dev_ioadj: Adjusted output settings for production runs.

See merge request !14

Modified registry files for most output changes.
Added run/iofieldmod_cactiles.txt for handling differenes between meso and LES domains.

Implements the PBL tracer that switches on at a given time and below a given...

See merge request !13

Implements the PBL tracer that switches on at a given time and below a given height above the surface.

bugfix: LASSO tracer emission and registry issues

See merge request !12

1) Fixed issue with ptracer4 accidentally getting emissions in addition to ptracer3.

2) Moved LASSO tracer changes within the registry back to Registry.EM from Registry.EM_COMMON. This seems clumsy, but it solves an issue with the order of the tracer 4-D array definitions for the initial blank entry.

3) Capitalized names of PTRACER variables for the wrfout files.

4) Switched checks for tracer_opt to use LASSO_TRACER instead of a number.

Inert tracers for LASSO-CACTI

See merge request !11

Resolved conflicts for Registry.EM_COMMON and module_em.F.

Moves the package and variable declarations to registry.EM_COMMON following Bill's suggestion and changes the tracer_opt value we use from 4 to 5 to avoid comflicts.

module_diffusion_em.F (from kg/kg m/s to kg/m^2/s); make very little
difference in terms of normalized anomalies.

(2) adds the same constant surface tracer flux at the surface for mesoscale domains in
module_pbl_driver.F in `diff4d`.

Merge "merge-ncar4.3.1-into-main" into "main" (had to use intermediary branch to handle conflicts from "main_ncar")

See merge request !9

Fixed conflicts for merging NCAR v4.3.1 into main. Using a side-branch
to handle this, which will then get merged into main.

Note that the changes to the handling of surface emissions for water and
ice-friendly aerosol for Thompson were not accepted because they did not
seem to play well with the handling of these variables when pulled from
GEOS-5. This may need to be addressed later if non-GEOS-5 approaches are
used for QNWFA and QNIFA.

First commit with two passive tracers with linear profiles and another surface emitted one. Have not added the PBL/subcloud-layer tracer yet.

cactiOutput: No nest interactions (du) for Thompson diagnostics

See merge request !8

Removed the "du" I/O settings in the registry for the new Thompson
diagnostic variables since they do not need to be transferred between
nested domains.

cactiOutput: Output variable adjustments for LASSO-CACTI

See merge request !7

Added output for z-component of wind total tendency for the
instantaneous value at the output time. This is calculated from the
simple difference between w at beginning and end of the time step.

Adding first batch of adjustments to wrfout output variables for
LASSO-CACTI. This it the bulk of them. Still missing are the handling of
buoyancy, vertical pressure gradient tendency, and/or vertical velocity
tendency, as well as inert tracers. Tracers will be handled under a
different branch.

ThompsonMPOutput: Restart and name fixes for new microphysics diagnostic output & MYJ surface scheme

See merge request !5

ThompsonMPOutput: Restart and name fixes for new microphysics diagnostic output & MYJ surface scheme

aerosolWRFsolar: Merging changes to ingest GEOS5 aerosol info for Thompson microphysics

See merge request !3

Finalize WRFV4.3.1 by merging fixes from release-v4.3.1 branch onto master.

TYPE: text only

KEYWORDS: README file, version_decl

SOURCE: internal

DESCRIPTION OF CHANGES:
Update the top-level README file, and the inc/version_decl file with the correct version.

LIST OF MODIFIED FILES:
M README
M inc/version_decl

TESTS CONDUCTED: NA

TYPE: bug fix


KEYWORDS: module_stoch, RAND_PERT_UPDATE arguments

SOURCE: internal

DESCRIPTION OF CHANGES:
Problem:
In the initialization of the rand_perturb=1 option, the arguments to RAND_PERT_UPDATE were wrong.
The results would be that initially the stochastic perturbation had no cutoff standard deviations, but during 
the run this would have been corrected. Effect likely minor and generic rand_perturb option not used much.

