|Review of “An ensemble Kaman filter data assimilation system for the whole neutral atmosphere” by Koshin et al.|
The authors have taken the previous reviews into consideration, and revised the manuscript accordingly. In general, I find that their revisions are satisfactory. However, I have a few remaining comments that the authors should address prior to publication.
1. I still do not think that the authors should refer to the Ctrl parameters as “optimal”. This implies that every possible combination of parameters were tested, and these values give the best results. However, only a subset of the near-infinite number of parameter settings were tested. The authors do not know if the results would be marginally improved by, for example, changing the localization length from 600 to 625 km. Because it is not possible to determine the truly optimal parameters, the Ctrl case should really be considered to be the best among what was tested. I thus still am of the opinion that the authors should not refer to these as the optimal parameters.
2. Lines 277-279: The 30 degree spacing is between observations on subsequent orbits. This is different then cross track observations, which would be observations along the same orbit. This statement should be revised accordingly.
3. Line 539: “collection” should be “correction”
4. Lines 538-545: The details about the bias corrections used in previous studies is not entirely correct. For example, Eckermann et al. (2009) state “Here we compute a similar profile over the May–June 2007 period using version 1.07 SABER data, but use it to bias-correct SABER temperatures only at altitudes below 2.7 hPa, where SABER has a 1–3 K warm bias relative to MLS and other instruments (Schwartz et al., 2008; Remsberg et al., 2008). At higher altitudes, V1.07 SABER temperatures appear to be accurate to within (Remsberg et al., 2008) whereas V2.2 MLS temperatures have a vertically structured bias near the stratopause and a systematic cold bias throughout the mesosphere (Schwartz et al., 2008). Thus, at altitudes above we bias-correct MLS temperatures using a profile based on a subjective fit to the various profiles of mean bias of MLS relative to other satellite, suborbital and analysis temperatures plotted in Fig. 26 of Schwartz et al. (2008).” Additionally, Pedatella et al. (2016, doi:10.1002/2016JA022528) indicates that a bias correction has been applied in WACCM+DART. McCormack et al. (2017) does not state explicitly whether or not a bias correction is applied, so it is unclear whether or not a bias correction was used.