This study retrieved the SIA/SIE uncertainties by taking into account the spatial and temporal error correlations of the underlying local sea ice concentration products. It shows that random SIC errors play a role in SIA uncertainties comparable to inter-SIC-product biases. This study also compiles the September SIA trend with an explicit representation of temporal error correlations from 2002 to 2017. Extending this research to regional studies would help to investigate the impact of the correlated SIC uncertainties on oceanic and atmospheric surface fluxes. I would suggest a minor revision before it can be published.

(1) It would be better to use the exact values of SIE and SIA uncertainties in the abstract rather than showing an approximate value.

(2) Please indicate why you are focusing on CCI SIC product at 50km grid spacing product instead of that at 25km grid spacing.

(3) Please further demonstrate the advantages of using Monte Carlo approach in this study.

(4) Why was 2015 chosen as a case to demonstrate the error correlation length scale and SIA/SIE standard deviation?

(5) It would be better to add more figures to make the Results and Discussion sections more convincing and easier to understand, e.g., superimposing a graph of SIA trend and the linear regression of CCI product in Figure 5(b).

Review of “Estimating the uncertainty of sea-ice area and sea-ice extent from satellite retrievals” by Wernecke et al.

Summary

This paper presents a method to estimate uncertainties in passive microwave-derives sea ice extent and area. It is derived from spatial and temporal errors in the gridded concentration fields. The approach yields estimates of uncertainty in daily and monthly extent and area values and the paper provides trend estimates with accompanying uncertainties.

General Comment

This is an excellent paper. It provides a logical, quantitative method to estimate extent and area uncertainties based on the characteristics of the gridded sea ice concentration fields. Such quantitative extent and area uncertainties have long been lacking, which is a significant limitation in the passive microwave products that are a key indicator of warming and climate change. The use of these extent and area uncertainties to derive uncertainties in trends is also highly valuable, particularly for the Antarctic where trends are near-zero and uncertainty is needed to assess if trends are significant. The paper is well-written and explains the methods and results well. I recommend publication after only minor revisions, noted below.

Specific Comments (by line number):

79: “tie points” is used here, but not defined. It is defined later in the paper in lines 244-245. Readers may not be familiar with the term, so it should be defined here when it is initially used.

83: “constant biases” – aren’t biases by definition a constant? I think you mean here that the biases are consistent throughout the various product – i.e., a land difference is a constant offset – as opposed to differences between products due to methodologies (channels used, tie point values) that have mean biases but with variable differences depending on conditions. I think it would be fine to just remove “constant” and just say “biases” as the source of these biases are mentioned.

115-116: This paper essentially uses the results from Kern (2021) and Kern (2021) as the basis for the whole approach. In light of that, I think a short summary of the method and data is warranted. Though the references obviously explain things in detail, I think having a brief explanation would be helpful to allow readers to have a sense of those papers without having to go to the external references. Again, it doesn’t need to be detailed, but at least 2 or 3 sentences summarizing the data and method used for both the spatial correlation (2.1.1) and temporal correlation (2.1.2) would be a good foundation for the rest of the paper. It could also be done for both spatial and temporal in Section 2.1, as an introduction, before going to the two subsections.

221-229: It is most useful to have trend values with the quantitative uncertainties derived based on the spread of the ensemble members. This provides the trend uncertainties based on the uncertainty in the extent and area values. However, there is also the significance of the trend based on the “noise” in the linear trend fit – e.g., the trend standard deviation and/or the P-value of the trend (e.g., P<0.05); this assesses the confidence level in the trend based on the length of the timeseries and the year-to-year variability. This is the number often calculated and quoted with trends. But that is different than your estimate based on the ensemble members. I think it would be worth making this clear and perhaps it would warrant a short discussion (maybe in Section 4 or 5) of what this means for understanding the trend significance. This is particularly key for the Antarctic where trends have been near-zero, but have varied between small positive and small negative trends – are these changes really significant given what you have shown about the uncertainties as well as the trend standard deviation values?