only the monthly mean GFED3 product is available.
2. RETRO anthropogenic VOC emissions inventory
3. Now use 0.03 cm/s for aerosol drydep velocity when over ice and snow
4. Implemented seasonality for NH3 emissions in Streets 2000 inventory
5. Overhaul of aerosol optical depth diagnostics to remove inconsistencies
6. Removal of legacy code constructs to facilitate Grid-independent GEOS-Chem model development

as well as a host of bug fixes. In addition, many new features in the offline Hg simulation have been standardized, including GEIA anthropogenic Hg emissions for 2005. For more information, please see the GEOS-Chem v9-01-02 wiki page.

We recommend that all GEOS-Chem immediately upgrade to GEOS-Chem v9-01-02. For downloading instructions, please visit Chapter 2 of the GEOS-Chem Online User's Guide.

Also note that there are several new new data directories and files that need to be downloaded before using GEOS-Chem v9-01-02.

Validating GEOS-Chem v9-01-02

We tested GEOS-Chem v9-01-02 rigorously according to the new benchmarking protocol developed at the 5th International GEOS-Chem Meeting (aka IGC5).

1-month benchmark simulations

Every time a significant change was added to the source code, the GEOS-Chem Support Team performed a 1-month benchmark simulation. The person who initiated the source code change had to examine the 1-month benchmark output plots (which are posted on Harvard's FTP site) and fill out a Benchmark Assessment Form. Final approval for each 1-month benchmark was given by Model Scientist Daniel Jacob (after the GEOS-Chem Steering Committee also had a chance to examine the benchmark output).

Please see our GEOS-Chem v9-01-02 benchmark history wiki page for a complete list of 1-month benchmarks that were done during the validation process.

1-year benchmark simulations

In accordance with the new benchmarking protocol, the GEOS-Chem Support Team also performed two 1-year NOx-Ox-HC-aerosol full-chemistry simulations. Each simulation used GEOS-5 met fields at 4° x 5° resolution for the year 2005 (with a 1-year spinup):

1. v9-01-02-Run0 (with RETRO and GFED3 emissions turned off)
2. v9-01-02-Run1 (with RETRO and GFED3 emissions turned on)

Comparing v9-01-02-Run1 to v9-01-02-Run0 is a clean comparison of the impact of the RETRO and GFED3 emissions inventories.

To view the output plots from each of these 1-year full chemistry simulations, point your browser to:

ftp://ftp.as.harvard.edu/gcgrid/geos-chem/1yr_benchmarks/v9-01-02/geos5/2005/Run0/output/pdf/

ftp://ftp.as.harvard.edu/gcgrid/geos-chem/1yr_benchmarks/v9-01-02/geos5/2005/Run1/output/pdf/

Rn-Pb-Be benchmark simulations

In addition to the 1-year full-chemistry benchmark simulations, the GEOS-Chem Support Team ran a 1-year Rn-Pb-Be benchmark simulation. Hongyu Liu computed the budget of Pb²¹⁰ and the [[Rn-Pb-Be simulation#Budget_of_Be7budget of Be⁷]] from these simulations.

Helen Amos ran several Rn-Pb-Be simulations GEOS-Chem order to investigate differences in wet scavenging when using GEOS-5 vs. MERRA met fields. We invite you to view the Comparison plots and comparison budgets that Helen created.

= Unresolved issues wiki page

We have created a

Numerical noise in ISORROPIA II

Daven Henze noted that the ISORROPIA II aerosol thermodynamic equilibrium package does not produce the same results on single processor as on multi processor. (This was a direct result of doing more frequent 1-month benchmarks as specified by the new benchmarking protocol).

Daven writes:

There is an unidentified OpenMP error which contributes to differences shown [in the v901-02a 1-month benchmark], as confirmed by independent tests of consecutive evaluations of a single executable compiled with and without parallelization. The impacts of this OpenMP error are typically less than a fraction of a percent, and negligible in terms of absolute value, with the exception of NIT and NH3. This is an issue to investigate further.

Bob Yantosca did extensive testing, which revealed that the numerical noise caused by ISORROPIA creeps into GC full-chemistry simulations of longer than about 2 weeks.

At present, this issue is still under investigation by Shannon Capps at Georgia Tech. The GCST will no longer look into this issue. If a new version of ISORROPIA is provided, then the GCST will install that into GEOS-Chem.

Recommendation of the GCST:

1. Most GC users will not have to do anything. The numerical noise is small enough that it should not affect the results of long simulations.

2. However:

a. If you are doing structural coding (i.e. rewriting source code but not changing any science), or

b. If you expect to see identical results before and after your updates have been made,

then you should turn off ISORROPIA during your testing & debugging phase. Otherwise, the numerical noise will make it difficult to assess if your own source code updates are functioning properly.

