TIME–SERIES COMPARISONS OF MIPAS LEVEL 2 NEAR–REAL–TIME PRODUCTSWITH CLIMATOLOGY

Vivienne Payne, Anu Dudhia, and Chiara Piccolo

Atmospheric, Oceanic and Planetary Physics, Department of Physics, University of Oxford, Oxford, UK, e–mail:payne@atm.ox.ac.uk

ABSTRACT

Monthly mean profiles have been calculated for theMIPAS Near-Real-Time Level 2 products (temperature,H � O, O � , HNO � , CH � , N � O and NO � ) from July 2002until March 2004. The Level 2 profiles have been splitinto six latitude bands (90S–65S, 65S–20S, 20S–0, 0–20N, 20N–65N, 65N–90N) for the calculation of themeans. Here we compare these monthly means with ref-erence climatologies for each of the Level 2 products toprovide an overview of the quality of these products overthe timescale that MIPAS has been operating. The ref-erence climatologies used in these comparisons are theCOSPAR International Reference Atmosphere (CIRA)(2) for temperature and the IG2 dataset (1) (used in theconstruction of the initial guess for MIPAS retrievals) forthe gases.

Key words: MIPAS; Level 2.

1. INTRODUCTION

Monitoring of the Level 2 products has been performedroutinely at the University of Oxford. Monthly mean pro-files have been calculated for the MIPAS Near-Real-TimeLevel 2 products (temperature, H � O, O � , HNO � , CH � ,N � O and NO � ) from July 2002 until March 2004. TheLevel 2 profiles have been split into six latitude bands(90S–65S, 65S–20S, 20S–0, 0–20N, 20N–65N, 65N–90N) for the calculation of the means, and the monthlymean profiles for each of the products plotted as time–series. Some spike detection has been used in the aver-aging process, mostly excluding values of N2O greaterthan 2ppmv, CH4 greater than 10ppmv and H2O greaterthan 1000ppmv. This examination of the time–series ofthe monthly means of the MIPAS Level 2 products is in-tended to give an overview of the quality of the data andof any persistent features in the profiles. The intentionwas also to try to identify any step changes in the datawith time that might have resulted from changes in thedata processing chain. Changes in processing for whichwe thought we might see changes in the monthly meanprofiles were a Level 1 fringe count error correction inMarch 2003, the implementation of cloud detection andthe updating of HNO � look–up–tables in July 2003, and anumber of improvements to the Level 1B processing thatwere implemented in November 2003.

In order to check whether the monthly mean profilesagree with what might be expected, we also comparethem with reference climatologies. The reference cli-

matologies used in these comparisons are the COSPARInternational Reference Atmosphere (CIRA) (2) for tem-perature and the IG2 dataset (1) (used in the construc-tion of the initial guess for MIPAS retrievals) for thegases. Further details of the monitoring that has been tak-ing place at Oxford can be found on the Oxford MIPASGroup web pages (3).

2. TEMPERATURE

Fig. 1 shows a contour plot of the MIPAS monthly meantemperature profiles from July 2002 until March 2004,for the six different latitude bands. Evidence of an-nual cycles can clearly be seen in the polar latitudebands of both hemispheres, with particularly cold tem-peratures in the southern hemisphere polar winter of 2003(June/July/August).

Fig. 2 shows the absolute difference of the MIPASmonthly mean temperature profiles from the COSPARInternational Reference Atmosphere (CIRA). The reasonthat the CIRA climatology, rather than the IG2, was usedfor this temperature comparison was that the IG2 onlyhas “polar winter” and “polar summer” profiles in six–month blocks, resulting in temperature differences fromthe MIPAS monthly mean temperature profiles of up to70K in polar regions. It was thought that the CIRA datawould show seasonal variations closer to those in the realatmosphere. Differences of as much as 30K are seen atpolar latitudes in the MIPAS-CIRA differences shown inFig. 2. However, these large differences are due to thelack of a strong seasonal cycle in the CIRA profiles, andare not a cause for concern. In mid–latitudes and equato-rial regions, it appears that the MIPAS temperatures aregenerally colder than the reference climatology through-out most of the stratosphere.

