AN ANNUAL CYCLE OF STRATOSPHERIC OZONE, WATER VAPOUR AND METHANEFROM MIPAS

Martin N. Juckes

British Atmospheric Data Centre, SSTD, Rutherford Appleton Laboratory, Chilton, Didcot, Oxfordshire, OX11 0QX,United Kingdom

ABSTRACT

MIPAS, on ENVISAT, has made high quality observa-tions of ozone, methane and water vapour. Gridded fields,at 4 hourly intervals and, have been calculated for all of2003 using data assimilation with isentropic advection asa constraint. The gridded fields are validated against in-dependent measurements (from 7 other instruments in thecase of ozone, 3 for water vapour and one for methane).For ozone the results are in agreement with previouslypublished results. For water vapour the bias relative toHALOE is below 10% between 20 and 48km, and thestandard deviation is below 12% in this range. Departuresfrom SAGE II and POAM III are substantially larger. Themethane analysis has a bias of less than 5% relative toHALOE between 23 and 40km, with a standard devia-tion less than 10% in this height range. The water vapourfield clearly reflects the upward motion in the lower trop-ical stratosphere, while both water vapour and methaneshow the signature of advection higher up. In the polarregions the descent in the vortex is clearly visible, withstrong descent in autumn giving way to weaker descentthrough the winter. Descent rates of around 10−3ms−1

are found during the formation of the polar vortices, slow-ing to around 3 × 10−4ms−1 during the winter. Ascentof around 2 × 10−4ms−1 in the tropics is revealed bythe water vapour and total observed hydrogen fields (4times the methane plus twice the water vapour concen-tration). The total observed hydrogen is depleted in thepolar upper stratosphere when air is advected down fromthe upper mesosphere.

Key words: ozone; methane; water vapour; assimilation;validation.

1. INTRODUCTION

Michelson Interferometer for Passive AtmosphericSounding (MIPAS) instrument (Fischer & Oelhaf 1996;Tsidu et al. 2003; Glatthor et al. 2005) on the ENVISAT1

1The European Space Agency’s ‘Environmental Satellite’, launchedon Friday March 1, 2002.

was operational at full resolution for 20 months. The cen-tral 12 month period (January to December 2003) hadnear continuous global coverage. This gives unprece-dented resolution of the day to day evolution of ozone,water vapour and methane. Here, assimilated fields ofthese tracers are validated against a wide range of inde-pendent observations. The assimilation method is pre-sented in Juckes (2006), and these results are discussedin more detail in Juckes (2007).

2. METHODS AND INPUTS

The analysis is defined as the minimum of a quadraticcost function (see Juckes 2006). The cost function con-strains the analysis to be close to the observations and toapproximately satisfy Dχ/Dt = 0, where χ is the tracermixing ratio and D/Dt is the isentropic Lagrangianderivative.

The solution is found by setting the gradient of the costfunction to zero:

−wapD2χ

Dt2− wnum

(c21∂2χ

∂t2+∇2χ

)3

+ (1)∑

i

σ−2o:i [χo:i − χ(λo:i, φo:i, to:i)]× (2)

cosφδ(λ− λo:i)δ(φ − φo:i)δ(t− to:i) = 0,(3)

λ, φ, t are longitude, latitude and time respectively, σ2o:i

is the observation error variance, with the subscript i hereand elsewhere labelling the observation. Constants usedhere are c1 = 0.5days per radian= 0.0068s m−1, wap =

50day2 ppmv−2 = 3.7 × 1011s2 ppmv−2, and wnum =1.45× 10−6radian6 ppmv−2 = 9.7× 1034m6 ppmv−2.The summation is over all observations. The large valueof the wnum in SI units reflects the scale-selectivity ofthis term, which ensures the solution is well behaved onthe grid scale. There are no prescribed error covariancestructures in this formulation: the variational problem iscompletely defined by the constants quoted above and theobservational data (χo:i, σo:i, i = 1, No, where No isthe number of observations). The solution method andthe choice of constants c1, wap and wnum is discussed inmore detail in Juckes (2006).