Solution:
Correction so that argument list is same as from the call in first_rk_step_part2

LIST OF MODIFIED FILES: 
M       dyn_em/module_stoch.F

TESTS CONDUCTED: 
1. mod has been tested - Judith Berner is OK
2. Jenkins is OK

RELEASE NOTE: 
Fix in initialization for rand_perturb=1 option for the stochastic capabilities, due to an incorrect argument list. The effect is minor.

Local variables in noahdrv require initialization.

TYPE: bug fix

KEYWORDS: uninitializaed local variables in noahdrv, soilw, flx4, fvb, fbur, fgsn

SOURCE: internal

DESCRIPTION OF CHANGES:
Problem:
MPAS tests revealed that Noah LSM variable SMSTAV does not give identical results when run using different number of processors. Note that this variable isn't in the standard wrfout file, and it was detected in the restart file comparison. Note also that this has no impact to model results, since SMSTAV is only a diagnostic variable.

Solution:
The problem stems from the local variable, soilw, which sets the values for SMSTAV. This local variable is not initialized in the i,j loops before it is used (it is outside the do loops). Initializing this variable to 0 fixes the problem. The same problem is also fixed in the LSM mosaic option. Similarly a few local variables used for ua_phys option, flx4, fvb, fbur, fgsn, should be initialized inside the i,j loops as well, and hence it is done here too, though no known problem is caused by these variables at this time. 

ISSUE: For use when this PR closes an issue.
Fixes #868

LIST OF MODIFIED FILES:
M phys/module_sf_noahdrv.F

TESTS CONDUCTED:

Tests show the problem is fixed.
The Jenkins tests are all passing.

Release Note: N/A.

TYPE: bug fixes along with general P3 update (update from P3v4.0.7 to P3 v4.2.2)

KEYWORDS: P3, microphysics

SOURCE: Hugh Morrison (NCAR)

DESCRIPTION OF CHANGES:
Problem: 
Bug in solve_em.F so that 3-moment P3 (MP option 53) was not working properly. Some additional minor bug fixes to the 
P3 module and lookup tables (see below) as well as code clean up (affecting MP options 50, 51, 52, and 53). Also 
updated the corresponding two lookup table generation codes (which do not affect WRF, but are included for users as 
a reference). This also required changing the head Makefile to update the 3-moment lookup table file name. Rather than 
back porting these minor bug fixes to the older version of P3 module (v4.0.7, in the release WRFv4.3.1), we have just 
swapped to a newer version of P3 (v4.2.2). Thus, it includes some other minor changes and clean up as well.

Solution:
Uncommented 1 line in solve_em.F in the call to microphysics driver so that 3-moment P3 is runs properly. Made 
several changes to the P3 module and all three P3 lookup tables to address minor bugs: 1) ice self-collection (need 
to multiply the normalized lookup table rate by nitot twice not once), 2) minor fix to ice particle area-size relationship 
in lookup table 1 generation code for ice self-collection, 3) fix to ice fallspeed array structure for collection between 
categories in lookup table 2, 4) remove extra factor of air density for collection between categories in P3 module. These 
fixes required modifying the head Makefile for the correct 3-moment lookup table file name. Two small tunings were 
also applied to compensate for the bug fixes: 1) reduction of aggregation efficiency at low temperatures (supported by 
lab studies in Connolly et al. 2012), 2) increase in parameter "nmltratio" from 0.5 to 1. Finally, we use a different 
diagnostic mu-lambda relationship for ice (detailed in Milbrandt et al. 2021, JAS, see below) which allows for a relaxed 
upper mean size limiter, improving the representation of large rimed ice. This affects the lookup table calculations for 
2-moment P3 (MP options 50, 51, 52) but not 3-moment P3 (option 53).

Impacts of all the above changes are minor overall -- testing (quasi-idealized squall line and supercell cases, and a real 
case using the GEM model with same P3 configurations) reveals the main impact is an increase in reflectivity above the 
melting level in stratiform precipitation regions.

