Interactive comment on “ Water Vapour and Methane Coupling in the Stratosphere observed with SCIAMACHY Solar Occultation Measurements

period 2003-2012, and the results shown here are scientifically valuable. However, in much of the text the authors seem to be trying very hard to create a mystery where there is none. There is (1) a QBO signature in H2O crossing the tropical tropopause and (2) a QBO signal due to changes in transport (age-of-air) which causes a variation in the amount of CH4 that has been oxidized to produce H2O. The authors repeatedly overemphasize the importance of small tropospheric CH4 variations on the observed variations in stratospheric H2O. While gradually increasing anthropogenic CH4 is a very important driver of long-term change in H2O, variations in CH4 entering the stratosphere are only marginally relevant to the variations observed in these measurements, which span a decade. Small changes in tropopause temperature are a far more important driver of interannual changes in H2O entering the stratosphere as has been shown by many authors (e.g. Dessler et al., JGR 2014).

e.g. Buchwitz et al., 2017, and references therein). Due to its long lifetime, tropospheric methane is then transported into the stratosphere.
Most of the water vapour is of natural origin and located in the troposphere. It enters the stratosphere mainly through the tropical tropopause layer (TTL; see e.g. Randel et al., 2004, Randel andJensen, 2013, andreferences therein). There, the cold temperatures of the tropical tropopause yield a 'cold trap' (see e.g. Read et al., 2004, Holton andGettelman, 2001) 5 causing lower concentrations of water vapour in the stratosphere than in the troposphere. The water vapour, which enters the stratosphere through the TTL, is then transported via the Brewer-Dobson circulation from the tropics to higher latitudes.
The Brewer-Dobson circulation also controls the tropical upwelling and freeze-drying process that in turn determines the stratospheric entry of water vapour in the tropics (Randel et al., 2006;Dhomse et al., 2008).
In the middle stratosphere and above, water vapour is in fact mainly produced from oxidation of stratospheric methane via 10 the reaction Via various photochemical processes (see e.g. le Texier et al., 1988) the CH 3 is converted first to HCHO and then to H 2 O resulting in the net reaction: CH 4 + 2O 2 → 2H 2 O + CO 2 (R2) methane product, Rong et al. (2016) presented results from a combination of SOFIE and MIPAS methane with water vapour profiles from the Aura Microwave Limb Sounder (MLS; Waters et al., 2006) on Aura.
The SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY (SCIAMACHY; Bovensmann et al., 1999;Gottwald and Bovensmann, 2011) on ENVISAT performed measurements in various viewing geometries over a large spectral range from the UV to the SWIR. Among these are solar occultation measurements, which cover -depending on season -5 the spatial region between about 50 • N and 70 • N. Noël et al. (2016) presented an updated data set for stratospheric methane derived from SCIAMACHY solar occultation using the onion-peeling DOAS (ONPD) method. Already some years ago, Noël et al. (2010) showed first retrieval results for stratospheric water vapour profiles from SCIAMACHY which were based on a similar algorithm. Recently, the improved method used by Noël et al. (2016) has also been applied to water vapour, resulting in a consistent set of SCIAMACHY stratospheric water vapour and methane data.