_____________________________________________________

Proc. ‘Envisat Symposium 2007’, Montreux, Switzerland 23–27 April 2007 (ESA SP-636, July 2007)

The analysis is based on the MIPAS level 2 retrievals dis-tributed by ESA (see Table 1 for more details). Windsfrom the operational analyses of European Centre forMedium-range Weather Forecasting (ECMWF) are usedto evaluate the Lagrangian derivatives in the analysisequation (1).

Data from several satellite instruments and from theWOUDC sonde data archive2 are used to validate theanalyses. More details about the different data sourcesare given in table 1.

The analysis is carried out on isentropic surfaces. Fieldsare interpolated with their own retrieved temperature pro-files where available (MIPAS, ozonesondes, HALOE),temperature profiles distributed with the data (SAGE,POAM, both of which use NMC temperatures), or withECMWF temperatures interpolated to the profile position(SBUV and OSIRIS).

3. RESULTS

Fig. 1 shows validation data for ozone on the 500K sur-face. The results are shown by latitudinal bins and bymonth. The figure also includes the observation minusanalysis statistics for MIPAS, showing that the analysisis unbiased relative to the input data and that the residualvariance is around 0.2ppmv at this level.

In the two southern bands (90S to 60S and 60S to 30S)the validation instruments are generally high relative tothe MIPAS analysis.

In the remaining four latitude bands the analysis is gen-erally lower than SBUV. Differences from other instru-ments, however, do not show a clear pattern.

Figure 2 shows a similar comparison for water vapour.There is a substantial offset between the analysis andPOAM III, but agreement with HALOE and SAGE IIis better, with monthly mean anomalies in the 30o binsgenerally less than 0.5ppmv. The analysis water vapourconcentrations are generally low relative to the latter twoinstruments.

In the winter polar regions, particularly in the southernhemisphere, the MIPAS observations minus analysis vari-ance increases markedly. In the southern hemisphere theobservation minus analysis has a small negative meanvalue in July, August and September.

The methane validation data at 500K (Fig. 3) show a clearseasonal cycle in the MIPAS observation minus analysisstatistics in both hemispheres, with larger residual vari-ances in the winter. The HALOE observations are gen-

2WOUDC is one of the World Data Centres which are part of theGlobal Atmosphere Watch programme of the World MeteorologicalOrganisation. The WOUDC is operated by the Experimental Stud-ies Division of the Meteorological Service of Canada (web address:www.woudc.org)

Figure 1. Ozone validation data, 500K, January to De-cember, 2003. Monthly means (observation minus analy-sis) and standard deviations by latitude bands. Each lat-itude band is divided into 12 months, with April, May,June and October, November, December shaded. Foreach month, where data is present, a horizontal barmarks the mean observation minus analyses, and the ar-rows plus and minus one standard deviation. A bar with-out arrows occurs when there is a single validating obser-vation in that month and latitude band. Bars are drawndashed when they overlap bars higher up the graphic.SAGE III [L] is the data from the lunar occultation modeof SAGE III. Sondes [S] is data from the WOUDC ozonesondes which has been vertically smoothed with a 2kmrunning mean.

Figure 2. As figure 1, but for water vapour.

Instrument Ref. Level 2version(s)

Ozoneppd

WaterVapourppd

Methaneppd

Comments

Haloe Bruhl et al. (1996) 19 13.2 12.4 10.9Mipas ESA (2004) 4.59, 4.61

& 4.62650 622 625

Osiris von Savigny et al.(2003)

1.2 165 – –

Poam III Lumpe et al.(2003); Pierceet al. (2003)

4 14.3 14.4 – High latitude only

SBUV Planet et al.(2001)

61610 1021 – – Daytime only

Sage II Thomason & Taha(2003)

6.2 12.5 12.5 –

Sage III Chu & MauldinIII (1999)

3 24.2 – – High latitude only

Sage III(L) 3 0.9 – –WOUDCOzonesondes

www.woudc.org 3.2 – – Fewer above 850K

Table 1. Information about the data used. Columns 4 to 6 give the average number of profiles per day (ppd) used in thisstudy.

