Supplement of Atmos. Meas. Tech., 8, 5133 5156, 215 http://www.atmos-meas-tech.net/8/5133/215/ doi:1.5194/amt-8-5133-215-supplement Author(s) 215. CC Attribution 3. License. Supplement of Cloud and aerosol classification for 2.5 years of MAX-DOAS observations in Wuxi (China) and comparison to independent data sets Y. Wang et al. Correspondence to: Y. Wang (y.wang@mpic.de) The copyright of individual parts of the supplement might differ from the CC-BY 3. licence.
Supplement: Analysis of the O 4 absorption The analysis of the O 4 dscds is performed in the spectral range of 35 nm - 391 nm covering two O 4 absorption bands using the DOAS method (Platt and Stutz, 28). For the analysis the WINDOAS software (Fayt and van Roozendael, 29) was used. Apart from the O 4 cross section at 296 K (Greenblatt et al., 199), also cross sections of NO 2 at 294 K (Vandaele et al., 1998) and O 3 at 293 K (Bogumil, 23) were included. Effects of broad band scattering and extinction were accounted for by including a third order polynomial. The effect of rotational Raman scattering is considered by including a Ring spectrum (Grainger and Ring, 1962; Solomon et al., 1987; Wagner et al., 29) computed by the DOASIS software (Kraus, 26). Also an intensity offest is included in the fitting routine. All spectra are allowed to be shifted and stretched against a Fraunhofer reference spectrum (FRS). Fig. S1 shows typical DOAS fit results for the O 4 analysis. It should be noted that we discard 69 measurement sequences, for which the root-mean-square (RMS) deviation of the residual structure larger than 3 1-2 or shifts larger than 1 detector pixels. The O 4 concentration is proportional to the square of the oxygen concentration (Greenblatt et al., 199) and thus varies only slightly with time (due to variations of temperature and pressure). Thus for the conversion to AMF (see equations 3 and 4 in the main text) we used a constant VCD O4 of 5 1 43 molecules 2 cm -5 (for the unit see Greenblatt et al. (199)). This value is calculated from average values of the surface temperature (29K) and surface pressure (11 hpa) at Wuxi. Here it should be noted that during the course of one year we found deviations of the O 4 VCD from the selected value of 1 %. Since in this study we use the O 4 measurements only for the identification of fog (see main text), these deviations don t have an effect on the cloud primary classification results. Also the effect of the fog classification is very small. For the interpretation of the O 4 AMF, another important aspect has to be considered: to remove the strong solar Fraunhofer lines, a FRS is included in a DOAS
differential optical density fit. Since the FRS used in our analysis also contains atmospheric O 4 absorption, the result of the DOAS fit actually represents the difference between the O 4 SCDs of the measured spectrum and the FRS. This difference is usually referred to as differential SCDs (dscd). Like the O 4 SCDs (eq. 3 in the main text), also the O 4 dscd can be converted into the corresponding O 4 damf (eq. 4 in the main text). We investigated the effect of the instrumental degradation on the O 4 dscds. We find systematic temporal variations of O 4 dscd (see Fig. S2a) that are probably caused by variations of the spectral resolution of the instrument (see Fig. S2b). Because of the systematic variations we did not use the O 4 absorption measured in zenith for the identification of optically thick clouds. Instead we identify optically thick clouds by largely reduced values of the measured radiances. 4 2-2 measured fitted Residual RMS= 1.1 1-3 offset polynomial 4 2.15.1 5-5 28 24 2 16 NO 2 dscd = 3.76 1 16 molecules cm -2-4 4 2 O 3 dscd = 4.54 1 18 molecules cm -2-2 9 6 3 O 4 dscd = 6 1 molecules 2 cm -5 -.1-8 -12 ring=-4.29 1 25 3-16 -2-2 -24-4 35 36 37 38 39-3 35 36 37 38 39 Wavelength [nm] 35 36 37 38 39 Fig. S1 Example of a DOAS fit for a spectrum measured at an elevation angle of 5 at 11:47 on September 15, 212 using the Fraunhofer reference spectrum from the same elevation sequence. The red lines indicate the trace gas cross sections scaled to the corresponding absorptions in the measured spectrum; the black lines indicate the respective fit results. The residual (top left) indicates the difference between the measured spectrum and all fitted spectra.