LIST OF MODIFIED FILES: 
Modified: 
dyn_em/solve_em.F
phys/module_mp_p3.F
Makefile

Added: 
p3_lookupTable_1.dat-5.3-3momI.gz
p3_lookupTable_2.dat-2momI_v5.2.2
p3_lookupTable_1.dat-5.3-2momI
create_p3_lookupTable_1.f90-v5.3
create_p3_lookupTable_2.f90-v5.2.2

Deleted: 
create_p3_lookupTable_1.f90-v5.1.6_b01
p3_lookupTable_1.dat-2momI_v5.1.6_oldDimax
p3_lookupTable_1.dat-3momI_v5.1.6.gz
p3_lookupTable_2.dat-4.1
create_p3_lookupTable_2.f90-v4.1

TESTS CONDUCTED: 
1. Idealized tests confirm the code is working. Additional real-case tests using the GEM model and same P3 module + 
lookup tables confirm reasonable results.

2. A new suite of tests using this modified P3 scheme (mp_physics=53) has been conducted. It was found that the large 
temperature bias in WRFV4.3 has been corrected. The results are improved and more consistent with the old P3 scheme 
( mp_physics = 51). Attached is the verification of the new tests  and also the results from WRFV4.3.    

![image](https://user-images.githubusercontent.com/17932265/137028700-677eafea-b972-45a1-8890-ea895f22d3df.png)

![image](https://user-images.githubusercontent.com/17932265/137028746-eaf744e8-8d92-4b62-9dec-7a6d93ac0256.png)

3. Jenkins tests passing.

RELEASE NOTE: Reference for the new mu-lambda relationship for ice in the P3 microphysics scheme is in:
Milbrabndt, J. A., H. Morrison, D. T. Dawson III, and M. Paukert, 2021: A triple-moment representation of ice in the Predicted Particle Properties (P3) microphysics scheme. J. Atmos. Sci., 78, 439-458.

TYPE: bug fix

KEYWORDS:  aerosol aware Thompson scheme

SOURCE: internal

DESCRIPTION OF CHANGES: 
Problem:
When running a nested case with the aerosol aware Thompson scheme, the surface aerosol number concentration 
emissions (QNWFA2D) differ by almost one order of magnitude between the parent and child domains (even when using
a recommended nest ratio such as 3:1). 

Solution:
This PR fixes the problem by removing the grid-size dependence of airmass, but still keeping the layer thickness 
dependence.

LIST OF MODIFIED FILES:
M   dyn_em/module_initialize_real.F
M   phys/module_mp_thompson.F

TESTS CONDUCTED:
1. A nested case with the original code is run to demonstrate the large difference in QNWFA2D between the parent 
and child domains (results shown with left figure).
2. The same nested case is done with the modified code. These results show a more reasonable and consistent 
behavior between the two domains (results shown with right figure).
3. Jenkins test is all pass.

![image](https://user-images.githubusercontent.com/17932265/128957778-62eba2d9-809c-4c00-821f-c315de7fa2dd.png)

RELEASE NOTE: Since V4.0 the aerosol aware Thompson (mp=28) option has not been initializing the surface aerosol source array consistently between a parent and nested domain due to grid-size dependence. This has been removed and the results will change if dx is not 20 km, but now nests will have the same source.

TYPE: enhancements

KEYWORDS: BEP+BEM, ROOFTOP MITIGATION STRATEGIES, WRF-URBAN, PHOTOVOLTAIC PANEL

SOURCE: Andrea Zonato (University of Trento)

DESCRIPTION OF CHANGES:
Problem:
The previous model, based on Salamanca et al 2016, was wrong and more simple than the one presented here.
In the radiation balance, the term regarding the reflected long-wave radiation should not be included since it does not contribute to the total energy balance. 

Solution:
We added a prognostic equation for the temperature of the Photovoltaic panels based on Jones and Underwood, 2002, avoiding to account of the reflected long-wave radiation in the sources.

LIST OF MODIFIED FILES: 
M phys/module_sf_bep_bem.F
M phys/module_sf_bem.F

TESTS CONDUCTED: 
1. Jenkins tests are all passing.

RELEASE NOTE: The photovoltaic panel (PVP) parameterization scheme, new in version 4.3, has been updated, since the previous one provided unreasonable results due to wrong calculation of the radiation balance. While before a simple energy balance was adopted to calculate the PVP temperature and related heat fluxes, now a prognostic equation is solved, accounting for the thermal coefficients of the photovoltaic panels.

