|Comments for the revised manuscript: “Intercomparison of MAX-DOAS vertical profile retrieval algorithms: studies on field data from the CINDI-2 campaign.” by Tirpitz et al.|
I have read the responses from my initial comments and check the revised manuscript. I appreciate that the author(s) reply to all my comments. With that being said, I still have minor (maybe major) comments below:
- It is very concerning that differences of individual participants vary by a factor of 10 for VCDs, AOT and surface concentration, considering that identical dSCDs are used and key retrieval parameters are prescribed. For the comparison against supporting/independent observations the differences are even larger. Note that if participants used their own dSCDs discrepancies degrade even more. Typically, MAX-DOAS (and other ground-based remote sensing instruments) are viewed as key measurements for validation of satellite products, however, after reading the revised manuscript it is not clear what to make of this work, again from a validation point of view. The work presented here is important and findings are key as well, but I highly suggest making the point of what is needed to improve these differences.
- At the end of the abstract and in the manuscript it is mentioned that for OE the underestimation of AOT compared to sun photometer is due to lack of sensitivity to higher altitudes and associated to smoothing effects and a priori assumptions. On my first review I recommended to use the ceilometer extinction profiles as a priori and compare retrievals with the exponentially decreasing profile to further improve this statement along with drawing better findings regarding the scaling factor (and PAC). However, authors decided not to perform and replied with the following two arguments:
1. The paper aims at the comparison and validation of MAX-DOAS profiles retrieved under typical measurement conditions. This includes using prior information as they are typically available for an arbitrary measurement location and season. Having daily radiosondes, ceilometer data and collocated sun photometer measurements at hand is not a very usual scenario. In fact, most MAX-DOAS studies have to resort to climatologies for their prior assumptions.
2. Since the MAX-DOAS results are validated by the supported observations (at least qualitatively, in the case of the ceilometer profiles), they need to be kept independent, which is not the case if one observation serves as a priori for the other.
Regarding 1, what is the point of having all supporting measurements if they are not used to improve retrievals and reduce errors in the forward models?, especially because also the aim of the study is to further compare with colocated supporting observations.
Regarding 2, I am not sure if I follow this, my understanding is that ceilometer observations are not used quantitatively, why do you need to keep them independent?
Furthermore, I asked if by using a ceilometer as a priori the sensitivity increases at higher layers? Do AKs change?
Author reply: This depends on the a priori covariance. Since the uncertainty of ceilometer data is surely smaller than that of an exponential profile, the sensitivity and DOFs will decrease.
I do not agree with this. Uncertainty of ceilometer data is not the same as the a priori covariance. A priori covariance should cover a range of expected variability.
To me it is not clear how authors draw this conclusion (written in the abstract): “for optimal estimation algorithms this can be largely explained and compensated by considering smoothing effects, namely biases arising from the reduced sensitivity of MAX-DOAS observations to higher altitudes and associated a priori assumptions”.
For my suggestion, participants do not need to repeat retrieval, but likely only one group performing OE can test the aerosol extinction profile obtained with the ceilometer and make an educated guess, probably based on the variability during the campaign, for the covariance matrix, is the PAC correction factor still needed?.
Authors included this:
“[…]even though the motivation for the application of the PAC and the SF are different: the application of the PAC is necessary solely for mathematical reasons related to the concept of OEM and prior constraints applied therein. In contrast, publications that suggest or discuss the application of an SF (e.g. Wagner et al., 2009; Clémer et al., 2010; Ortega et al., 2016; Wagner et al., 2019).....They do not make use of optimal estimation or prior constraints similar to those used in our study. Thus their findings can be considered independent from any kind of PAC.”
I might be wrong but I recall that Clemer et al. (2010) used OE in the retrieval. Note also that Ortega et al. (2016) suggested that aerosol extinction aloft might be important, which is related with your statement about the lack of sensitivity aloft due to assumptions in the a priori but it is not reflected or mentioned in the manuscript.