A COMPARISON OF DAYTIME AND NIGHT-TIME OZONE PROFILES FROM GOMOS AND MIPAS

Transcription

1 A COMPARISON OF DAYTIME AND NIGHT-TIME OZONE PROFILES FROM AND P. T. Verronen 1, E. Kyrölä 1, J. Tamminen 1, V. F. Sofieva 1, T. von Clarmann 2, G. P. Stiller 2, M. Kaufmann 3, M. Lopéz-Puertas 4, B. Funke 4, and D. Bermejo-Pantaleon 4 1 Finnish Meteorological Institute, Earth Observation, 1 Helsinki, Finland 2 Forschungszentrum Karlsruhe (IMK-ASF), 721 Karlsruhe, Germany 3 Forschungszentrum Jülich (ICG-I), 524 Jülich, Germany 4 Instituto de Astrofisica de Andalucia (CSIC), 188 Granada, Spain ABSTRACT We present a comparison of daytime and night-time ozone data from and. The comparison covers stratospheric and lower mesospheric altitudes, from to 68 km. For the night-time data, the results indicate good agreement between and. For the daytime data, the agreement is poor below km and the results depend strongly on the properties of the target star. The poor daytime agreement below km is likely due to very low signal-to-noise ratios of the daytime observations. Key words: Ozone; Stratosphere; Mesosphere; ;. 1. INTRODUCTION The European Space Agency s Envisat satellite was launched into a polar orbit on March 1, 2. On board are instruments observing the Earth and its atmosphere, these include and. (Global Ozone Monitoring by Occultation of Stars) is a stellar occultation instrument that measures altitude profiles of several atmospheric species, including O 3, and temperature [1, 2, 3]. The altitude range of measurements is km for ozone and km for the other gases. makes several hundred measurements per day with good global coverage including the polar areas. (Michelson Interferometer for Passive Atmospheric Sounding) is a high-resolution (.5 cm 1, apodised) mid-ir limb sounder that allows measurement of the kinetic temperature and a large number of atmospheric species, including O 3, with good global coverage including the poles [4, 5, 6]. standard mode of observation covers tangent altitudes from 6 to 68 km. In the stratosphere, mesosphere, and lower thermosphere, case studies have indicated a good agreement between and night-time ozone observations [7], with data from observations made in the upper atmosphere mode #1 [8]. At 7km and 85 km, showed in general about 5% and % larger values, respectively. Around the 8 km ozone minimum the differences were larger but still within ±%. 2. DATA CHARACTERISTICS ozone data were produced with the current ESA operational ground segment, version /5.. The altitude sampling resolution depends on the measurement geometry but is always 1.7 km. In the retrieval a Tikhonov-type regularisation is applied so that the altitude resolution of the ozone product is 2 3 km. The measurement error depends on the star temperature and magnitude. In the case of ozone, the bright and hot stars result in a good accuracy with 5% estimated errors around both the primary and secondary maximum [9]. To ensure good accuracy, we preselected data with following criteria: 1) target star visual magnitude was required to be 1.9 and 2) stellar temperature of night and day observations was required to exceed 7 K and K, respectively. Hereafter, this preselected set is referred to as all stars. ozone data are from the IMK-IAA retrieval processor [, 11]. The data are based on the ESA Level 1b spectra, version 4.61 or higher (V3O), and have been processed with retrieval baseline version of 7 or 8. In case of ozone, the differences in the retrieval set-up between these two baseline versions are small, so that mixing of them does not produce any significant artifacts. Altitude resolution of the data is 4 km in the stratosphere and 8 km in the mesosphere. The estimated total error of ozone retrieval, including random and systematic errors, varies between 7 and 14% above km. All data used in this comparison are from the year 3, selected mainly because most of the IMK-IAA data exist, at the moment, for this year. Proc. Envisat Symposium 7, Montreux, Switzerland April 7 (ESA SP-636, July 7)

