wiki.met.no

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Model Diagnostic Tools======

This page links to tools used for the NorESM model evaluation.

===== NorESM Diagnostic Packages =====


**NB: The wiki page for the NorESM diagnostic tools is moved!!**

Last updated: 12-Dec-2020.

Please go to the following NorESM documentation page for the latest description:

https://noresm-docs.readthedocs.io/en/latest/diagnostics/diagnostics.html
----


====== Aerosol and Chemistry, Clouds and Forcing Diagnostics ======


In both the default CAM5-aerosol packages (MAM3,MAM7) and the Oslo-aerosol packages, the budget terms can be taken out using a variable in the namelist : 

===Configuring a run with more aerosol diagnostics in (NorESM2) ===

<file>
&phys_ctl_nl
history_aerosol = .true.
/
</file>

Two more diagnostics are useful:
  * Enable estimates multiple calls to radiation which are necessary for effective radiative forcing estimates
  * Enable diagnostics for AEROCOM


To enable this, take the file cam/src/physics/cam_oslo$ vim preprocessorDefinitions.h and copy it to your SourceMods/src.cam folder

Change both preprocessor definitions to true

<file>
#define AEROCOM
#define AEROFFL
</file>

The AEROCOM-token turns on diagnostics needed for AEROCOM
The AEROFFL-token tells the model to do additional radiation-diagnostics for aerosol indirect effect

===== Tracer Budget terms ===== 

===Fields produced in monthly average files when running with budgets activated===

Running with budgets activated will produce the following terms in the monthly output files: 

^ Output variable name    ^ Meaning                      ^ Comment               ^
| SF{Tracer}              | Emissions from surface       |                       |
| GS_{Tracer}             | gas phase chemistry          | 3D-emissions and gas phase washout included in this term | 
| AQ_{Tracer}             | aquous chemistry             |                                                          |
| {Tracer}_Mixnuc1        | Activation in clouds and evaporation of cloud droplets |                                | 
| {Tracer}_DDF            | Dry deposition flux (aerosol tracers)                  |                                |
| {Tracer}_SFWET          | Wet deposition flux (aerosol tracers)                  |                                |
| {Tracer}_condtend       | loss/production in condensation/nuclation              | (CAM-Oslo only)                |
| {Tracer}_coagTend       | loss/production in coagulation                         | (CAM-Oslo only)                |
| DF_{Tracer}             | dry deposition flux (gas tracers)                      | output with history_aerosol with CAM-Oslo only|
| WD_A_{Tracer}           | wet deposititon flux (gas tracers)                     | output with history_aerosol with CAM-Oslo only|
| {Tracer}_CLXF           | 3D-emissions ("external forcing")                      | output with history_aerosol with CAM-Oslo only|
| {Tracer}_clcoagTend     | loss of tracer due to coagulation with cloud droplets  | output with history_aerosol with CAM-Oslo only|

Note: Since 3D-emissions and and gas washout rates are included in the term GS_{Tracer} in the mozart chemistry solver, the individual terms can be found like this (example for SO2): 
<file>
ncap2 -O -s GS_ONLY_SO2=GS_SO2-WD_A_SO2-SO2_CLXF infile.nc outfile.nc
</file>

More info on SO2 budgets (see /models/atm/cam/tools/diagnostics/ncl/ModIvsModII/ for scripts with info on all tracers): 

GS_SO2 contains the SO2 budget terms for all that goes on in the chemistry-routine, which is \\
1) Gas phase chemistry, 2) Wet deposition, and 3) 3D-emissions.\\
Gas phase chemistry is both production from DMS (GS_DMS) and loss through OH (GL_OH) \\
For calculations of net loss, e.g. used to calculate SO2 life-times, we're interested in the \\
loss through OH from the chemistry-term (GL_OH).\\
GS_SO2 = GL_OH + SO2_CLXF - WD_A_SO2 - GS_DMS*64/62 \\
or \\
GL_OH = GS_SO2 - SO2_CLXF + WD_A_SO2 + GS_DMS*64/62 \\

