Dear Patrick, Thank you so much. I do not know what is wrong. I also try without water vapor (just temperature and O3) and the problem is the same. If it's not too much trouble, could you try to do the same as before but adding the perturbation at 13km, where the column of the AVK (Figure 6) is different to zero?
Thank you very much Facundo 2016-01-18 18:59 GMT+09:00 Patrick Eriksson <[email protected]>: > Dear Facundo, > > No, the change at 7 km should not have any impact. I did a small test > myself. I used qpack2_demo and tested this > > --- > > % Add a second measurement > % > Y(2) = Y(1); > % > Y(2).LONGITUDE = pi; > Y(2).ZA = 45; > > % Add 0.5 ppm around 7 km: > Q.ABS_SPECIES(1).ATMDATA.DATA(10) = ... > Q.ABS_SPECIES(1).ATMDATA.DATA(10) + 0.5e-6; > > % Calculate simulated spectra > % > Y = qpack2( Q, oem, Y ); % Dummy oem structure OK here > > return > > --- > > Note thr return, to avoid adding noise to the spectra. > > > > That gave (just including result of first inversion) > > >> qpack2_demo; > Simulating spectrum 1/2 > Simulating spectrum 2/2 > > /--------------------------------------------------------------------\ > | Inversion case 1 (of 2) | > | Gamma Total Profile Spectrum Converg. | > | Iteration factor cost cost cost measure | > | 1 NaN 0.000 0.00 0.00 NaN | > | 2 0.0 0.000 0.00 0.00 0.00 | > \--------------------------------------------------------------------/ > > > That is, the disturbance had no impact. If I instead added 0.5 ppm at > altitude 20 (i.e. disturbing DATA(20)), I got: > > >> qpack2_demo; > Simulating spectrum 1/2 > Simulating spectrum 2/2 > > /--------------------------------------------------------------------\ > | Inversion case 1 (of 2) | > | Gamma Total Profile Spectrum Converg. | > | Iteration factor cost cost cost measure | > | 1 NaN 0.118 0.00 0.12 NaN | > | 2 0.0 0.000 0.00 0.00 2.85 | > | 3 0.0 0.000 0.00 0.00 0.00 | > \--------------------------------------------------------------------/ > > > That is, the "measurement" is disturbed from the a priori state. > > I tested this both for v2.2 and v2.3, to be totally sure. > > > So, you must do something else in your test that gives a disturbance of > the spectrum. But I can not figure out what it can be. In any case, all > looked OK in my test. > > Regards, > > Patrick > > > > > > > > > > > > > On 01/15/16 09:16, Facundo Orte wrote: > >> Dear Patrick, >> Thanks a lot for your reply. >> Related with the first suggestion, I interpolate the a-priori ozone >> profile and synthetic ozone profile to Q.P_GRID to be sure that dO3 is >> affecting only the altitud that I want to affect. >> Temperature and H2O are same in both cases and also all other parameters. >> I prepared a .pdf that I am sending attached to explain better which is >> the problem. >> I can not understand why ozone below Y.Z_PLATFORM is affecting my >> retrieval. >> >> Thank you so much. >> Regards >> Facundo >> >> >> 2016-01-14 16:46 GMT+09:00 Patrick Eriksson >> <[email protected] <mailto:[email protected]>>: >> >> >> Dear Facundo, >> >> What do you mean with "a little bit different"? A figure would have >> helped. >> >> As soon as you involve a retrieval in test of this kind, you need >> also to consider the retrieval grid. In this case, the O3 profile >> will be interpolated, from the retrieval grid to Q.P_GRID and only >> this can introduce some disturbance. To keep this "disturbance >> small", it is a good idea that the retrieval grid is sub-set of the >> P_GRID, such every second point in P_GRID. Or if you do a test like >> this, why not set the retrieval grid to P_GRID. >> >> Just to be clear, no change in the profile is totally local, it has >> an effect out to the adjecent grid points. But