2 Table 1. Number of coincidences and median distances for the preselected set of target stars. Night Star # Coincidences Distance (km) Day Star # Coincidences Distance (km) COINCIDENCE CRITERIA The data analysis began with a search for temporal and spatial coincidences of observations. Because of the strong diurnal variation of ozone in the upper stratosphere and mesosphere, the data were first divided into daytime and night-time observations which were then treated separately. The selected daytime observations have SZA 85 while night-time observations have SZA 1. The twilight measurements were therefore excluded from this study. When searching for coincidences, the following criteria were used: 1. difference in latitude lat 5 2. difference in longitude lon 3. difference in SZA sza 4. difference in time time 3h With criteria 1 4 applied we have a total of 91 and 8 coincidence cases for the nigh-time and daytime, respectively. The median distance between coincidence locations is 4 km for night-time and 5 km for daytime. Table1 lists the coincidence numbers/distances on a starby-star basis. 4. CALCULATING DIFFERENCES For each pair of / coincidence profiles of ozone we performed the following comparison procedure: 1. data were interpolated into the fixed MI- PAS altitude grid, which has 1-km and 2-km steps between 44 km and km, respectively. 2. concentrations (cm 3 ) were converted to mixing ratios (ppmv) using the ideal gas law pv = nrt and the coincidence pressure and temperature. 3. mixing ratio profile was multiplied by the averaging kernel matrix in order to compensate for the different altitude resolutions (see, e.g., [12]). 4. Absolute difference between the and profiles was calculated in mixing ratio: = vmr MIP AS vmr, i.e. we took as the reference point which was then compared to. We characterise the distribution of the differences by the following statistical parameters. = median( ), i.e. median of absolute difference, i.e. bias. σ vmr = sid( ), i.e. semi-interquartile deviation of absolute bias (% of the data are within ± σ vmr ). = /median(m vmr ), i.e. median of relative differences, i.e. relative bias. σ rel = σ vmr /median(m vmr ), i.e. semiinterquartile deviation of relative bias. median(m vmr ) is the median of mixing ratio profiles. 5. COMPARISON RESULTS Figure 1 shows the comparison for the night-time data, including all target stars. The median mixing ratio profiles are very similar in shape and magnitude, and are in a good agreement. The relative bias,, varies from 6% to +3% at km, generally shows larger ozone values. Above km, shows 3 9% larger values than. In absolute numbers, is in the range from.5 to +.2 ppmv at all altitudes. For the daytime data, shown in Figure 2, the situation is different. The median mixing ratio profiles are similar above km, but at km gives significantly lower values of ozone, and much higher values below 33 km. rel and are rapidly increasing below km, exceeding %/1 ppmv at about 46 km. At these altitudes, the agreement between GO- MOS and is poor. Above km, the difference is smaller, i.e. ranging from % to +11%, or from.1 to +.1 ppmv in mixing ratio.

3 9 Coincidences / all stars N = 91 SZA = o LAT = o all stars Figure 1. Comparison of and ozone, night-time observations. The panels from left to right are: (1) geographic locations of the coincidences, (2) median profiles, (3) relative bias (black) and its semi interquartile deviation (blue), and (4) absolute bias (black) and its semi-interquartile deviation (blue). 9 Coincidences / all stars N = 8 SZA = o LAT = o all stars Figure 2. As Figure 1, but with daytime observations.

4 2 Altitudes: km, night time Altitudes: km, daytime Altitudes: 68 km, night time Altitudes: 68 km, daytime star number star number Figure 3. Night-time and daytime relative bias at two altitude ranges: km and 68 km. is shown individually for each target star. The dashed horizontal lines indicate minimum, average, and maximum bias. Note that for clarity, not all star labels are shown. Coincidences / Star #29 9 N = 85 SZA = o LAT = o star #29 T = K Mv = Figure 4. As Figure 1, but showing the results for nigh-time observations using target star #29 exclusively.

5 Coincidences / Star #5 9 N = 87 SZA = o LAT = o star #5 T = 1 K Mv = Figure 5. As Figure 1, but showing the results for daytime observations using target star #5 exclusively. Because of the known dependency of the accuracy on the target star spectra, we present in Figure 3 the difference between and on a star-by-star basis, and also divide the altitude into two regions: below and above km. Night-time data show variations of from star to star. The minimum, average, and maximum numbers of bias are listed in Table 2. For night-time observations at altitude range km, target star selection is not crucial. Most stars have 2%, but even the worst ones have 6%. At 68 km, however, most of the stars show +4% while some show bias of several tens of percent. It should be noted that these results may not be conclusive for all stars because the number of coincidences varies significantly from star to star (see Table 1). For daytime observations at altitude range km, most stars have in tens of percent. Star #5 compares to clearly much better than the rest, although it still has 19%. At 68 km, stars #5, #19, #22, and #28 compare very well with MI- PAS, within ±5%, while some others show bias up to %. In Figures 4 and 5 we compare and using target stars #29 and #5, respectively. In principle, best ozone observations are made with hot and bright stars, but Figure 4 shows that at night even relatively dim stars like #29, with visual magnitude 1.67, give good agreement with the observations. On the other hand, Figure 5 shows that at daytime even the brightest stars like #5 have a rather poor agreement with at altitudes below km. 6. SUMMARY We have compared daytime and night-time ozone measurements from the and instruments in the stratosphere and lower mesosphere. The night-time data show good general agreement. The comparison results are not strongly dependent on the target star selection. The night-time difference of to is 6% +3% at km, and +3 +9% at 68 km. The daytime data shows poor agreement at altitudes below km. The comparison results differ significantly from star to star. Daytime difference of to worse than % at km, and % +11% at 68 km. Table 2. Variation of ozone bias with target star. Minimum / average / maximum values of and are given. Night Altit. range vmr (ppmv) rel (%) km.36 /.14 / / 3 / km.2 / +.5 / +. 2 / +4 / +33 Day Altit. range vmr (ppmv) rel (%) km 4.7 / 2.22 / / 44 / km.3 / +.4 / / +4 / +17