Estimating chemical loss w.r.t. S (instead of SO2 or DMS), for comparison with CAM4-Oslo numbers:\\
net chemial loss gas phase = (GS_SO2/1.998 - SO2_CLXF + WD_A_SO2)/1.998  + GS_DMS/1.938 \\
net chemical loss = net chemial loss gas phase + AQ_SO2/1.998 \\

Finally, total net loss (used to calculate life-time = -load/(net loss), where load = cb_SO2/1.998):\\
net loss = \\
- WD_A_SO2/1.998    ;wet deposition in kg/m2/sec (positive in output file) \\
- DF_SO2/1.998      ;dry deposition in kg/m2/sec (positive in output file) \\
+ AQ_SO2/1.998      ;wet phase production of SO4 in kg/m2/ses (negative in output file) \\
+ (GS_SO2 - SO2_CLXF + WD_A_SO2)/1.998 + GS_DMS/1.938  ; net chemical loss gas phase \\

===Looking at the aerosol budgets (CAM-Oslo only) ===

  * Go to the directory models/atm/cam/tools/diagnostics/ncl/budgets
  * Change the filename to use in the file budgets.ncl ("myFileName" around line 18). Should be for example yearly average of month-avg file in a run with budgets
  * Run the script budgets.sh to create a pdf-file (output.pdf)


===== NCL Model Version Comparison package (Alf K) ======

===Making ncl plots of often used aerosol and cloud fields, including ERFs, for two model versions (CAM-Oslo only) ===

  * Make a local copy (on Linux) of the directory models/atm/cam/tools/diagnostics/ncl/ModIvsModII
  * Assuming that you have produced output data from 4 simulations: two different model versions, each with PD and PI emissions, and all run with #define AEROCOM & AEROFFL:   
  * In ModIvsModII.csh (note: read the header info):
  * - edit model info for the first model (shown to the left in the plots): modelI = CAM4-Oslo or modelI = CAM5-Oslo ?
  * - provide paths and partial file names of the model data (PD and PI) for Model I (CAM4-Oslo or CAM5-Oslo) and Model II (must be CAM5-Oslo)
  * - choose desired plot format (plotf=ps, eps, pdf or png)
  * Run the script: ./ModIvsModII.csh
  * Furthermore, to display the plots in an organized form by use of a web browser (only possible if the chosen plot format is png):    
  * - download htm template files from ftp://ftp.met.no/projects/noresmatm/upload/NorESM2Diagnostics/ModIvsModII/htm-templates/
  * - edit general model info (only) in ModIvsModII.htm, and manually cut and paste the mass budget numbers from the script output into this file 
  * - copy all png (plots) and htm files to the desired output (common) directory
  * - open ModIvsModII.htm in your browser: hyper-links to all other htm files, including plots, are found here
  * Example: ftp://ftp.met.no/projects/noresmatm/upload/NorESM2Diagnostics/ModIvsModII/revision610inclSOA-Nudged_1984-12to1985-11_vs_CAM4-Oslo/ModIvsModII.htm

===== Cloud water mass and number analysis (budgets) ======


===Configuring a run with more cloud diagnostics in NorESM2 ===

To switch on extra output for cloud diagnostics (mass and number tendencies for liquid water and mass) change the following namelist variable:

<file>
&phys_ctl_nl
history_budget = .true.
/
</file>

A python script for plotting the mass and number budgets for the cloud microphysics can be found under:

models/atm/cam/tools/diagnostics/ncl/cloudBudgets 

in the same branch. Copy the script to your local computer or lustre and edit the script to read the correct input file(s) (instructions inside the script). Run the script by typing:

<file>
python scriptname.py
</file>

in your terminal.

