The Potential of Satellite Sounding Observations for Deriving Atmospheric Wind in All-Weather Conditions
Round 1
Reviewer 1 Report
Review of the paper
The Potential of Satellite Microwave Sounding Observations for Deriving Atmospheric Wind in All-Weather Conditions
By: Yijia Zhang , Hao Hu, and Fuzhong Weng
General comment
The paper to obtain pixel-wise 3D wind fields by tracking features of 3D water vapor fields using the Farneback OF method.
It is shown that the wind accuracy from tracking the specific humidity is higher than that from tracking the relative humidity. For fast-evolving weather systems, and in the paper is considered the Lekima typhoon, the shorter time step allows for more accurate wind retrievals, whereas for slow to moderate evolving weather conditions, the longer time step is needed for smaller retrieval errors.
The topic of the paper is interesting and well fit the scope of the journal. The main problem with this paper is that it is too cryptic in some sections. The logical passes and the metrics used to compute errors are difficult to understand for people not working in the field.
Major point
The description of the methodology (Material and methods, Section 2.1) and of sensitivity tests (Section 3.1) are too cryptic and not understandable to non-expert reader. The error metrics must be described in more detail, including logical passages, to understand which computations have been done (for reader to replicate the methodhology).
In addition, how was the resolution degraded from 3 km to coarser resolutions. Taking a grid point every few grid points? Explain. How was the time resolution changed? Saving WRF output every 5 minutes and calculating the differences at certain time intervals? Explain.
The initial part of the Section 2.1 start describing WRF simulation using a language that is not common to atmospheric modelers. I suggested changes to be more understandable to the meteorological community. If not changed, a motivation should be provided for the choice at the start of the section.
Section 3.4 is hardly understandable in the current form. Again, more details should be provided to clearly show what was done, including logical passages.
Minor points
The minor points refer to typos errors, language and figure adjustments. They are reported in the in sticky notes attached to the pdf of the paper.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper presents a technique, Farneback Optical Flow, to be applied to water vapor fields to derive atmospheric motion vectors. This is a topic that is very interesting and will become useful for, mostly, the geostationary hyperspectral infrared sounders that are or will be in space. It can also be applied to microwave sounders in polar orbits in locations where the revisit time is high enough (near the poles). The paper is very well written and presented.
Unfortunately, the claim in the title and introduction that they are deriving winds from Microwave (MW) sounders, but no such simulation seems to exist in the paper. They seem to apply the technique directly to the WRF NWP model water vapor fields. This is clearly and ideal, and not very realistic simulation, since they know perfectly the level at which the water vapor is and the water vapor fields are very well defined within their level. In practice, a MW sounder will have quite a big uncertainty in the determination of the exact height of the water vapor (several hundreds of hPa) and the algorithm might not prove as useful. As such, the simulation they are doing is not at all realistic.
They should also stress more the fact that they are doing all the exercise with an NWP simulation with all the drawbacks that that implies. NWP models usually do not reproduce many features which are present in reality and that will complicate the determination of AMVs, such as, for example, gravity waves.
I would therefore suggests two improvements:
- Simulate the radiance of any, selected by the authors, MW instrument. This can be done with a radiative transfer model, such as RTTOV for example (https://nwp-saf.eumetsat.int/site/software/rttov/). After radiances are simulated, try to derive the AMVs from them. It will certainly prove more difficult. This will mean more work for the authors, but will make the paper more meaningful. They could also apply it, with very little extra effort to hyperspectral sounders such as IASI or MTG-IRS or GIIRS.
- Stress in the abstract and conclusions that this is a simulation with potential significant differences with the reality. Features such as gravity waves and others can have a big impact in the quality of the AMVs
I would therefore recommend to have a major revision of the paper as outlined above.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors answered to all my comments and suggestions satisfactorily . The paper an be published in the present form.
Author Response
Thank you for your very helpful advice before!
Author Response File: Author Response.docx
Reviewer 2 Report
The main conclusion of the paper is still based on tracking direct water vapor fields from the model. This is not realistic, since these water vapor fields will be derived in practice by a 1D retrieval from MW instruments. This was suggested in the first review and they have done this only partially.
Because of this, I think the paper is not very useful because it is not reproducing a realistic situation. I would strongly recommend to change:
- The title of the paper and remove Microwave from it, since it is just a study of tracking winds from WV fields from an NWP model.
- Also, they should stress in the abstract that this is study is made with simulated data and as such it will have many limitations in practice. Many practical details from RTM in the MW region, and particularly at 183 GHz are still unknown. See for example https://amt.copernicus.org/articles/9/2207/2016/ or https://amt.copernicus.org/preprints/amt-2018-181/amt-2018-181.pdf). Other effects, not properly reproduced by models could also come into play in a real situation. These sort of considerations should be stated as on sentence in the abstract and in the conclusions.
As such I can only recommend publishing the paper after major revisions
Author Response
Please see the attachment.
Author Response File: Author Response.docx