Grid-Independent GEOS-Chem

Grid-Independent GEOS-Chem Mike Long recently returned from a trip to GSFC. He consulted with Arlindo da Silva to begin the Grid-Independent (GI) GEOS-Chem code implementation within the GEOS-5 GCM. The ESMF sandbox test code that Arlindo and Bob Yantosca had developed will provide the template for testing and ultimate completion of the Grid Independent GEOS-Chem at Harvard. This will end up saving us a lot of time. The GI code is being bootstrapped into the main source through preprocessor switches. Unless the switches are intentionally activated, the user community SHOULD never have to interact with the GI code modifications. Furthermore, most or all of the GI coding and ESMF interface will be at a high-level such that it will be invisible to scientific developers. There will be no GI functionality in the v9-01-02 release (although several modifications have been made in v9-0102 in order to better implement Fortran-90 coding standards into GC).

The advantage of using preprocessor switches to block off GI-specific code is that this approach will allow GC users to decide whether they want to invoke the ESMF and MPI functionality or not. During the transition to the GI model, some GC users (especially those who utilize the simpler offline chemistry simulations) may wish to continue running GC in serial mode (i.e. with OpenMP) and without having to install ESMF or MPI libraries. Our approach will make this possible and will enable GC users to continue their research without interruption.

Here is a list of ongoing tasks and their state-of-completion: 1. Emissions: The GI treatment of emissions routines is nearly complete. In the development code, all emissions will be stored in a single emissions tendency array with global scope. Emissions routines will no longer interact with the tracer (STT) or species (CSPEC) arrays directly, but will instead update the appropriate slot of the emissions tendency array. The emission tendency array will then be passed to the chemical solver setup routine (setemis.F) so that the emissions can be entered into the reaction rate matrix. Also, the units of all emissions will be standardized. 2. Chemistry: The primary development effort in the chemistry routines is tied into reading and writing to/from generalized tracer & species arrays. This is progressing in tandem with Task (1). However, restructuring the spaghetti-esque legacy code routines has proven to be challenging. The work is ongoing. 3. ESMF wrapping: Development of the ESMF interface wrappers is ongoing and the system is not yet capable of a full 'sandbox' test simulation. This is due in large part to the state of Tasks (1) and (2) listed above. The ESMF wrapper code is, as stated above, drawing heavily from column-model work and it already has matured somewhat. There are a few major tasks that remain: 1. File I/O libraries: These will be necessary for emissions files, history & diagnostic file writing, met fields, etc. for both the GEOS-5 coupled and the MPI-Parallel GEOS-Chem CTM. ESMF provides good library tools for this, while Bob & Philippe Le Sager had worked on some NetCDF routines that were not implemented; but are proving useful. 2. Transport & Physics: It is becoming clear that we may be able to preserve the current structure of the physics and transport routines for use by GEOS-Chem as an MPI-parallel model. We won't make any predictions about scalability at this time. General timeline to completion:

1. GI standalone (not parallel): The GI code may be in working form by the end of November or December. We do not expect to pass out any of the interim code to scientific developers until after completing a passing benchmark. Once the GI code modifications have proven to be mature, it will be incorporated into the standard repository (possibly in time for the v9-01-03 release).

2. GI ESMF-wrapped: We expect this to be completed by mid-2012.

3. GI stand-alone parallel using ESMF: This is not a priority at this time; therefore we

cannot make a prediction as to when this will be completed. This task is contingent on the completion of the standalone GI (Task 1 above) and the ESMF wrapping (Task 2 above).

Met field update

GEOS-5.7.2 data product

GEOS-5.7.2 met field update

GMAO recently updated to the GEOS-5.7.2 data product. This data will be 0.25

lon horizontal resolution. This will replace the older GEOS-5.2.0 product, which was 0.5

lat x

0.666 o

lon horizontal resolution.

The file naming convention, timestamps, etc. used by GEOS-5.7.2 data are very similar to what is used in the MERRA data set. Therefore, the MERRA data processing software can be used as a starting point. However, it should be noted that the data file format for GEOS-5.7.2 is netCDF4, whereas previous GEOS data products used HDF4-EOS. This will require some rewriting of source code.

Milestones:

1. 18 Jul 2011: GMAO released the file specification document for the GEOS-5.7.2 (v1.2) data. This document describes the file naming structure, time stamp structure, and which fields are contained in which files.

2. 22 Aug 2011: GMAO sent sample GEOS-5.7.2 data to Harvard in order to test the file transfer.