I assume that you >> have some grids points between 7 and 12 km, and this is not an issue >> here. >> >> Anyhow, a change of O3 at 7 km should not affect upward measurements >> at 12 km. Note that things can be different for temperature and H2O >> as these quantities affect hydrostatic equilibrium, and a local >> effect propagates through the atmosphere. >> >> Regards, >> >> Patrick >> >> >> On 01/14/16 07:14, Facundo Orte wrote: >> >> Dear Patrick and Ole Martin, >> I am writing again from this old mail because I have a problem >> in Qpack2 >> that I think that it is related with the setting that Ole Martin >> proposed in the previous mail, but I am not sure. >> To explain it I want to share the experiment that I did: >> >> 1. I set Y.Z_PLATFORM=12Km. My O3 a-priori profile goes from >> ground to >> 90km (I will call it 'APRIORI') [So, then of the 3. step I will >> have a >> simulated spectrum from the a priori (SPECTRUM_AP)] >> >> 2. I make a synthetic profile (taking it as O3 in the atmosphere >> or true >> state vector)which is same as the a priori profile, but I add a >> delta O3 >> (0.5ppm) at 7km (PROFILE_7km). So, I arrange a syntetic "true >> state" >> that I know. Then I calculate a spectrum (SPECTRUM_7km) from >> PROFILE_7km >> using QPack2/ARTS. I took this spectrum as the measurement. >> >> 3. I run Qpack2/ARTS inserting SPECTRUM_7km as measurement, and >> APRIORI >> as a priori profile and stop in the first iteration. >> >> (Note that I am setting Y.Z_PLATFORM=12km, and I the add a delta >> O3 the >> ozone in the true state vector at 7km. ) >> >> The result was that SPECTRUM_AP (yf) is a little bit diferent to >> SPECTRUM_7km (y). I think that it must be equal (SPECTRUM_AP = >> SPECTRUM_7km) because the Y.Z_PLATFORM is above to 7km and a >> perturbation at 7km should not affect the retrieval. Is it >> correct? >> >> the Q is: Why changes below Y.Z_PLATFORM in the true state vector >> affect the retrieval? >> >> Thanks in advance. >> Best regards >> Facundo >> >> >> >> >> >> >> >> 2014-11-18 21:58 GMT+09:00 Ole Martin Christensen >> <[email protected] >> <mailto:[email protected]> >> <mailto:[email protected] >> >> <mailto:[email protected]>>>: >> >> Hi, >> >> the standard way of doing this is to do it before you do the >> retrieval. You manually adjust your measured spectra >> assuming that >> your measured opacity is correct (called "tropospheric >> correction"). >> >> In arts you can then model your instrument as being above the >> troposphere, i.e. by setting the altitude of your sensor >> (Y.Z_PLATFORM if you use qpack2) to 15 km. The simulated >> spectra can >> then be compared to your corrected spectra for a successful >> retrieval. >> >> Another option is using continuum models to model O2, >> H2O(gas and/or >> liquid) across the tropopause. These can be added by adding >> by >> writing e.g.: >> >> Q.ABS_SPECIES(2).TAG = { 'H2O-PWR98' }; >> >> With a following fields like Q.ABS_SPECIES(2).ATMDATA, and >> Q.ABS_SPECIES(2).GRIDS. >> >> However, in general the manual tropospheric correction is >> the most >> commonly used method, as it is simpler, and would recommend >> starting >> with this. >> >> If anything is unclear please feel free to send more >> questions. For >> further inquiries I would also strongly recommend the qpack >> and arts >> mailinglists. >> >> https://www.sat.ltu.se/mailman/listinfo/qpack >> https://www.sat.ltu.se/mailman/listinfo/arts-users >> >> Regards >> >> Ole Martin >> >> >> >> >> ------------------------------------------------------------------------ >> *From:* Facundo Orte [[email protected] >> <mailto:[email protected]> >> <mailto:[email protected] >> >> <mailto:[email protected]>>] >> *Sent:* Tuesday, November 