10
In this manuscript, we shortly describe the updated water vapour algorithm in section 2. We then present the new water vapour results in section 3, which also includes a first validation by comparison with independent data sets and a combination of the new water vapour data with the methane data from Noël et al. (2016). The results are discussed in section 4. The conclusions are then presented in section 5.
The retrieval method used in this study is essentially the same as described in Noël et al. (2016), therefore only the principle idea is explained here.
We use transmission spectra as function of viewing (tangent) altitude derived from SCIAMACHY solar occultation measurements. For the water vapour retrieval, we take data in the spectral range 928 nm to 968 nm. The ONPD retrieval is then based on a combination of a weighting function DOAS fit (see e.g. Perner and Platt, 1979;Burrows et al., 1999;Coldewey-Egbers et al., 20 2005) with a classical onion peeling method (see e.g. Russell and Drayson, 1972). The retrieval altitude grid is 0 to 50 km in 1 km steps. The measured spectra are interpolated to this grid. The analysis starts at the top level and then proceeds downwards, taking into account the results from the upper levels. At each level, we determine the water vapour density from the difference between the measured transmission and a modelled one. This is done by fitting to the data a set of factors describing the change of an atmospheric parameter in combination with corresponding weighting functions. Such a weighting function describes the 25 change of the spectrum when changing a certain parameter, e.g. the water vapour concentration at this altitude. In the present case we consider in addition to water vapour also changes in ozone (which also absorbs in the spectral window used). Actual pressure and temperature profiles have been taken from ECMWF ERA Interim data (Dee et al., 2011). The related weighting functions have been determined from radiative transfer calculations using the SCIATRAN model .
To account for spectrally broadband effects resulting from e.g. aerosols we also fit a polynomial to the spectra. A possible 30 misalignment of the wavelength axis of the measured data is considered by fitting corresponding shift and squeeze parameters.
An example for the results of the fitting procedure is shown in Fig. 1  Bottom: resulting residual, i.e. relative difference between measurement and fit.
After the retrieval several additional corrections are performed as described in Noël et al. (2016): -The retrieved profiles are smoothed with a 4.3 km boxcar to account for the vertical resolution of the measurements and to reduce oscillations in the retrieved number densities.
-Additional correction factors are applied for non-linearity and saturation effects (due to the limited spectral resolution of the measurements).

5
-The resulting errors are multiplied by a factor of 0.66 to correct for correlations between different layers not considered in the fit (see Noël et al., 2016, for details).
The resulting number density profiles are converted to volume mixing ratios (VMRs) using ECMWF pressure and temperature. The useful vertical range of the SCIAMACHY ONPD data is currently considered to be 17 to 45 km, mainly limited by noise and numerical effects at the upper altitudes and by tropospheric effects (e.g. clouds and increased refraction)) at the 10 lower altitudes.  denote the errors given in the products. Obviously, the new SCIAMACHY product is closer to the ACE-FTS results and the reported error is largely reduced compared to the older version. This is due to the improved retrieval method and to the updated calculation of errors as described in Noël et al. (2016).

H 2 O validation
A large number of water vapour data products have contributed to the second SPARC (Stratosphere-troposphere Processes And 5 their Role in Climate) water vapour assessment (WAVAS-II; see e.g. Lossow et al., 2017, further publications in preparation).
One activity of WAVAS-II was the inter-comparison of the different data sets, including a preliminary earlier version (V4.2.1) of the SCIAMACHY ONPD product. The performance of the V4.2.1 product is very similar to the V4.5.2 product described in this manuscript. We therefore show in this section only two comparisons with collocated ACE-FTS (see e.g. Nassar et al., 2005) and MLS (see e.g. Carr et al., 1995;Lambert et al., 2007) data as an example. In both cases the spatial collocation criterium   Fig. 4. The SCIAMACHY water vapour profiles agree with both data sets within less than 5%. The SCIAMACHY water vapour VMRs are usually slightly higher than those of ACE-FTS, but (except for the lowest altitudes) typically smaller than MLS VMRs. A small vertical oscillation of 1-2% amplitude is visible in the differences; this is caused by the SCIAMACHY data and probably a retrieval artifact which was also seen in the SCIAMACHY ONPD methane and CO 2 data . The observed deviations are within the typical error of the products.

Time series
The ONPD algorithm for water vapour has been applied to the whole set of SCIAMACHY measurements from have therefore a specific temporal and spatial sampling.
The SCIAMACHY water vapour profiles perform in general as expected: Highest VMRs (up to about 8 ppmv) occur at upper altitudes, lowest VMRs at lower altitudes. The variation with time follows roughly the tropopause / latitude pattern.

