Assessment of the Hyperspectral Infrared Atmospheric Sounder (HIRAS)

Round 1

Reviewer 1 Report

This study “Assessment of the Hyperspectral Infrared Atmospheric Sounder (HIRAS)” is a novel study that compares the two hyperspectral infrared sounder measurements: HIRAS and CrIS. Using radiative transfer model (RTTOV), the radiance derived from Met Off NWP short-term forcast datasets are compared with FSR and NSR datasets of HIRS, and FRS results between HIRAS and CrIS (double differening). Calibration/validation of a new sensor is important compared to legacy and well calibrated sensors.  This study compares the HIRAS and CrIS on three infrared bands, and characterizes the channel bias and standard deviation and links the possible large bias/standard deviation with the lack of proper specification of trace gas in the forecast model. Particularly, the author investigates the causes of large bias of detector 3 and attribute to the sun-glint contamination in the space view. This study is significant for preparing the HIRAS data for assimilation into different NWP model and being used as references for cross calibration of similar sensors. The logic of the analysis of the sun-light contamination of detector 3 is sound and convincible.  The contents of the paper fall well in the scope the remote sensing. I recommended this paper to be published on Remote Sensing with minor revision.

Comments:

For a new instrument, the validation of the geolocation is often accompanied with the radiometric calibration. Before the radiometric comparison/validation with CrIS, do the author know how good the HIRAS FOV geolocation is? Is there any publication for comparing the geolocation with on-board imager or other platform observations? How does this geolocation accuracy affect the bias and standard deviation in comparison? If authors select certain uniformity levels as screening creteria, will the bias and standard deviation reduce? From FSR to NSR, the apodization used in CrIS is a three point Hanning Window smoother. The authors used the same as CrIS, does this affect the standard deviation of NSR comparison? Line 19, in abstract, can you explain what this mean “ where confidence in NWP is the largest”? Line 39, first … hyperspectral infrared instrument in polar orbit. I don’t think there is another Chinese hyperspectral infrared sounder in other orbits (e.g. geostationary). Line 54, please also explain what’s the internal calibration target. The black body for warm view? Line 95, please make it clear that you use the Met Off NWP forecast results as your radiative transfer model input. Line 101, explain briefly why land and ice pixels are excluded Line 104, the link should be more specific to point to the coefficient webpage. Line 215, ‘detector 1-2’ should be ‘detectors 1-2’ Line 259, I understand the double difference can detect the relative bias using bias differencing, how is the standard deviation able to be differenced, even assuming the reference error free? Line 281, isn’t HIRAS NSR data derived fromFSR? How does the CMA cloud mask screening affect differently between FSR and NSR of HIRAS?’ Line 307, spell out ISEM at first referenced. Line 310, Are the waves, refractive indices and saliniry needed by ISEM? Line 311, spell out IREMIS Line 421, Spell out and use consistenly later (MERSI-2) Line 429, either remove +/- sign, or use for all numbers as range. Line 430, does it make more sense using “no more than” instead of ‘ less or equal to’. Figures are generally fine, for Figures 2 and 3, move legends inside the box.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors presented an initial study of the accuracy of the hyperspectral IR instrument HIRAS by comparing with UK-MET forecasts and CrIS brightness temperatures using a double difference method. 

This paper is well written and clear in it's presentation, but several major points need to be addressed. These are:

 

1.  It is not clear what the orbit HIRAS is, and whether there are any observations that can be collocated with CrIS.   The comparisons with CrIS that were done don't indicate whether these are collocations. If they aren't, a comparison using just collocated observations should be done. 

 

2.  There is no point to compare HIRAS FSR with CrIS NSR.  These comparisons are not meaningful.  NSR lacks the structure of FSR, so it will always be different.  The standard product for CrIS is NSR, but there is an FSR product.  I would recommend either obtaining the FSR product from CrIS and comparing with that, or drop the FSR to NSR comparison altogether. 

 

3. The plots of the double differences also don't tell us the relative accuracy of HIRAS compared to CrIS.  It would really be helpful to see (O-F) for HIRAS and (O-F) for CrIS in the same plot. I realize that CrIS is assimilated so that it will be closer to the forecasts, but this still tells us whether the forecasts are also drawn towards HIRAS. Ideally, the system would assimilate another instrument and leave both CrIS and HIRAS passive. In fact, both CriS and CrIS2 are flying during the Feb-March 2019 period of analysis. So this could in fact be done. 

 

4. It would also be helpful to include a description of the analysis methods used for other IR satellites (CrIS, IASI, AIRS) in the introduction.  And in the discussion, describe how the accuracy of HIRAS compares to these.

 

Minor points:

 

Line 115: Do you mean thin parameters or thin observations? Line 118: What is the temporal resolution of the forecasts.  Line 199: I believe that this plot is the MEAN departure. This should be indicated in the caption and on the plot title.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors have made the corrections to the paper needed for publication.