Hi,
Patrick will look into exactely what happens with Y.HSE, however if you want to
ensure consistent results you can start by setting Q.HSE.ON = false in both
your cases. This ensures that hydrostatic equilibrium adjustments do not change
anything.
Then you can move on to compare the raw files directely as you described.
t_field.xml, z_field.xml and vmr_field.xml should be indendical in both cases.
To load these files use "xmlLoad" (xmlLoad('t_field.xml')). opening in the
editor will not work out of the box. Smilar for the other files.
Finally if you still don't find any differences you can directely compare the
two generated arts controlfiles "cfile.arts" and see if you see any
difference.
Ole Martin
________________________________________
From: [email protected] [[email protected]] on behalf of
Pauline Martinet [[email protected]]
Sent: Monday, September 07, 2015 9:50 AM
To: Patrick Eriksson
Cc: [email protected]
Subject: Re: [Qpack] differences between Qpack and arts simulation
Dear Patrick and Ole Martin,
Thank you very much for helping me tracking down the error.
I have done the following checks:
- F grid identical in Q and Y ok
- T field identical to the first dimension of Q.T.ATMDATA.DATA
- VMR field identical to the ABS_SPECIES.ATMDATA.DATA : ok
- Y.HSE_P equals to Q.HSE_P
- Y.ZA is equal to Q.SENSOR_LOS (for the zenith angle)
- Q.SENSOR_POS equals Y.Z_PLATFORM
I have noticed that have a value in Y.HSE_Z (=60 in my example) and I do not
see this HSE_Z argument in my Q structure but
Q.SENSOR_POS is equal to Y.HSE_Z. Did I miss something in the Q structure ?
To be sure one more time that I have exactly the same values, I ran atmlab in
debug mode. A directory is created with the following files but I don't know
which one I should compare to:
backend_channel_response.xml f_grid.xml mblock_aa_grid.xml
sensor_los.xml vmr_field.xml z_surface.xml
backend_channel_response.xml.bin f_grid.xml.bin mblock_aa_grid.xml.bin
sensor_los.xml.bin vmr_field.xml.bin z_surface.xml.bin
cfile.arts lat_true.xml mblock_za_grid.xml
sensor_pos.xml y_aux.xml
cfile.propmat_clearsky_field.xml lat_true.xml.bin mblock_za_grid.xml.bin
sensor_pos.xml.bin y.xml
f_backend.xml lon_true.xml p_grid.xml
t_field.xml z_field.xml
f_backend.xml.bin lon_true.xml.bin p_grid.xml.bin
t_field.xml.bin z_field.xml.bin
Moreover, there is data when I open vmr_field.xml or t_field.xml.
Should I download the: atmlab-trunck.tar.gz to have the last version of atmlab ?
To use it instead of the last version in my code, do I have to do something
else except rerunning atmlab_init in the atmlab repertory ?
Thank you again for your help.
Best regards,
Pauline
----- Météo-France -----
Dr. Pauline Martinet
Chercheur CNRM/GMEI/LISA
[email protected]
Fixe : +33 561079031
Site web: www.sites.google.com/site/martinetpauline31
----- Mail original -----
De: "Patrick Eriksson" <[email protected]>
À: "Pauline Martinet" <[email protected]>, [email protected]
Envoyé: Vendredi 4 Septembre 2015 20:05:33
Objet: Re: [Qpack] differences between Qpack and arts simulation
Hi Pauline and all,
Nice that you are so careful!
First I should clarify one thing. Y.F is in fact not used by qpack
itself. But it should be good practice to set Y.F and run qp2_check_f as
part every retrieval.
I had a similar problem some days ago. I tracked that down, and found
that it was due to that f_grid could was stored as float in
arts_oem_init. This gave a difference of 3 kHz, and this caused a
difference of 5 K for limb sounding of the 118 GHz O2 line! I fixed this
problem in v2.3 and this made me remember to do it for v2.2 as well.
But I don't think this your problem (but maybe update atmlab anyhow).
But there are some possible pitfalls here. As Ole Martin wrote, qpack
takes data from Y. Note that some HSE settings are taken from Y. Do you
have exactly the same values in the Q sent to arts_y?
Also note that qpack uses the AT
MDATA fields, but not arts_y. For
example, if T.ATMDATA is set, qpack overwrites Q.T_FIELD.
I hope these things are mentioned in the qpack pdf, but things can
always be explained clearer ...
Bye,
Patrick
On 09/04/15 17:45, Pauline Martinet wrote:
> Hi everybody,
>
> I have noticed differences between the forward model calculations when I
> call Qpack or the function arts_y.
> I am simulating a microwave radiometer between 22 and 60 GHz.
> I have a matlab routine where I define the structures Q and O.
> Then I call the two functions:
> TB=arts_y(Q);
> and L2=qpack2(Q,O,Y);
> with Y.Y=[] : from what I read in the Qpack documentation it should give
> the same results.
>
> But then I do not obtain the same values of brightness temperatures when
> simulating the brightness temperature with Qpack or arts_y:
>
> This is from Qpack:
>
> 37.6242
> 36.7163
> 32.3708
> 24.4462
> 22.0529
> 19.4580
> 18.6558
> 110.6579
> 152.7556
> 250.9260
> 279.8277
> 284.9573
> 285.2151
> 285.3064
>
> This is from arts_y:
> 37.5430
> 36.6337
> 32.2927
> 24.3829
> 21.9958
> 19.4087
> 18.6113
> 110.5074
> 152.5850
> 250.8197
> 279.7959
> 284.9563
> 285.2181
> 285.3116
>
> For the call to arts-y I had to define: Q.T_FIELD, Q.Z_FIELD, Q.VMR_FIELD
>
> but I normally use the same values as the ones defined in Q.ABS_SPECIES
> and Q.T for Qpack.
>
> Do you have any idea on what I am doing wrong ?
>
> Thank you very much.
>
> Best regards,
>
> Pauline
>
> ----- Météo-France -----
> Dr. Pauline Martinet
> Chercheur CNRM/GMEI/LISA
> [email protected]
> Fixe : +33 561079031
> Site web: www.sites.google.com/site/martinetpauline31
> <https://www.sites.google.com/site/martinetpauline31/>
>
>
>
> _______________________________________________
> qpack mailing list
> [email protected]
> https://www.sat.ltu.se/mailman/listinfo/qpack
>
_______________________________________________
qpack mailing list
[email protected]
https://www.sat.ltu.se/mailman/listinfo/qpack
_______________________________________________
qpack mailing list
[email protected]
https://www.sat.ltu.se/mailman/listinfo/qpack