3. H � O

Fig. 3 shows a contour plot of the MIPAS monthly meanH � O profiles from July 2002 until March 2004. In gen-eral, the profiles show a minimum around the tropopause,with the volume mixing ratio (VMR) gradually increas-ing with height, as would be expected due to the oxida-tion of CH � . Also, drier “tongues” of air can be seenpropogating upwards from the tropical tropopause in theequatorial regions. This would appear to be consistentwith the “tropical tape recorder effect” (4). However,there are some obvious problems with the MIPAS watervapour profiles. The profiles tend to show a reasonablysharp decrease at the highest retrieval altitude, which is

_______________________________________________________Proceedings of the Second Workshop on the Atmospheric Chemistry Validation of ENVISAT (ACVE-2)ESA-ESRIN, Frascati, Italy, 3-7 May 2004 (ESA SP-562, August 2004) EPOMIVP

not a feature expected in the real atmosphere. Also, val-ues at the second–highest retrieval altitude seem ratherlarger than would be expected. There are also some unre-alistically high values of H � O VMR in the southern polarmonthly means for the 2003 southern hemisphere winter.It is thought that this may have been due to polar strato-spheric clouds, which would be consistent with the lowtemperatures observed for this time period, which couldhave caused difficulties in the MIPAS retrievals.

Fig. 4 shows the percentage difference between the MI-PAS monthly mean H � O and the IG2 climatology pro-files. The sharp decrease in MIPAS H � O at the highestretrieval altitude, the otherwise large MIPAS H � O val-ues at the second–highest altitude and the problems inthe southern polar regions in winter can be clearly seenin these difference plots. In addition, it appears that theMIPAS H � O at the lowest retrieval altitude is generallymuch higher than would be expected from the climatol-ogy. It is thought that this may be due to a problemwith the implementation of the cloud detection algorithmin the MIPAS retrieval. Cloud detection was supposedto have been implemented in July 2003, and we mighthave expected to see a step change in the monthly meansaround this time. However, it seems that there have beensome problems in the implementation of the cloud de-tection in the Near–Real–Time processing. This is cur-rently under investigation. It should, however, be notedthat the cloud detection is working in the Off–Line pro-cessing (the results of which are not dealt with in this pa-per), but that there are difficulties there in distinguishingcloudy sweeps from those with very high water vapouramounts.

MIPAS H � O is, in general, much higher than the IG2climatology at the lowest retrieval altitude (12 km, oraround 100 mbar), around 20% lower than the IG2 clima-tology in the lower stratopshere (below 10 mbar), higherthan the IG2 by up to 40% between about 4 mbar and1 mbar, and around 30–50% lower than the IG2 at thehighest retrieval altitude.

4. O �

Fig. 5 shows a plot of the time-series of the MIPASmonthly mean O � profiles from July 2002 to March 2004.The plots show a maximum in the ozone VMR at around10 mbar, with the highest O � values in the tropics and ev-idence of ozone depletion in the autumn seasons of bothhemispheres, with the strongest depletion in southern po-lar latitudes, as would be expected.

Fig. 6 shows the percentage difference between the MI-PAS monthly mean O � and the IG2 climatology profiles.Large differences from climatology can be seen in the po-lar profiles, but the reason for this is that the climatologydoes not contain such strong seasonal cycles as the realatmosphere, so this is not a cause for concern. In themid–latitude and equatorial regions, MIPAS O � is muchhigher at the lowest retrieval altitude than would be ex-pected from the climatology. As in the case of H � O ( see

section 3.), it is suspected that this might be due to prob-lems with cloud in the MIPAS retrieval. It can also beseen from the plots in Fig. 6 that the MIPAS O � is gener-ally lower than climatology in the lower stratosphere andhigher than climatology in the upper stratosphere. This isbecause the ozone peak in the MIPAS profiles occurs at ahigher altitude than in the climatology.

5. HNO �

Fig. 7 shows a plot of the time-series of the MIPASmonthly mean HNO � profiles from July 2002 to March2004. Strong sesonal cycles can be seen in the polar re-gions and mid–latitudes, with the highest values of HNO �occurring in polar winter.