10
For a more detailed analysis including the combination of water vapour and methane results, we computed monthly anomalies from the SCIAMACHY H 2 O data in the same way as described Noël et al. (2016) and put them in relation to the CH 4 data from this study. This is done by first averaging the daily data over the months and then subtracting the long-term average  we use only years for which data for all months are available.
In Fig. 6 the time series of the H 2 O and CH 4 anomalies are shown. There is a clear bi-annual structure visible in both data sets with opposite sign. As already mentioned in Noël et al. (2016), this structure is related to the Quasi-Biannual-Oscillation (QBO), see e.g. Baldwin et al. (2001). be identical. This is in fact nearly the case for altitudes above about 25 km, where the water vapour variations follow quite well the methane variation. At 17 km, however, the methane anomaly does not vary much whereas the water vapour anomaly still shows a clear QBO signature, which is shifted in phase with respect to 25 km.
The downward peak in the water vapour anomalies in the middle of 2009 is related to the eruption of the Sarychev volcano on 12 June 2009 which reached these altitudes (Jégou et al., 2013). Note that this observed reduction of water vapour after 5 the Sarychev eruption may be introduced by the remaining sensitivity of the retrieval method to aerosol. In the retrieval only spectrally broadband contributions of aerosols are considered, but there are also (second order) effects e.g. caused by the vertical integration of the signal over the field of view, which may play a role in case of large aerosol concentrations. This issue is still under investigation.
The relation to QBO is illustrated in Fig. 8  about the same altitude), which is commonly used as proxy for the QBO (see e.g. Gebhardt et al., 2014). The Singapore wind data have been provided by Freie Universität Berlin (2014). Negative wind direction corresponds to Easterly winds (marked blue in Fig. 8), positive direction to Westerly winds (marked red). Water vapour negative and positive anomalies are also plotted in blue and red, respectively. For the methane plot, the vertical axis and colouring has been inverted, because an increase in water vapour should correspond to a reduction of methane according to (R2). Potential Water VMR Anomaly (ppmv) Figure 9. Potential water anomalies derived from combination of SCIAMACHY H2O and CH4 anomalies (Fig. 6).

Potential water
To further investigate the temporal variabilities a time series has been derived by adding to the water vapour VMR anomalies two times the methane anomalies. As mentioned above this combination, referred to as potential water (Nassar et al., 2005), is assumed to be conserved if water vapour is solely produced from methane oxidation, and temporal variations of this quantity can be related to changes in transport or additional sources and sinks. The result is displayed in Fig. 9.

Trends
A linear trend model has been fitted to the water vapour anomalies at each altitude similar to what has been applied in the earlier methane study, see Noël et al. (2016). The trend profiles are displayed in Fig. 10. water vapour during the time interval considered in this study has been observed by Urban et al. (2014) and Weigel et al. (2016) mainly in the tropics. As already discussed in Noël et al. (2016) methane trends are also not significant except for the lowest altitudes, where they are in general agreement with tropospheric trends. However, it should be noted that errors of the data and 10 autocorrelation of noise have not been considered in the trend fits, which might affect the trend errors.
The potential water vapour trend is the sum of the water vapour trend and two times the methane trend. This is an estimate an estimate for water vapour changes or methane changes not related to the stratospheric production of water vapour by methane. If potential water is conserved, this trend should be zero. The potential water trend profile is shown in the right plot of Fig. 10. The error of the potential water trend has been derived via propagation of the errors of the methane and water vapour 15 trends. Considering this error, the combined trend above about 20 km is in a statistical sense not significant, meaning that the assumption that all water vapour is produced from methane via the net reaction (R2) is not disproved by the measurements.
This is especially the case between 25 and 30 km where the trend itself is close to zero. At the lower altitudes, a significant deviation of the potential water trend from zero is observed (up to about 0.02 ppmv year −1 ).