REFERENCES:
Salamanca, F., Georgescu, M., Mahalov, A., Moustaoui, M., & Martilli, A. (2016).
Citywide Impacts of Cool Roof and Rooftop Solar Photovoltaic Deployment on Near-Surface Air Temperature and Cooling Energy Demand. Boundary-Layer Meteorology, 161(1), 203–221. https://doi.org/10.1007/s10546-016-0160-y

Jones, A. D., & Underwood, C. P. (2002). A thermal model for photovoltaic systems. Fuel and Energy Abstracts, 43(3), 199. https://doi .org/10.1016/S0140-6701(02)85831-3

TYPE: bug fix
    
KEYWORDS: maxtimes, 10000, climate runs, LBC
    
SOURCE: internal, reported by David Carvalho (CESAM-Aveiro University, Portugal)
    
DESCRIPTION OF CHANGES:
Problem:
When processing a LARGE number of time periods in a single lateral boundary file, the WRF
model fails. The failure is always just after time period 9999 is processed. This points to
a likely culprit of 10000 being a maximum number of time periods. Sure enough, there is a
variable named `MaxTimes` in a few of the netcdf IO directories, where each is currently set to 
this value of 10000.
    
Solution:
Bump this fixed number of maximum times to a larger fixed number that is sufficient to handle 
proposed regional-climate simulations at a typical data frequency (20 years at 3-h intervals).
    
ISSUE:
Fixes #1383
    
LIST OF MODIFIED FILES:
modified:   external/io_netcdf/wrf_io.F90
modified:   external/io_pio/wrf_data_pio.F90
modified:   external/io_pnetcdf/wrf_io.F90
modified:   share/input_wrf.F
    
TESTS CONDUCTED:
1. The user who originally saw the problem indicates that this fixes the problem. Additionally,
the fix works for not only the LBC files, but for the FDDA and SST update files.
2. All jenkins tests are a PASS.
    
RELEASE NOTE: For climate-length WRF simulations, the restriction of 10000 time periods in a single file has been increased to 60000 (a value that allows 3-hourly data for 20 years).

These edits incorporate the remaining changes from wrf-solar that are relevant to LASSO.
Changes not ported into lasso-wrf-cacti include the aerosol-radiation interactions and
shallow cloud parameterization changes. Note that the black carbon emissions and their
treatment within Thompson have been ported, but they do not impact the cloud activation,
so they are not really necessary for LASSO. Thompson only does removal of the BC to help with
the aerosol-radiation interactions.

New/modified namelist variables associated with GEOS-5 aerosol ingest and use:
qna_update = update aerosol from wrfqnainp file  0=no, 1=yes
use_aero_icbc = (now an integer instead of logical) use 3D aerosols IC/BC data in Thompson-MP-Aero, 0=no, 1=climatology, 2=first guess
do_fire_emit = activate biomass burning emissions
do_fire_inj = vertically distribute biomass burning emissions
wif_input_opt = water and ice-friendly aerosol processing: 0=do not process the Water Ice Friendly Aerosol input from metgrid, 1=Only Water/Ice Friendly Aerosol, 2=Water/Ice Friendly and Black Carbon Aerosol

Other namelist settings to turn on:
auxinput17_inname                   = "wrfqnainp_d<domain>"
io_form_auxinput17                  = 2
auxinput17_interval                 = 180, 180, 180, 180

So far, we can run real.exe to generate the aerosol initial conditions
and boundary conditions. This requires the associated changes to WPS in
lasso-wps-cacti. The code for these new features is coming from
wrf-solar.

TYPE: bug fix

KEYWORDS: BEP+BEM, longwave radiation, TSK

SOURCE: Andrea Zonato (University of Trento, Italy)

DESCRIPTION OF CHANGES:
There was an error in the previous version on the calculation of long-wave radiation. The variable rl_emit now accounts for the emissivity by the photovoltaic panel. Moreover, the variable rld, the downward longwave radiation flux, was incorrectly calculated. This will affect the calculation of TSK, in urban areas, that in the previous version was wrong. The fix reduces the predicted skin temperature by about 20+ K in a June case over Italy.