Fig. 8 shows the percentage difference between the MI-PAS monthly mean HNO � and the IG2 climatology pro-files. Large differences from climatology are seen in thepolar regions, where the climatology does not have sucha strong seasonal cycle as the real atmosphere. The prob-lem with high MIPAS values at the lowest retrieval alti-tudes can be seen here as well as in the other gases. Amarked change can be seen in the northern hemisphereequatorial difference plot around 10 mbar between Juneand July 2003. Before July 2003, MIPAS HNO � is lessthan climatology, and after, MIPAS HNO � is greater thanclimatology. The look–up–tables (effectively the spec-troscopy data) were updated for HNO � in July 2003, andit is thought that perhaps this is the cause for the suddenchange in MIPAS HNO � concentrations. However, thischange is not seen in the southern hemisphere equatorialmonthly means. The reason for this is not clear.

6. CH �

Fig. 9 shows a plot of the time-series of the MIPASmonthly mean CH � profiles from July 2002 to March2004. A strong seasonal cycle can be seen in polar re-gions, due to the general descending motion of the air atcertain times of year. CH � is produced at the surface, iswell-mixed in the troposphere, and broken down in thestratosphere by oxidation and photolysis. It is thereforeexpected that the CH � profiles should decrease monoton-ically with height. This is not the case for some of theequatorial monthly mean profiles. The MIPAS CH � in theequatorial latitude bands for some of the months shows alocal minimum at around 15–18 km. The reason for thisis not known, but it is thought that it may be due to the3 km vertical resolution of MIPAS being unable to re-solve temperature features around the tropopause region.

Fig. 10 shows the percentage difference between the MI-PAS monthly mean CH � and the IG2 climatology pro-files. As with the other gases, the MIPAS CH � at thelowest retrieval altitude seems unrealistically high whencompared with climatological profiles. The problem withthe local minimum in the equatorial tropopause regioncan be seen as blue regions in the equatorial plots. Ingeneral, MIPAS CH � is around 20% higher than would

be expected over most of the stratosphere, for all latitudebands.

7. N � O

Fig. 11 shows a plot of the time-series of the MIPASmonthly mean N � O profiles from July 2002 to March2004. Like CH � , N � O is produced at the surface, is well–mixed in the troposphere, and falls off with height in thestratosphere. The shape of the N � O plots would thereforebe expected to be qualitatively similar to the CH � plots,and indeed this is the case. The problem of the local min-imum in the tropical tropopause region can also be seenin the MIPAS N � O profiles.

Fig. 12 shows the percentage difference between the MI-PAS monthly mean N � O and the IG2 climatology pro-files. There are particularly large differences from cli-matology in polar regions, but again, as with the othergases, this is likely to be attributable to the lack of strongseasonal variations in the climatology. The differenceplots for N � O show some sort of banded structure: ingeneral, MIPAS is higher than climatology at the lowestaltitudes, lower than climatology in the region between80 and 20 mbar, higher than climatology in the regionbetween around 10 and 2 mbar and lower than climatol-ogy at the highest retrieval altitudes. Since we are look-ing at monthly means, this banded difference structure isunlikely to be due to the sort of retrieval–induced oscil-lations that are seen in individual N � O profiles, as thesetend to average out. It would appear that there is somekind of structure in the MIPAS profiles that is not seen inthe climatology.

8. NO �

Since NO � has a diurnally–varying concentration, theprofiles have been further split into day and night pro-files, separated based on the value of the solar zenith an-gle. Fig. 13 and Fig. 14 show plots of the time-seriesof the MIPAS day and night monthly mean NO � profilesfrom July 2002 to March 2004. The difference betweenday and night profiles can be clearly seen. For monthswhere there is complete polar day or night, there are maybe no NO � profiles in the polar latitude bands.

The IG2 climatology for NO � does not take diurnalvariation into account. Therefore, comparisons withthe IG2 dataset show very large differences. In an at-tempt to obtain a more useful comparison, the MIPASmonthly means for day and night were compared withmid–latitude day– and night–time climatologies. (Thesemid-latitude day and night profiles were also producedby Remedios (1) with MIPAS in mind, but, unlike theIG2 profiles, are not used in the construction of the ini-tial guess for the MIPAS retrieval. Also, they show noseasonal variations at all). It can be seen from theseplots that the MIPAS mid–latitude day–time NO � is verymuch lower (by 50–100%) than we might expect fromthis climatology. The night time profiles are also gener-ally around 20–50% lower than we would expect from

the climatology. It is difficult to say anything more aboutthese plots, since the climatology here does not changewith time. In addition, the real atmospheric NO � willshow variation with time of day that cannot be adequatelyaccounted for simply by splitting the profiles by solarzenith angle.