20
The findings of the previous section can be summarised as follows:  Figure 11. Simplified schematic view of transport pathways within the Brewer-Dobson circulation.
-Water vapour and methane time series and trends look different above and below about 20 km.
-In the upper altitudes both water vapour and methane time series show a pronounced QBO signature.
-In the lower stratosphere, QBO signature is only visible in the water vapour data.
-There is a phase shift in the water vapour QBO signal between upper and lower altitudes.
-Potential water, the combination of methane and water vapour VMRs, is essentially conserved at upper altitudes except 5 for some short-term events and a longer-term variation with a period of about 5-6 years.
-The QBO signal is also visible in the potential water data at lower altitudes until about 2009/2010; after that potential water increases slowly.
These observations can be explained by a separation of the stratosphere into two vertical regimes. The lower region is mainly affected by the shallow (or lower) branch of the Brewer-Dobson circulation (see e.g. Butchart, 2014, and references According to the data of the present study, this separation occurs at about 20 km; however it has to be kept in mind that this is an approximated value and that the vertical resolution of the SCIAMACHY solar occultation data is about 4 km. In the lower region, variability is determined by water vapour variations due to QBO effects on tropopause temperature and/or stratospheric transport and due to tropospheric methane variations; above, water vapour is mainly produced from methane oxidation and potential water anomalies are more homogeneous with altitude and change on longer time scales. Water vapour and methane below 20 km are therefore dominated by the variations imprinted on them from their tropospheric sources especially during their vertical transport into the stratosphere at tropical regions. The amount of water vapour entering 5 the tropical stratosphere is related to the tropopause temperature which varies with QBO, see e.g. Fueglistaler and Haynes (2005). This is not the case for methane, which could explain the missing QBO signature in the methane time series at 17 km (Fig. 7).
The missing balance between methane and water vapour at lower altitudes is in fact not surprising, because the photochemical processes involved in the conversion of methane to water vapour are less effective there since less UV radiation reaches 10 these altitudes (le Texier et al., 1988). Furthermore, since the transport via the shallow branch is comparably fast (less than about one year from the entry point in the tropics to mid-latitudes, see Birner and Bönisch, 2011) the balance between water vapour and methane is also not reached in the extratropical lowermost stratosphere. This could explain the phase shift in the water vapour QBO signal between 25 and 17 km (Fig. 7) and is in line with measurements of age of air by e.g. Haenel et al. (2015) which show that the air at 17 km is younger than the air above.  Haenel et al., 2015), this could explain the increase of potential water at lower altitudes after 2009/2010 shown in Fig. 9. Until end of 2011 the positive potential water anomaly extends to higher altitudes. This is in agreement with the increasing age of air at higher altitudes. 20 However, from the current data set an additional influence of varying tropospheric water vapour input on the observed increase of potential water cannot be ruled out.
Above 20 km, in the region of the deep branch of the Brewer-Dobson circulation, air is older such that the conversion process from methane to water vapour has reached an equilibrium, variations of both gases are in phase and potential water is essentially conserved (Fig. 7). A remaining open issue is the QBO signal observed in both methane and water vapour at higher 25 stratospheric altitudes. The conservation of potential water indicates that at these altitudes water vapour changes are mainly related to changes of methane. Therefore the QBO signal has to be carried by methane, but as can be seen at lower altitudes the methane entering the stratosphere is not varied by QBO. The QBO signature in the upper altitude data can be explained by a QBO-related modulation of the transport to higher latitudes via the deep branch of the Brewer-Dobson circulation, similar to the variation in tropical aerosol extinction coefficients as seen by Brinkhoff et al. (2015) at 30 km. Also Randel et al. (1998) 30 observed a QBO signal in tropical methane from HALOE measurements on UARS above about 35 km but not below, correlated with the residual mean wind circulation. This is also in line with results from e.g. Niwano et al. (2003)

Conclusions
A new stratospheric water vapour data set based on SCIAMACHY solar occultation measurements is available. It covers the latitude range between about 50 and 70 • N and the altitude range from 17 to 45 km. It has been generated in a similar way as the corresponding methane product   Variations in water vapour are clearly correlated with those of methane. A QBO signature is visible in both water vapour and methane anomaly time series, showing that transport from the tropics affects essentially the whole altitude range under investigation in this study.

10
The analysis of the combined water vapour and methane data sets reveals, that potential water, the sum of water vapour VMR and two time methane VMR, seems to be overall conserved between about 20 and 40-45 km. However, potential water is not constant over time. In addition to short term fluctuations a variation on a timescale of 5-6 years is observed, which needs further investigation.
At altitudes below about 20 km the QBO signature is only visible in water vapour but not in methane data. As a consequence, 15 potential water also shows a significant QBO variation, but also a continuous increase after about 2009.