===== Automatic AEROCOM analysis =====

To prepare output so that it is processed automatically by the aerocom tools, use the script located at **models/atm/cam/tools/aerocom/** in the svn repository. The script prepares files such that the idl aerocom tools prepare plots for the aerocom webinterface:
[[http://aerocom.met.no/cgi-bin/aerocom/surfobs_annualrs.pl?PROJECT=NorESM&MODELLIST=NorESM-development&FULL=explicit&INFO=nohover&PERFORMANCE=ind&YEARFILTER=ALLYEARS&PSFILTER=ALLVARS&Type0=ZONALOBS&Ref0=AERONETSun&Run0=CAM53-Oslo_r773bNudge_151215AG_PD_DMS_733b&Parameter0=OD550_AER&Station0=WORLD&Year0=an9999&Period0=mALLYEAR | URL link to NorESM on AeroCom webinterface]]

The script requires <ModelName>_<ExperimentName> and <Period> as input.

<Period>: for a climatological average and run choose 9999 , for nudged simulations choose the year of the meteorology

<ModelName>_<ExperimentName>: is the dataset identifier under which the plots appear on the AeroCom webinterface \\ 
in the required format 
**NorESM-CAM5_svn{RevisionNumber}_YYMMDD{initials}_Freetext**. \\

Example: 
"**NorESM-CAM5_svn1094_151201AG_CMIP6endelig**" \\
Initials AG: Alf Grini, AK: Alf Kirkevåg, DO: Dirk Olivie...

Where the date YYMMDD corresponds to the time when the  AeroCom data preparation script has been executed. 

The script creates files named like

“aerocom3_<ModelName>_<ExperimentName>_<VariableName>_<VerticalCoordinateType>_<Period>_<Frequency>.nc”

<ModelName> ⇒ eg NorESM-CAM53 \\
<ExperimentName> ⇒ svn{RevisionNumber}_YYMMDD{initials}_Freetext \\
<VariableName> ⇒ aerocom variable names \\
<VerticalCoordinateType> ⇒ “Surface”, “Column”, “ModelLevel”, “SurfaceAtStations”, “ModelLevelAtStations” \\
<Period> ⇒ eg “2008”, “2010”, "9999" \\
<Frequency> ⇒ “timeinvariant”,”hourly”, “daily”, “monthly”, “sat1000”, “sat1330”, “sat2200”, “sat0130” \\

Note that VerticalCoordinateType is dependent on the variable!! It is not a question about "vertical coordinate type used in model simulations"!

The script copies files on norstore into
**/projects/NS2345K/CAM-Oslo/DO_AEROCOM/<ModelName>_<ExperimentName>/renamed/**

====== ESMval CIS JASMIN platform and tools ======

ESMVALtool
http://www.geosci-model-dev-discuss.net/8/7541/2015/gmdd-8-7541-2015-discussion.html

cis tools
http://www.cistools.net

JASMIN
http://www.jasmin.ac.uk/services/jasmin-analysis-platform/

====== Post analysis and workup of CAM diagnostics output tables ======

A tool for post analysis of (multiple) CAM diagnostics ASCII tables can be found in 
the following repository:

GitHub https://github.com/jgliss/noresm_diag_postproc

To get started, please follow the instructions in repository README (displayed in repository). Currently, the main 
analysis tool is a jupyter IPython notebook called 

//**analysis_tool.ipynb**// (https://github.com/jgliss/noresm_diag_postproc/blob/master/analysis_tool.ipynb)

which includes more detailed instructions about setup and options. 

Use the notebook 

https://github.com/jgliss/noresm_diag_postproc/blob/master/download_tables.ipynb

to download local copies of result tables using a list of URL's.


**Short summary:** 
The notebook reads multiple diagnostics files (runs) into one long table and creates heatmap plots
of //Bias, RMSE and RMSE relative error// for a subset of variables (rows -> y-axis of heatmap) 
vs. the individual runs (columns -> xaxis). 

**NOTE:** In the current version, you need to download all tables that you are interested in as csv or 
ascii into one directory, that is specified in the header of the notebook.

Variable groups can be defined in this config file:

https://github.com/jgliss/noresm_diag_postproc/blob/master/config/var_groups.ini

**NOTE:** If you add groups to this file in your local copy of the repository, please consider sending
the updated to [[mailto:jonasg@met.no|jonasg@met.no]] or to submit a pull request, so that the remote repository 
remains up to date.

**Troubleshooting**

If you run into problems, please raise an issue in the repository or contact [[mailto:jonasg@met.no|jonasg@met.no]]