LIST OF MODIFIED FILES:
M phys/module_sf_bep_bem.F

TESTS CONDUCTED:

This error affects the calculation of TSK, which now is correctly calculated.
Jenkins tests are OK.

RELEASE NOTE: There were two errors in the previous version (v4.3) of solar panel physics on the calculation of long-wave radiation, which can cause large positive bias in skin temperature (TSK) in urban areas. Both are fixed with this PR.

TYPE: bug fix

KEYWORDS: snow radar reflectivity

SOURCE: Ted Mansell (NOAA/National Severe Storms Lab)

DESCRIPTION OF CHANGES:
Fix:
Restored previous formulation of snow reflectivity (before version 4.1.3), as it was realized that the last change incorrectly assumed a fixed density independent of size. Generally lower snow reflectivity values as a result (no effect on microphysics).

Other:
Generic fall speed coefficients (axx,bxx) to accommodate future frozen drops category (no effect).
Reordered collection coefficients (dab1lh) to be consistent (no effect).
Switched to full calculation of rain number loss via collection by graupel (chacr) to be consistent with collection by hail (chlacr) (minor effects).

LIST OF MODIFIED FILES:
M phys/module_mp_nssl_2mom.F

TESTS CONDUCTED:

Model tests to confirm snow reflectivity and minimal impact from graupel-rain collection (chacr).
Are the Jenkins tests all passing? Yes

RELEASE NOTE: For the NSSL 2 moment MP scheme (mp_physics=17, 18, and 22), reverted formulation of snow radar reflectivity back to version (pre-4.1.3) using size-dependent snow density. The diagnosed bright band is still turned on by default: iusewetsnow=1. (Setting iusewetsnow=3 will switch back to old formulation.)

TYPE: bug fix

KEYWORDS: w_damp, cfl print criterion, IEVA

SOURCE: internal

DESCRIPTION OF CHANGES:

IEVA = Implicit Explicit Vertical Advection

Problem:
Code in w_damp subroutine was changed in 4.3 to turn on w_damping at CFL = 1.2, and print CFL when it exceeded 1.2. Prior to v4.3, w_damping was turned on when CFL = 1.0, and printed the message CFL message when the threshold value of 2.0 was exceeded. This change results many more CFL prints. We want to revert to the original behavior.

Solution:
This PR fixes the CFL warning issues by changing the threshold for notification back to what it was before 4.3, and will turn on w_damping at CFL = 1. Importantly, this modification does not change the behavior that sets a higher CFL value to turn on w_damping when IEVA is used.

LIST OF MODIFIED FILES:
M dyn_em/module_big_step_utilities_em.F

TESTS CONDUCTED:

Do mods fix problem? Yes.
Jenkins tests are all PASS.

RELEASE NOTE: Code in w_damp subroutine was changed in 4.3 to turn on w_damping at CFL = 1.2, and print CFL when it exceeded 1.2. This change resulted in many more CFL prints. This PR reverts both changes to what it was before v4.3, where w_damping starts at CFL = 1.0, and CFL prints appear when CFL exceeds 2.0. When IEVA is on, w_damping starts at CFL = w_crit_cfl set in the namelist, and CFL prints only appear for those exceeding 2.0.

TYPE: enhancement

KEYWORDS: CMIP6, GHG, SSP, climate, radiation

SOURCE: Jesus Fernandez (Instituto de Fisica de Cantabria CSIC-UC, Santander, Spain)

DESCRIPTION OF CHANGES:
Problem:
Missing GHG concentration files for the new Shared Socioeconomic Pathways used in CMIP6 and IPCC AR6 
([Meinshausen et al. 2020](https://doi.org/10.5194/gmd-13-3571-2020)). They are required to nest WRF into 
CMIP6 global models for future regional climate simulation.