O 4 dscd [1 4 molecules 2 cm -5 ] (a) (b) 15 1 5 Jun Dec Jun Dec Jun Dec 211 212 213 month Fig. S2 (a): O 4 dscd for zenith view in the SZA range of 49 to 51 from the retrievals with a fixed reference spectrum (from Sep 15, 212). (b): The full width at half maximum and spectral shift (red points) of the measured spectra derived from a fit to a high-resolution solar spectrum. TOA reflectance at 65 nm for SZA of 1 polynomial fitting curve 1..8.6.4.2 2 4 6 8 1 cloud optical thickness Fig. S3 Top of the atmosphere (TOA) reflectance at 65 nm for nadir view for SZA of 1 as a function of the cloud optical thickness derived from radiative transfer simulations. The cloud optical properties are described by the Henyey Greenstein approximation with an asymmetry parameter of.85 and a single scattering albedo of 1. The cloud layer range is from 1 to 2 km for COT of 1 to 15, from 1 to 3 km for COT of 2 to 3, from 1 to 4 km for COT of 35 to 4, and from 1 to 5km for COT of 5 to 1.
aerosol optical depth ration of AOD at 44 nm / 34 nm.4.3.2.1 at 44 nm at 38 nm at 34 nm ratio of AOD.75.7.65.6.55.5 Fig. S4 Time series of the AOD at 34 nm (black), 38 nm (red) and 44 nm (green) on 29 July 212 as well as the ratio of the AOD at 44 and 34 nm (black) (right axis) in the sky condition of extremely high midday CI. Comparison of cloud products from MODIS, OMI and GOME-2 We briefly discuss the satellite cloud products in more detail and compare them with each other. Here we make use of the fact that the overpass time of MODIS on Terra is similar to that of GOME-2, and the overpass time of MODIS on Aqua is similar to that of OMI to compare the cloud products from GOME-2 and OMI with MODIS. In Fig. S5a the effective CFs from OMI are compared with the corresponding MODIS results (on Aqua). Because the effective CF is clipped between and 1, we excluded the data with effective CF of 1% and smaller than 1% for this comparison. A good agreement with the fitted linear slope of.71, the intercept of 6.7% and the square of correlation coefficient (R 2 ) of.61 is found. The different spatial resolution and slightly different overpass time of the two instruments contribute to the scattering of the points. The different retrieval techniques may also contribute to the discrepancy of effective CFs between both satellite instruments. In Fig. S5b in the same way the effective CFs from GOME-2 are compared with the corresponding MODIS results (on Terra). A good agreement with the fitted linear
CP from OMI [hpa] CP from GOME-2 [hpa] effective CF from OMI [%] effective CF from GOME-2 [%] slope of.86, the intercept of 14% and the R 2 of.61 is found. The larger effective CF from GOME-2 than that from MODIS is found, especially when the COT is low. This phenomenon has been found by Acarreta et al. (24). Besides different spatial resolution and overpass time, the different spatial collocation criteria of GOME-2 and MODIS also contribute to the discrepancy between them. The retrieval algorithms using the O 4 absorption for OMI and the O 2 (FRESCO+) for GOME-2 normally retrieve much larger CP than MODIS, especially for optically thick clouds. The reason is that multiple scattering leads to the fact that the cloud pressure from the O 4 absorption algorithm and FRESCO+ is closer to the middle of the clouds than to the top of the clouds (Acarreta et al., 24; Wang et al., 28; Wang and Stammes, 214). The comparisons of the CPs from OMI and GOME-2 with those from MODIS in Fig. S5c and d also show the same feature. (a) R 2 =.62, slope=.71, intercept = 6.7 1 8 6 4 COT from MODIS 1 2 3 4 5 (b) 1 8 6 4 R 2 =.64, slope=.86, intercept=14 2 2 2 4 6 8 1 effective CF from MYD [%] (c) 1 8 6 4 2 R 2 =.22, slope=.31, intercept=499 2 4 6 8 1 effective CF from MOD [%] (d) 1 8 6 4 2 R 2 =.48, slope =.49, intercept=46 2 4 6 8 1 CP from MYD [hpa] 2 4 6 8 1 CP from MOD [hpa] Fig. S5 Comparison of cloud products derived from GOME-2 or OMI and MODIS: effective CF from OMI and GOME-2 are plotted against those from MODIS on Aqua (MYD) a) and on Terra (MOD) b), respectively; CPs from OMI and GOME-2 are plotted against those from MODIS on Aqua (MYD) c) and on Terra (MOD) d), respectively. The black dashed line indicates the 1:1 line. The colours indicate the COT from MODIS. The comparisons are performed for satellite measurements over the Wuxi site. For the comparison of CPs, the data with corresponding effective CF <
2% are excluded to only keep the cloudy conditions, because the CP can t be accurately retrieved for effective CF < 2%. Selection of threshold values The procedure for the selection of threshold values is described. Firstly the days with specific sky conditions were selected based on visual images from MODIS (Fig. S6), time series of AOD at 34 nm from Aeronet (Fig. S7a), and the visibility at 55 nm derived from the visibility meter (Fig. S7b): -24 September 212: clear sky with low aerosol -29 September 212: clear sky with high aerosol -2 July 212: broken clouds -9 September 212: continuous and thick clouds (the COT from MODIS is around 5) -28 July 212: cloud holes -1 June 212: fog Fig. S8 depicts the time series of the quantities described in Table 1 for the six selected days. From visible inspection of the different quantities for the different sky conditions on the selected days we determined the respective threshold values, which are listed in Table 1. In Fig. S9 exemplary results of the cloud classification scheme are presented for the selected days (including also the days shown in Figs. 7 and 8 of the main manuscript). Overall the deduced sky conditions are consistent with those identified by the MODIS images, and the measurements from AERONET and the visibility meter. 24 September 212 29 September 212 (clear sky with low aerosol) (clear sky with high aerosol)
Normalized Radiance AOD at 34nm Visibility at 55 nm [km] 2 July 212 9 September 212 (broken clouds) (continuous and thick clouds) 28 July 212 1 June 212 (cloud holes) (fog) Fig. S6 Visual images from MODIS (left Terra; right: Aqua) for the six selected days with different sky conditions near Wuxi. (a) 2.5 2. "clear with low aerosol" on 29 Sep 212 "clear with high aerosol" on 24 Sep 212 "cloud holes" on 28 July 212 "broken clouds" on 2 July 212 4 3 (b) "broken clouds" on 2 July "cloud holes" on 28 July "fog" on 1 June 1. 2.5 1 Fig. S7 (a) Time series of the AOD at 34 nm from AERONET on the days with clear sky, cloud holes or broken clouds (AOD data are not available for the other days because of clouds); (b) Time series of the visibility at 55 nm from the visibility meter on the days with broken clouds, cloud holes and fog (visibility data are not available for the other days due to instrumental problems). a) 29 September 212, "clear sky with low aerosol" Normalized CI z.9.6 Normalized TSI z 2.4x1-7 1.6x1-7 8.x1-8 Normalized TSI L 7.5x1-7 5.x1-7 2.5x1-7 Spread CI.8.4 2. 1..5 Spread O4 2 1
Normalized Radiance Normalized Radiance b) 24 September 212, "clear sky with high aerosol" Normalized CI z.9.6 Normalized TSI z 2.4x1-7 1.6x1-7 8.x1-8 Normalized TSI L 7.5x1-7 5.x1-7 2.5x1-7 Spread CI.8.4 2.5 2. 1..5 Spread O4 2 1 Normalized CI z.9.6 c) 2 July 212, "broken clouds" Normalized TSI z 2.x1-6 x1-6 1.x1-6 5.x1-7 Normalized TSI L 2.x1-6 x1-6 1.x1-6 5.x1-7 Spread CI.8.4 2.5 2. 1..5 Spread O4 2 1