6 7. CONCLUSIONS AND FUTURE WORK 1. Based on this comparison, confidence in the GO- MOS and night-time data is high. 2. and daytime observations are in reasonable agreement in the mesosphere but more validation is needed, especially at altitudes above 68 km. 3. Poor daytime agreement below km is likely due to very low signal-to-noise ratios of the daytime observations. It should be noted that currently the same data processor is used for both day and night observations, so that a more carefull optimisation might improve the daytime retrievals. A new processing line based on the limb scattering measurements is foreseen in the near future. A combination of limb and stellar measurements would give a new way to obtain better daytime ozone profiles covering both the stratosphere and the mesosphere. We will continue the / comparisons. Especially, we are interested to study the mesospheric altitudes using special-mode data which extends up to km [8]. From 7 on, more mesospheric data will be available because the special, upper atmosphere mode is operated more frequently, i.e. every tenth day. We are planning to compare also to the official ESA products, which due to larger amount of data available will give us more coincidence cases. REFERENCES [1] J. L. Bertaux, G. Megie, T. Widemann, E. Chassefiere, R. Pellinen, E. Kyrölä, S. Korpela, and P. Simon. Monitoring of ozone trend by stellar occultations: The instrument. Adv. Space Res., 11(3): , [2] J. L. Bertaux, A. Hauchecorne, F. Dalaudier, C. Cot, E. Kyrölä, D. Fussen, J. Tamminen, G. W. Leppelmeier, V. Sofieva, S. Hassinen, O. Fanton d Andon, G. Barrot, A. Mangin, B. Théodore, M. Guirlet, O. Korablev, P. Snoeij, R. Koopman, and R. Fraisse. First results on /Envisat. Adv. Space Res., 33:29, 4. [3] E. Kyrölä, J. Tamminen, G. W. Leppelmeier, V. Sofieva, S. Hassinen, J.-L. Bertaux, A. Hauchecorne, F. Dalaudier, C. Cot, O. Korablev, O. Fanton d Andon, G. Barrot, A. Mangin, B. Theodore, M. Guirlet, F. Etanchaud, P. Snoeij, R. Koopman, L. Saavedra, R. Fraisse, D. Fussen, and F. Vanhellemont. on Envisat: An overview. Adv. Space Res., 33: 28, 4. [4] H. Fischer and H. Oelhaf. Remote sensing of vertical profiles of atmospheric trace constituents with limb-emission spectrometers. Appl. Opt., (16): , [5] M. Endemann and H. Fischer. Envisat s high resolution limb sounder:. ESA bulletin, 76:47 52, [6], Michelson interferometer for passive atmospheric sounding, an ENVISAT instrument for atmospheric chemistry and climate research, scientific objectives, mission concept and feasibility, instrument design and data products. European Space Agency, [7] P. T. Verronen, E. Kyrölä, J. Tamminen, B. Funke, S. Gil-López, M. Kaufmann, M. López-Puertas, T. von Clarmann, G. Stiller, U. Grabowski, and M. Höpfner. A comparison of night-time and ozone profiles in the stratosphere and mesosphere. Adv. Space Res., 36: , 5. [8] S. Gil-López, M. Kaufmann, B. Funke, M. García- Comas, M.E. Koukouli, M. López-Puertas, N. Glatthor, U. Grabowski, M. Höpfner, G. P. Stiller, and T. von Clarmann. Retrieval of stratospheric and mesospheric O 3 from high resolution spectra at and µm. Adv. Space Res., 36: , 4. [9] E. Kyrölä, J. Tamminen, G. W. Leppelmeier, V. Sofieva, S. Hassinen, A. Seppälä, P. T. Verronen, J.-L. Bertaux, A. Hauchecorne, F. Dalaudier, D. Fussen, F. Vanhellemont, O. Fanton d Andon, G. Barrot, A. Mangin, B. Theodore, M. Guirlet, R Koopman, L. Saavedra, P. Snoeij, and T. Fehr. Nighttime ozone profiles in the stratosphere and mesosphere by the Global Ozone Monitoring by Occultation of Stars on Envisat. J. Geophys. Res., 111:D246, 6. [] N. Glatthor, T. von Clarmann, H. Fischer, B. Funke, S. Gil-López, U. Grabowski, M. Höpfner, S. Kellmann, A. Linden, M. López-Puertas, G. Mengistu Tsidu, M. Milz, T. Steck, G. P. Stiller, and D. Y. Wang. Retrieval of stratospheric ozone profiles from /ENVISAT limb emission spectra: a sensitivity study. Atmos. Chem. Phys., 6: , 6. [11] T. Steck, T. von Clarmann, H. Fischer, B. Funke, N. Glatthor, U. Grabowski, M. Höpfner, S. Kellmann, M. Kiefer, A. Linden, M. Milz, G. P. Stiller, D. Wang, M. Allaart, T. Blumenstock, P. von der Gathen, G. Hansen, F. Hase, G. Hochschild, G. Kopp, E. Kyrö, H. Oelhaf, U. Raffalski, A. Redondas Marrero, E. Remsberg, J. Russell III, K. Stebel, W. Steinbrecht, G. Wetzel, M. Yela, and G. Zhang. Bias determination and precision validation of ozone profiles from -Envisat retrieved with the IMK-IAA processor. Atmos. Chem. Phys. Discuss., 7: , 7. [12] C. D. Rodgers and B. J. Connor. Intercomparison of remote sounding instruments. J. Geophys. Res., 8(D3), 3.