Figure 3. As figure 1, but for methane

erally lower than the analysis, with an offset of around0.1ppmv.

Figure 4 shows the ozone validation data at 1250K. Thesonde data is sparse at this altitude, but is included forcompleteness. Osiris shows large negative biases relativeto the analysis. Between 60S and 60N the SAGE mea-surements, POAM III, HALOE and SBUV all measurehigh relative to the analysis, though for HALOE the off-sets are marginal. The neglect of chemistry in the analysisis probably having a detrimental effect at this altitude.

In the water vapour (Fig. 5) we again (as in Fig. 3) seea seasonal cycle in the polar MIPAS observation mi-nus analysis variances, with higher values in the winter.All three validation instruments have a consistent nega-tive bias of around 0.5ppmv, suggesting that MIPAS ismeasuring high at this altitude. The results presented in

Figure 4. As figure 1, but for the 1250K isentropic surface

Figure 5. As figure 1, but for water vapour on the 1250Kisentropic surface

Figure 6. As figure 1, but for methane on the 1250K isen-tropic surface

Juckes (2007) show, however, that it is only at this onelevel that the three occultation instruments all measurelower water vapour than the assimilated MIPAS: POAMIII is high relative to the analysis below 1250K and SAGEII is high relative to the analysis above 1250K. The resid-ual variances are large for POAM III and SAGE II, butmoderate, at a standard deviation of around 0.3ppmv, forHALOE.

The methane validation shows good agreement at 1250K(Fig. 6), with no consistent offset relative to HALOE andstandard deviations of around 0.04ppmv.

Results in the above format for the 850K surfaceare presented in Juckes (2007). Results for otherlevels (from 460K to 3000K) will be posted onhome.badc.rl.ac.uk/mjuckes/mista.

4. CONCLUSIONS

MIPAS measurements of ozone, water vapour andmethane have been validated against a wide range of in-

dependent data.

The assimilated MIPAS concentrations generally liewithin the range spanned by other instruments.

ACKNOWLEDGEMENTS

This work was funded by NERC through BADC. I amgrateful for all the data used in this study, which was ofhigh quality and provided freely from the sources cited. Iwould like to thank Anu Dudhia and Chiara Piccolo forhelp with the MIPAS data.

REFERENCES

Bruhl C., Drayson S.R., II J.M.R., et al., 1996, J. Geo-phys. Res., 101, 10217

Chu W., Mauldin III L., 1999, In: Larar A.M. (ed.) Opti-cal Spectroscopic Techniques and Instrumentation forAtmospheric and Space Research III, vol. 3756, 102–109, SPIE

ESA, 2004, MIPAS product handbook, European SpaceAgency, ESRIN, Frascati, Italy, available from en-visat.esa.int/dataproducts

Fischer H., Oelhaf H., 1996, Appl. Opt., 35, 2787Glatthor N., Von Clarmann T., Fischer H., et al., 2005, J.

Atmos. Sci., 62, 787Juckes M.N., 2006, Atmospheric Chemistry and Dynam-

ics, 6, 1549Juckes M.N., 2007, Atmospheric Chemistry and Dynam-

ics, 7, 1879Lumpe J.D., Fromm M., Hoppel K., et al.,

2003, J. Geophys. Res., 108(D5), 8316, doi:10.1029/2001JD000472 [2002]

Pierce R.B., Al-Saadi J., Fairlie T.D., et al.,2003, J. Geophys. Res., 108(D5), 8317, doi:10.1029/2001JD001063 [2002]

Planet W.G., Miller A.J., Horvath K., et al., 2001, J. Geo-phys. Res., 106(D15), 17,471

von Savigny C., Haley C.S., Sioris C.E., et al., 2003, Geo-phys. Res. Lett., 30, 1755

Thomason L.W., Taha G., 2003, Geophysical ResearchLetters, 30, 33

Tsidu G.M., Kiefer M., von Clarmann T., et al., 2003, In:H.-L. Huang Y.S. D. Lu (ed.) Optical Remote Sensingof the Atmosphere and Clouds III, vol. 4891, 483–496,SPIE