Solution:
Add corresponding GHG concentration files using global average data provided via input4MIP (namely, via CSV files 
provided by https://greenhousegases.science.unimelb.edu.au). Note that yearly hemispheric averages are also 
available from this source, and also 15-degree resolution monthly values. These could be used to tailor GHG 
evolution files for a particular WRF domain (see the code used to generate these files in 
https://github.com/SantanderMetGroup/wrf-ghg-files/tree/v0.1.2). The files included in this PR are global averages, 
consistent with previous CAMtr RCP files.

The files are used as the existing RCP CAMtr files, by renaming or linking the appropriate file with the exact name 
CAMtr_volume_mixing_ratio in the WRF running directory. For this to work, WRF needs to be compiled with the macro 
`-DCLWRFGHG` and the long-wave radiation parameterization set to one of: RRTM, CAM, RRTMG (ra_lw_physics = 1, 
3, 4).

A successful reading of the files leads to messages such as the following at the beginning of the WRF standard output log:
```
d01 2033-12-27_00:00:00  CLWRF reading...: istatus:  0 idata: 1 year: 1765 co2:    277.913000000000      n2o:    274.372000000000      ch4:   738.986000000000
d01 2033-12-27_00:00:00  CLWRF reading...: istatus:  0 idata: 2 year: 1766 co2:    277.958000000000      n2o:    274.392000000000      ch4:   739.831000000000
d01 2033-12-27_00:00:00  CLWRF reading...: istatus:  0 idata: 3 year: 1767 co2:    278.017000000000      n2o:    274.426000000000      ch4:   740.729000000000
d01 2033-12-27_00:00:00  CLWRF reading...: istatus:  0 idata: 4 year: 1768 co2:    278.078000000000      n2o:    274.447000000000      ch4:   741.222000000000
```

LIST OF MODIFIED FILES: 

M       run/README.physics_files
A       run/CAMtr_volume_mixing_ratio.SSP119
A       run/CAMtr_volume_mixing_ratio.SSP126
A       run/CAMtr_volume_mixing_ratio.SSP245
A       run/CAMtr_volume_mixing_ratio.SSP370
A       run/CAMtr_volume_mixing_ratio.SSP585

TESTS CONDUCTED: 
1. Model runs with the new files, which are just ASCII files formatted exactly as the existing ones for RCPs used in 
CMIP5.
2. The new files were compared to the existing ones for RCPs to ensure a consistent order of magnitude and 
evolution of the concentration values. No long-term climate simulation has been yet conducted with the new data 
files, but the files are within the order of magnitude of existing RCP files, which were extensively tested. 
![CAMtr file comparison](https://raw.githubusercontent.com/SantanderMetGroup/wrf-ghg-files/v0.1.2/CAMtr_files.png)
  
RELEASE NOTE: Added new CAMtr_volume_mixing_ratio data files for CMIP6 SSPs (https://doi.org/10.5281/zenodo.5507554) containing yearly global average concentrations from input4MIPs (v1.2.0 for historical part and v1.2.1 for scenario evolution).

TYPE: bug fix
    
KEYWORDS: moving nest, vortex tracking, segfault, Intel
    
SOURCE: internal
    
DESCRIPTION OF CHANGES:
Problem:
When the Intel compiler is used with the moving nest option (vortex tracking), the tools/registry program seg-faults,
and does not complete building the automatically generated files that are included via cpp into the WRF source code.
    
Solution:
This is a quick-fix for users. Simply swap the order of the list of registry files.
    
ISSUE: For use when this PR closes an issue.
Fixes #1521
    
LIST OF MODIFIED FILES:
M   Registry/registry.em_shared_collection
    
TESTS CONDUCTED:
1. This mod allows the code to build successfully for intel 19.0.5
2. The code compiles with option 15/3 using Intel OneAPI:  `ifort (IFORT) 2021.1 Beta 20201112`
3. The code compiles with option 15/3 and run using Intel HPC suit: `ifort (IFORT) 19.0.5.281 20190815`. The code 
runs at least 1-h with a triple-nests 501 x 601 x 841 Tropical Cyclone with 55 levels.
4. Jenkins is all pass.

RELEASE NOTES: A temporary fix has been put in place for the moving nest capability when using the Intel compiler. Previously, the WRF source code did not build due to a segmentation fault in the execution of the tools/registry program.