18, 2014 12:44 PM >> *To:* Patrick Eriksson >> *Cc:* Ole Martin Christensen >> *Subject:* Re: Question about ARTS >> >> Dear Patrick Eriksson, >> thanks for your reply. >> The MWR is installed in Río Gallegos city (Latitud: -51° 36' >> 02,03",Longitud: -69° 19' 09.96") since 2011. The name of the >> observatory is "Observatorio Atmosferico de la Patagonia >> Austral" >> which depends to Lidar Division (CEILAP). You can see the >> site in >> this link http://www.division-lidar.com.ar/sitios_rg.html. >> It is >> quite near to Ushuaia. >> The MWR is working at 110.83GHz. For calculate the opacity >> from the >> MWR we can use, for example, the method described by Zafra >> et al. >> (Zafra, R. L., Parrish, A., Solomon, P. M and Barrett, J. >> W., A >> Quasi-continuous record of atmospheric opacity at λ=1.1 mm >> over 34 >> days at Mauna Kea Observatory, International Journal of >> infrared and >> Millimeter Waves, Vol. 4, No 5, 1983.). Comparisons between >> the >> opacity measured and obtained from radiosonde profiles >> presents good >> agreement. >> If I may to introduce the measured opacity (by MWR) as >> input of ARTS >> maybe it could be good. My question is:where can I input the >> measured opacity in ARTS? >> Thanks a lot >> Best Regards >> Facundo >> >> >> 2014-11-17 15:53 GMT-03:00 Patrick Eriksson >> <[email protected] >> <mailto:[email protected]> >> <mailto:[email protected] >> >> <mailto:[email protected]>>>: >> >> Dear Facundo Orte, >> >> Nice to hear that you have started to use Qpack. >> >> Getting curious, where exactly is the radiometer >> placed? Did >> some travelling around Latin America in the 90-ies. >> Passed e.g. >> Uschaia. What frequency is the radiometer using? >> >> Don't understand your question exactly. How do you >> observe an >> opacity with the radiometer? I don't know how that could >> be >> done. Measured brightness temperatures can be converted >> to an >> opacity, but then it is better to use the original >> Tb-data. Or >> is the opacity determined by some other instrument. If >> you >> describe this closer, we will try to help you. >> >> Regards, >> >> Patrick >> >> >> >> On 2014-11-17 19:05, Facundo Orte wrote: >> >> Dear Patrick Eriksson, >> My name is Facundo Orte and I am a PhD student of >> the Lidar >> Division of >> CEILAP (Laser Research Center) from Argentina >> (division-lidar.com.ar >> <http://division-lidar.com.ar> <http://division-lidar.com.ar> >> <http://division-lidar.com.ar>__). >> >> I am writing you because in our laboratory there is a >> Millimiter Wave >> Radiometer (MWR) instaled in South Patagonia, near >> to the >> spring ozone >> hole, and we retrieve ozone profiles with it. To >> analyze the >> signal we >> are trying to use the ARTS model that you and your >> team >> developed. I am >> using MATLAB to introduce inputs changing >> Q.DEFINITION.m >> function and I >> retrieve some profiles but there are quite close to >> "a >> priori" ozone >> profile (I am using MLS climatology as a priori O3 >> profile). >> I think >> that it is because I do not have well caracterized >> the >> atmosphere and I >> am trying to do that. For example, at this moment, >> I do not >> know how to >> introduce the opacity measured for the MWR >> instrument from >> MATLAB as >> input and I think that it is an important parameter >> to >> introduce. >> Could you help me with it? Do you know how I can to >> introduce the >> opacity from MATLAB? >> I will very appreciate if you can help me with it. >> Thank you in advance >> Best regards >> Facundo Orte >> >> >> >> >> >> >>
_______________________________________________ qpack mailing list [email protected] https://www.sat.ltu.se/mailman/listinfo/qpack