Normalized Radiance Normalized Radiance Normalized CI z.9.6 d) 9 September 212, "continuous clouds" Normalized TSI z 2.4x1-7 1.6x1-7 8.x1-8 Normalized TSI L 7.5x1-7 5.x1-7 2.5x1-7 Spread CI.8.4 2. 2 1..5 Spread O4 1 e) 28 July 212, "cloud holes" Normalized CI z.9.6 Normalized TSI z 4x1-7 3x1-7 2x1-7 1x1-7 Normalized TSI L 1.x1-6 7.5x1-7 5.x1-7 2.5x1-7 Spread CI.8.4 2. 2 1..5 Spread O4 1
Normalized Radiance Normalized CI z 1..5 f) 1 June 212, "fog" Normalized TSI z 2.4x1-7 1.6x1-7 8.x1-8 Normalized TSI L 7.5x1-7 5.x1-7 2.5x1-7 Spread CI.8.4 2. 1..5 Spread O4.8.4 Fig. S8 Time series of the normalized Cl z, normalized TSI z, normalized TSI L, spread Cl, normalized radiance and spread O4 on the six selected days with the typical different sky conditions. thick cloud fog primary conditions clear with low aerosol extremely high midday CI clear with high aerosol zenith cloud holes off zenith cloud holes broken clouds continuous clouds (large SZA) continuous clouds (small SZA) fog optically thick clouds Sep. 29, 212 typical clear with low aerosol Sep. 24, 212 typical clear with high aerosol thick cloud fog primary conditions Jul. 2, 212 typical broken clouds Sep. 9, 212 typical continuous clouds thick cloud fog primary conditions Jul. 28, 212 typical cloud holes Jun. 1, 212 typical fog thick cloud fog primary conditions Jul. 8, 212 typical continuous clouds (small SZA) Jul. 29, 212 typical extremely high midday CI Fig. S9 Results of the cloud classification scheme for the selected days. Each symbol represents an individual elevation sequence. In the lower part of the indiviual figures, the primary sky conditions are indicated. In both upper parts the secondary sky conditions optically thick clouds and fog are indicated.
References Bogumil, K., Orphal, J., Homann, T., Voigt, S., Spietz, P., Fleis-chmann, O. C., Vogel, A., Hart-mann, M., Kromminga, H., Bovensmann, H., Frerick, J., and Burrows, J. P.: Measurements of molecular absorption spectra with the SCIAMACHY pre-flight model: instrument characterization and reference data for atmospheric remote-sensing in the 23 238 nm region, J. Photoch. Photobio. A, 157, 167 184, 23. Fayt, C. and van Roozendael, M.: WinDOAS 2.1 Software User Manual, IASB/BIRA available at: http://www.oma.be/gome/gomebro/windoas-sum-21b.pdf(last access: 2 May 214), 29. Grainger, J. F. and Ring, J.: Anomalous Fraunhofer line profiles, Nature, 193, 762, doi:1.138/193762a, 1962. Greenblatt, G. D., Orlando, J. J., Burkholder, J. B., and Rav-ishankara, A. R.: Absorption measurements of oxygen be-tween 33 and 114 nm, J. Geophys. Res., 95, 18577 18582, 199. Hermans, C., Pinardi, G., and Van Roozendael, M.: The simultane-ous retrieval of tropospheric aerosol extinction and NO2 vertical profiles from MAXDOAS measurements in Beijing, in: Proceed-ings of the 8th International Symposium on Tropospheric Pro-filing, ISBN: 978-9-696-233-2, Delft, the Netherlands, 19 23 October 29. Kraus, S.: DOASIS A Framework Design for DOAS, Ph.D. thesis, Combined Faculties for Mathematics and for Computer Science, University of Mannheim, Germany, 25. Vandaele, A. C., Hermans, C., Simon, P. C., Carleer, M., Colin, R., Fally, S., Mérienne, M.-F., Jenouvrier, A., and Coquart, B.: Mea-surements of the NO2 absorption cross section from 42 cm -1 to 1 cm -1 (238 1 nm) at 22 K and 294 K, J. Quant. Spectrosc. Ra., 59, 171 184, 1998. Wang, P. and Stammes, P.: Evaluation of SCIAMACHY Oxy-gen A band cloud heights using Cloudnet measurements, At-mos. Meas. Tech., 7, 1331 135, doi:1.5194/amt-7-1331-214, 214.