MIPAS Quality Working
Group Meeting #12: Level 2
No NRT processing for now Ð new L2 products for validation dataset processed with GRIMI-2. Have a new format Ð no tools available yet. New BEAT available hopefully in next 1-2 weeks.
When D band is corrupted, retrieval skipped (re: 21-22 Aug 2004) as whole sweep is dubbed corrupted. Discussion on why this happens and what would be better É what if MWs used not corrupted? Clouds Ð start retrieval above corrupted sweep? Final word (Thorsten Fehr) is: Continue with baseline as is, think of updating baseline but move forward Émake software change request to Sven É
Based on baseline for full resolution Ð going to make updates to account for new configuration. Monthly monitoring to be resumed when data becomes available. Summary page to continue when data available.
EADS - Astrium Sven Bartha
Several issues raised (SCR40, SPR41, SPR42) have been cured by a patch. Two typos in documentation (averaging kernel -> double precision, spare field length in PS2 ->2 bytes) and new available is version 5.0. Contact at EADS Astrium is Michael Schmidt. Michael.Schmidt@astrium.eads.net. Tentative schedule for open work listed.
Is increase in NESR of 25% ok for L2, from decontamination perspective? Trade off Ð decontamination = loss of data, but better data quality. Lots of maths on information content, time frames etc. , but basically exponential. Optimal time to have decontamination found for different operation modes. Information cost-effective decontamination scheme in 3-d on, 4 off mode when decontamination during off period . Generally 66 days good, but delay in decontamination of up to 100% ok. TFÕs final word: decontaminate 4 times per year (May, August, October, January). Anu suggests doing the decontaminations around the MA modes É
IFAC - Simone Ceccherini
Used new MWs on orbit 17540 (reduced resolution). See small increase in c2 for pT, but smaller for all others. Anomalous peak at 21km has disappeared. Compared two retrievals: R1 done on full resolution data with old MWs and R2 done on reduced resolution with new MWs. R2 has almost universally higher c2. For full resolution, night has smaller c2 than does day, for all species except NO2. Possible NLTE effects in some MWs used for reduced resolution. Anu defends his MWs, since using more MWs as is done in reduced resolution will inherently increase c2.
IFAC - Piera Raspollini
Band-dependent parameter a used to modify the width of ILS
provided by L1B. Discrepancies between IFAC and Ox considered: 2nd
derivative of sinc function looked at Ð discrepancy because theyÕre actually
looking at different things É apodised vs unapodised!?
ESA wants to investigate possibility of using same MIP_PS2_AX for all floating altitudes. Floating altitudes basically follow along the tropopause so that you get at least one spectrum within troposphere which is not cloud contaminated. Engineered and actual tangent heights differ Ð why??
Friday 16
February 2007
ADF 2 v6.0
Comparison of nominal and UTLS-1 modes, in terms of scan patterns, mean c2 of species etc. Comparison of species (temp, CH4, O3 etc). Problem in that in ORM retrieval many examples donÕt actually converge Ð c2 increases Ð to be discussed in Marco RidolfiÕs presentation. (IÕm too tired to be creative and catty. Maybe after coffee break.)
Bologna Ð Marco Ridolfi
Need to optimize LM-parameters for pT. Using microwindow series 300, convergence criterion Ver.3.4, and a-posteriori Tikhonov regularisation with EC method. Retrieval is said to fail to converge if it hasnÕt done so within 10 iterations. End up with 7 triplets of optimized LM parameters. New problem: sometimes pT retrieval doesnÕt converge because there are too many microiterations. Due to intrinsic weakness of algorithm that updates simultaneously tangent pressures and temp profile Ð but can be fixed in code. Presents final performance achieved by using new optimized LM-parameters.
Bologna Ð Gabrielle
870 Ð 871.5 cm-1 spectral interval used to retrieve heavy nitrogen. Retrieved scan-by-scan, assuming homogeneous atmosphere. Use 3 MWs, including new 870-871.5 cm-1 MW. High altitudes, retrieval stays close to initial guess. Lower altitudes, retrieval moves towards real value. Open issues: error budget, vertical resolution, required number of scans, noise error vs strength of regularization É
Bologna - Massimo Carlotti
CO2 used to derive pT in stratosphere usually. CMTR updated to do CO2 line-mixing and a target-dependent retrieval grid. MWs used are in bands A and AB. O3 responsible for most of error in VMR. Do GMTR of p, T, H2), O3. Then GMTR for p, T, CO2 É Monthly averages over 10 degree latitude bins. Systematic error still remains significant, between 3-8%.
Leicester Ð John Remedios
New cloud MWs for reduced resolution - MWs are more or less the same, but might want to change thresholds used. Also looked at error to go with CI Ð used NESR. Looked at CI and associated error as a function of temperature and altitude. At thresh about 2, error is small at about 0.6. As threshold shifted upwards, error gets bigger (which makes sense since will include more clear data as well). As temperature is lowered, errors go up. Maybe a latitude dependent cloud index threshold required? V4.61 doesnÕt completely flag cloud Ð but V4.62 does. Needs to check upper altitude limits (present A band uses 6-45 km). Tech note available soon, will be sent around to everyone.
Shows old and new pT microwindows Ð extra MW near 790 cm-1 makes the difference and smoothes over the 21 km problem.
Now contains complete orbit in L1B file, instead of just 1 san per file. Overlap in latitude in AE and NOM mode measurements could allow for comparisons, if assuming consistent atmospheric conditions. AE mode seems to offset by 4-5 km from actual.
Using only nominal measurements from AE-flagged measurements. Temp, H20 stripy near south pole. Temp very different from initial guess. Enhancement of HNO3 near North Pole (due to solar proton event). Double peak in ozone plot over equator Ð not sure of its cause. Methane shows equatorial oscillation, as does N2O.
Temperature higher at poles when use NOM MWs instead of MA MWs. H2O vapour, again high over south pole when using NOM Mws instead of MA MWs. Stripes at equator and south pole in H2O. Forward and reverse pattern when using MA MWs. HNO3 and NO2 have higher values when using NOM MWs. O3 doesnÕt show double peak anymore in these set of days.
IMK Ð Ann
Kleinert
Temperature fields compared with ECMWF and IMK-UTLS retrieval. Compared with MLS temperature as well. Retrieved H2O compared with ECMWF and MLS Ð better agreement with MLS data. Comparison with MIPAS-B temperature (balloon profile and closest MIPAS satellite measurement) : temperature, H2O, O3, HNO3 within error, although some species show biases. Retrieval of polar stratospheric clouds, CI threshold of 4.5 used.