9. SUMMARY

This analysis of monthly means of the MIPAS Level2 products shows that in general, MIPAS is producinggood, useful data. There are, however, some persitentproblems with the profiles visible in the monthly means,which have been outlined in the sections above.

ACKNOWLEDGMENTS

We would like to acknowledge all of the MIPAS team,particularly those responsible for the development of theoperational retrieval, and the MIPAS QWG for their in-put.

REFERENCES

Remedios, J.J., Extreme atmospheric constituent pro-files for MIPAS, Proceedings of the European Sympo-sium on Atmospheric Measurements from Space, ES-TEC, Noordwijk, Netherlands, 20-22 January, Vol 2,779-783, 1999

http://badc.nerc.ac.uk/data/cira

http://www.atm.ox.ac.uk/group/mipas

Mote, P.W. et al, An atmospheric tape recorder: The im-print of tropical tropopase temperatures on stratosphericwater vapour, J. Geophys. Res., 101, 3989-4006, 1996

Figure1. Timeseriesof MIPASnear–real–timetemperature retrievals,fromJuly 2002 to March 2004.

Figure 2. Time seriesof absolute differenceof MIPASnear–real–time temperature retrievals from CIRA climatology(MIPAS - CIRA), from July 2002 to March 2004. Redshowswhere MIPASis hotter than CIRA andblue showswhereMIPASis colder.

Figure3. Timeseriesof MIPASnear–real–timeH � O retrievals,fromJuly 2002 to March 2004.

Figure 4. Timeseriesof percentage differenceof MIPASnear–real–timeH � O retrievals fromtheIG2 climatology, fromJuly 2002 to March 2004. Plots show

�������� ��� ���� � ������� . Redshowswhere MIPASH � O values are higher than theclimatology andblueshowswhereMIPASvaluesare lower thantheclimatology.

Figure5. Timeseriesof MIPASnear–real–timeO � retrievals,fromJuly 2002 to March 2004.

Figure 6. Time seriesof percentage differenceof MIPAS near–real–time O � retrievals from the IG2 climatology, fromJuly 2002 to March 2004. Plots show

�������� ��� ���� � ������� . Redshowswhere MIPASH � O values are higher than theclimatology andblueshowswhereMIPASvaluesare lower thantheclimatology.

Figure7. Timeseriesof MIPASnear–real–timeHNO� retrievals,fromJuly 2002to March 2004.

Figure8. Timeseriesof percentage difference of MIPASNear–real–timeHNO � retrievalsfromtheIG2 climatology, fromJuly 2002 to March 2004. Plots show

�������� ��� ���� � ������� . Redshowswhere MIPASH � O values are higher than theclimatology andblueshowswhereMIPASvaluesare lower thantheclimatology.

Figure9. Timeseriesof MIPASnear–real–timeCH � retrievals,fromJuly 2002to March 2004.

Figure 10. Timeseriesof percentage differenceof MIPASnear–real–timeCH � retrievals fromtheIG2 climatology, fromJuly 2002 to March 2004. Plots show

�������� ��� ���� � ������� . Redshowswhere MIPASH � O values are higher than theclimatology andblueshowswhereMIPASvaluesare lower thantheclimatology.

Figure11. Timeseriesof MIPASnear–real–timeN � O retrievals,fromJuly 2002to March 2004.

Figure 12. Timeseriesof percentage differenceof MIPASnear–real–timeN � O retrievals fromtheIG2 climatology, fromJuly 2002 to March 2004. Plots show

�������� ��� ���� � ������� . Redshowswhere MIPASH � O values are higher than theclimatology andblueshowswhereMIPASvaluesare lower thantheclimatology.

Figure13. Timeseriesof MIPASnear–real–timeNO� retrievals,fromJuly 2002 to March 2004 (day–timeprofilesonly).

Figure14. Timeseriesof MIPASnear–real–timeNO� retrievals,fromJuly 2002to March 2004(night–timeprofilesonly).

Figure 15. Time seriesof percentage differenceof MIPASnear–real–timeday–timeNO � retrievals froma mid–latitudeday–timeclimatology (1).

Figure 16. Timeseriesof percentage differenceof MIPASnear–real–timenight–timeNO � retrievalsfroma mid–latitudenight–timeclimatology (1).