Cloud radiative feedbacks and rapid adjustments, both prime sources of climate model uncertainty, are increasingly diagnosed in climate models and observations using the cloud radiative kernel (CRK) technique. A limitations of how CRKs are typically applied is that they rely on either passive satellite data or model output that mimics passive satellite data which, in either case, can provide a misleading representation of the low cloud behavior as the passively sensed high clouds obscure lower level changes. Likewise, nonlow cloud radiative changes can be misinterpreted when the observation/simulator is misrepresenting the low cloud state.This technical note addresses this issue, providing a guide and code for overcoming this issue as best as possible. It then demonstrates the extent to which this obscuration effect can bias the magnitude (and in some cases sign) of the low and nonlow cloud feedbacks (or adjustments). This manuscript is well written, very polished, and timely, as the passive satellite simulators needed to apply this method in models figure to play a large role in the upcoming CMIP7.  I have a few comments below I hope the authors can address, but otherwise this manuscript is in good shape for publication.

Equation 1 and surrounding text: If I understand correctly, framing of fractionally unobscured, or clear-sky fraction, is really only relevant in the context of grid-scale histograms,. Whereas, at the actual satellite pixel-level, we can’t really differentiate between fractions of cloudiness/obscuration. I think it’s worth clarifying if so that this is specifically applicable to the use of CRKs /joint histograms and not pixel-level analysis.

Section 3: One can imagine a scenario where high cloud fraction changes between the perturbed and control state while a low cloud appears in the perturbed state that truly was not present in the control state.  How is this scenario differentiated from the scenario in Figure 1 where it is assumed there is no low cloud in the control state even though it is present and just fully obscured?  The former scenario is a covariance term case while the latter is an obscuration term case, but can output provided by the simulator actually distinguish between the two?

Page 8 footnote: The text mentions that the ISCCP retrieval algorithm reports a single cloud type using the optical depth integrated across all cloud types, including a lower-level cloud beneath an upper-level cloud. Does this suggest a disconnect between the model simulator, which would know a low-cloud is present as generated by the model subcolumn, vs. an actual ISCCP passive satellite retrieval which could not see the low level cloud and thus would not be accounting for any low cloud in the optical depth/integrated extinction estimate?  Or am I missing something?

Line 240-248: Total feedback magnitude is conserved with these corrections as they are essentially just moving radiative changes from one category to another, but is total feedback inter-model spread not conserved?  If both low and noncloud cloud amount feedback spread are reduced as noted in this paragraph, that must mean either total feedback spread is able to decrease, or it means another type of cloud feedback’s spread is increasing after these corrections in order for total cloud feedback inter-model spread to remain the same. 

Review ‘Recommendations for Diagnosing Cloud Feedbacks

and Rapid Cloud Adjustments Using Cloud Radiative Kernels’ by Zelinka et al.

This paper presents a methodology for diagnosing the impact of changing obscuration on cloud feedbacks and adjustments, and it quantifies these effects across climate models. It documents the effects of obscuration across models, demonstrating that, when obscuration is considered, the multi-model mean radiative effects of both low and non-low clouds are reduced in both cloud feedback and cloud adjustment. Additionally, this paper offers best practice recommendations and provides a codebase.

The paper is well-structured and well-written. I have only minor comments and recommend that it be published once these points are addressed.

1. According to Fig. A-2, the inter-model spread in global mean low cloud feedback and non-low cloud feedback appears similar, particularly after modification. However, the spatial maps of the standard deviation of cloud amount feedback reveal additional details. For instance, cloud feedback in stratocumulus deck regions is known to vary significantly among climate models, though the spatial coverage of these regions is limited. It has been suggested that the area-weighted contributions should be considered to identify which regions most influence the inter-model spread of cloud feedback. After accounting for obscuration effects, the standard deviation of low cloud feedback over stratocumulus deck regions remains the largest, but the spread is considerably reduced. Could you comment on the cloud types and/or regions that most contribute to the inter-model spread of global mean cloud feedback after modification?
2. I noticed that the authors added equations for obscuration effects at three vertical levels: high, middle, and low. If the threshold for 'low clouds' is defined at 680 hPa, then high and middle clouds combined would correspond to 'non-low clouds' in a two-level categorization. I am interested in (1) which of the middle or high clouds are primarily responsible for the obscuration effects on low clouds, and (2) how significant the obscuration effects of high clouds are on middle clouds.
3. I understand that the authors have developed cloud radiative kernels for radiative feedbacks and adjustments and documented the obscuration effects of clouds on these radiative processes. However, the modified cloud distributions themselves would also be valuable information. Is the provided code capable of outputting these modified cloud distributions for low and non-low, as well as high, middle, and low cloud levels?
   
   
   
   

L113-114: Can you describe a (few) example(s) of the context or needs of having different definitions of upper- and lower-level clouds?

L126: (I cannot type in upper score here, so use [], instead.) [ L_RF] must be [ L_R][F] 

L131 & L145: The authors explain their approach by using the areal fraction of low cloud and clear sky. They then describe breaking this fraction down into 'amount,' altitude, optical depth, and residual components. However, I find the use of 'fraction' and 'amount' confusing. It is unclear how a 'fraction' can be divided into 'amount,' altitude, optical depth, and residual components. Could you please clarify this?

L283: halved ‘in the opposite sign of feedback’

L304-307: Cannot follow. Please clarify.  Positive rapid cloud adjustment: 1) ‘ … being dominated by nonlow clouds’: yes, but it is because adjustment by low clouds in the original is negative. 2) ‘switches to being dominated by a large positive low cloud contribution…. is opposed slightly by a small nonlow cloud contributions’: are you talking about negative values in low cloud adjustment in Fig 8(e) and small values in nonlow cloud adjustment in Fig8(b)?