3. Late August 2011: GMAO turned on the live GEOS-5.7.2 data stream.

4. Late October 2011 (need to confirm date): GMAO turned off the GEOS-5.2.0 data stream.

At present, Bob Yantosca has completed the following tasks:

1. Created pages on the GC wiki with information about the GEOS-5.7.2 data.

2. Contacted GSFC to establish a data subscription for the GEOS-5.7.2 data.

3. Installed netCDF4 and HDF5 libraries locally. These are necessary for reading the GEOS-5.7.2 raw data files from disk. The netCDF4 library is an update from the previous (and more commonly used) netCDF-3 library. netCDF4 requires the HDF5 library to also be installed.

4. Written scripts to transfer GEOS-5.7.2 data from NASA to Harvard and to automate the data download process.

5. Created a new Git repository for GEOS-5.7.2 data processing code and scripts.

6. Completed initial source code preparation and documentation. The source code is based on the existing data processing code for MERRA, with modifications for netCDF dat

I/O. Regridding will be done with the MAP_A2A algorithm (as was used for the regridding of previous GMAO data products).

We plan to store the GEOS-5.7.2 following data at Harvard (which will be downloadable by the community). Data will be processed for Aug 2011-onward. Priority will be given to these grids:

1. Global 2 x 2.5

2. Global 4 x 5

3. SEA4CRS Nested grid 0.25 x 0.3125 : 20S-50N; 60E-150E

Consistent with our established nested grid protocol, we will distribute the GEOS-5.7.2 data processing code & scripts to any user groups who wish to pull the nested data themselves.

GEOS-5.2.0 data product

The current GEOS-5 data product, aka GEOS-5.2.0, (at 0.5° x 0.666° resolution) will be produced concurrently with GEOS-5.7.2 for approximately two months. After this time, the GEOS-5.2.0 data stream will be turned off and only GEOS-5.7.2 data shall be produced going forward.

Please note: Presently, the Harvard data server (ftp.as.harvard.edu) contains GEOS-5 met data thru July 31, 2010 regridded to the following GEOS-Chem resolutions:

1. 4° x 5°
2. 2° x 2.5°
3. 0.5° x 0.666° nested China
4. 0.5° x 0.666° nested North America

You may download data for August 1, 2010 and onwards from the Dalhousie data server (rain.ucis.dal.ca). In addition, the Dalhousie server contains met data for the 0.5° x 0.666° Europe nested grid.

MERRA data product

The MERRA data product is still being produced. We have currently processed the 30-year data archive (1979-2010) for GEOS-Chem to 4° x 5° resolution. Matthew Cooper (Dalhousie U.) is working on regridding MERRA to 2° x 2.5° resolution.

Several of you have asked us about how to obtain the "raw" 0.5° x 0.666° MERRA met field data for your own research projects. NASA's GES-DISC department has created a user interface from which you can request "subsetted" data files (i.e. data files that have been trimmed down to a specific geographic region and/or contain a reduced number of data fields). You can then download the subsetted files via FTP file transfer to your own disk space. For more information, please see the NASA GES-DISC user interface for requesting subsetted MERRA data.

GEOS-Chem reference

Website

Please take a moment to review your group's entry on the GEOS-Chem People and Projects web page. Kindly inform the GEOS-Chem Support Team if anyone has recently joined or left your research group.

We invite you to consult our Credits and References web page for you are unsure how to give credit to GEOS-Chem developers.

Wiki

We encourage all GEOS-Chem code developers and users to check the wiki frequently, as this is the place where the latest information about GEOS-Chem will be posted. The wiki is designed to be a two-way street of communication. Users should feel free to add content to the wiki pages that are most closely related to their research.

In particular, the GEOS-Chem Support Team has started a new effort to make sure that all 3rd-party code and data that is submitted into GEOS-Chem has a corresponding wiki page. This will ensure that all information can be shared transparently.

Please see the Recent Changes wiki page to see which wiki pages have been updated.

Finally, if you have not already done so, consider signing up for automatic wiki notification. You will be updated whenever someone modifies the GEOS-Chem wiki...you'll never miss a discussion!

Subscribing to the GEOS-Chem email lists

We use several email lists to share information among the GEOS-Chem user community:

• We send information about new GEOS-Chem releases, bug fixes, and other notices of importance to the the GEOS-Chem "general" email list (geos-chem@seas.harvard.edu)
• Each GEOS-Chem Working Group uses its own email list to exchange information between group members.

We invite you to join the Working Group that is most appropriate to your research, and to subscribe to the relevant email list. Consider joining more than one Working Group (and email list) if your research spans several topics.

For complete instructions on how to subscribe to (or unsubscribe from) the GEOS-Chem email lists, please see our Subscribing to the GEOS-Chem email lists wiki page.

Search engines

We have set up several GEOS-Chem search engines which will let you search for information in the GEOS-Chem website, wiki, and manual pages.

You can also search the GEOS-Chem wiki by means of the search box in the left column. It is often more efficient to search for a topic rather than to follow links down from the main page.

--Bob Y. 15:02, 4 August 2011 (EDT)