Estimating the volcanic emission rate and atmospheric lifetime of SO 2 from space : a case study for Kı̄lauea volcano , Hawai ‘ i

Reviewer comment: 1) I think the paper is weak in the references. This is especially true for the introduction section which does not give a good overview of past studies on the subject. Instead, the authors put forward the “new potential” of satellite sensors for monitoring volcanic activity. Estimation of SO2 fluxes and lifetime from space have been reported in several papers, recently but also in early studies using TOMS data. Moreover, when referring to satellite measurements, the authors should also mention SCIAMACHY, OMI, GOME-2, IASI, AIRS (among others) in addition to the GOME-1 sensor. In the first two paragraphs of the introduction (on the role of SO2 on the atmosphere and its lifetime), it would be good to have references to key papers.

complete transects does not pose a problem.However, due to the close proximity to the vent, the encountered SO 2 column densities and aerosol optical depths are orders of magnitude higher than those e.g.measured by satellite tens to hundreds of kilometers farther downwind, and perhaps the more interesting question is whether or not the core of the plume is adequately sampled by the ground-based traverses.Due to the high optical thickness of these plumes at close proximity, light is often prevented from penetrating the plume core.Instead, light paths that only penetrate the plume edge before being scattered towards the instrument become more likely, therefore skewing measurements towards lower column densities (Kern et al., 2012).These non-linear radiative transfer effects are now discussed in more detail in the text (Section 4.4).
Reviewer comment: Section 4 gives a description of the uncertainties of the emission rate and lifetime estimates but the latter uncertainties only relate to the method used and not to the SO2 column retrievals (see next point).

Authors' reply:
We agree that the discussion of uncertainties misses the important aspects of SCDs and VCDs.We have thus extended the discussion of uncertainties considerably.In new subsections (4.2.1 and 4.2.2),we now deal with the uncertainties of SO 2 SCDs and VCDs.
Reviewer comment: 3)In view of the above, the description of the SO2 retrieval (section 2.1) is insufficient.For an SO2 plume at a presumable altitude of 1.5-2.5 km, many parameters can influence the retrieval which are not even discussed here.In reality, the results of the SO2 algorithm have not been validated and it is hard to know what it is the accuracy of the retrieved slant columns.

Authors' reply:
We have expanded the description of the SO 2 retrieval, now providing the relevant information for the DOAS set-up and the calculation of AMFs, and expanded the discussion of uncertainties accordingly (see above).In order to check the consistency of our GOME-2 SO 2 data product to other satellite data products, we also investigated the operational NASA OMI SO 2 Level 2 product for August 2008 exemplarily.Therefore, the results from the Planet Boundary Layer product (PBL-SO2, assuming a mean altitude of 0.9 km) and the Lower Tropospheric Column product (TRL-SO2, assuming a mean altitude of 2.5 km) were interpolated to an altitude of 2 km.The resulting SO 2 fluxes as function of time are very similar compared to our results from GOME-2:  The derived lifetime based on OMI is longer by 33%, as in the OMI monthly mean the downwind VCDs are slightly higher than for GOME-2.The estimated emission rates, however, agree within 11%.The deviations are thus smaller than the total error of about 40% estimated in the revised manuscript.We conclude that our GOME-2 retrieval is in good agreement with the operational OMI product, derived by a completely independent retrieval algorithm, and that monthly means are meaningful for this approach, i.e. sampling effects are negligible.
wavelength dependence of the AMF is strong especially for a low plume (as it is the case here).The surface albedo used is not given, although it is arguably a large source of uncertainty on the SO2 column retrieval.
Authors' reply: We agree that our description of the method was too scarce.The extended method section now includes the respective details on the chosen wavelength (315 nm) and albedo (5%), and the discussion section now also investigates the impact of a-priori wavelength and surface albedo on the resulting AMFs.
Reviewer comment: GOME-2 is also known to suffer from several limitations at the edge of band 2 and this is not developed in the text.
Authors' reply: Due to some restrictions concerning the correction for Etalon structures of GOME-2 spectra, the lower edge of the GOME-2 band 2A (<316.5 nm) is officially not supported as "valid spectral region" (pers.communication with Rüdiger Lang, EUMETSAT, Darmstadt).The Etalon correction has been expanded to lower wavelength only since June 13, 2013, but has so far not been confirmed to significantly improve this spectral region for Level 2 data products.In order to investigate the possible influence of Etalon structures on our SO 2 retrieval, we compared the fit residues and SO 2 SCDs over a remote area over the Pacific (145 • -180 • W, +-15 • N) for the time period 2 weeks before and after the Etalon correction has been expanded to the lower UV.Although the overall fit quality was improved (as indicated by a decrease of the fit residues), the influence on the mean SO 2 SCDs as well as the corresponding standard deviation was found to be negligible.Additionally, data for moderately enhanced SO 2 SCDs during increased activity of Nyiamuragira volcano in June/July 2013 had been analysed with and without expansion of the Etalon correction to wavelengths <316.5 nm (by courtesy of Rüdiger Lang).Like for the Pacific area, an overall improvement of the fit could be observed.However, the corresponding SO 2 SCDs were found to be within ≈ 1% of the uncorrected results.As this is a quite complex technical detail with almost no impact on our results, we do not include this discussion in the manuscript.

Reviewer comment:
GOME-2 has undergone severe degradation since its launch and it is not clear whether the elevated background values of a few kT/day observed from 2009 onwards (Figure 7) are real or not.
Authors' reply: As mentioned in Section 2.1, the detection limit for SO 2 has doubled from 2007 to 2011, and fit residues gradually increase over time.However, due to averaging (monthly means) and spatial integration (line densities), statistical errors are largely reduced (compare the vertical error bars of Fig. 4).Possible systematic effects of degradation on the SO 2 retrieval, on the other hand, are eliminated by our background correction east of Hawai'i.We thus conclude that the observed peak of SO 2 in 2008 and the enhanced values in 2009-2012 (compared to 2007) are real (in terms of SCDs).A qualitative comparison of our GOME-2 results to operational SO 2 data products for the OMI and SCIAMACHY instruments (available e.g. at http://so2.gsfc.nasa.gov/and http://sacs.aeronomie.be/),which do not show that strong degradation effects, also reveal clearly increased SO 2 over Hawaii for 2008 and onwards, compared to 2007.

Reviewer comment:
It is anyway in clear contradiction with the ground-based data.The authors shall expand the data description and include an error analysis to confirm (or not) if their space-based emission rates estimates are larger than the ground-based values.
Authors' reply: We have substantially expanded the retrieval description and discussion of uncertainties for both our satellite retrieval and the ground-based measurements, and substantiated our arguments why we think the satellite measurements are more accurate for this specific case, at least in 2008.
Reviewer comment: -"..spatial fluctuations, and total emissions are still highly uncertain": please provide a reference.

Fig
Fig. R1: SO 2 flux as function of time (crosses) and the fitted model function (smooth line) for GOME-2 (red) and OMI (green) in August 2008.