The manuscript by Pirk et al. investigates links between net ecosystem CO₂ and water vapour exchange and snowpack dynamics at an alpine tundra over a three-year observation period. The authors combine observations using the eddy covariance technique with remote sensing observations of snow cover and of vegetation indices and with land cover classifications. One of their main findings is that the site turned into an annual net CO₂ source during the year with most snow accumulation and consequently the latest snowmelt date, while the site was a net CO₂ sink during the other years when snow cover was close to the long-term mean. Understanding interactions between snowpack dynamics and land-atmosphere exchange of CO₂ and water vapour is crucial to better predict climate change impacts on alpine ecosystems and the topic of this study is thus timely and addresses an important research topic. However, in my opinion, the study would strongly benefit from clearly defined research questions that better link to the authors’ analyses. The submitted manuscript presents various analyses that are only loosely connected. The lack of coherence makes it difficult for the reader to grasp the most important findings/implications of this study. The authors may consider reframing the study and to link each analysis to a specific objective. For example, it remains unclear how the spatial variability in melt-out dates across the landscape links to interannual variability in NEE and ET. Furthermore, the authors demonstrate a link between snow accumulation and synoptic atmospheric circulation patterns. However, it remains unclear how this analysis then links to the flux tower measurements. The authors present a range of interesting findings that would have a much stronger impact if they were logically connected.

Specific comments

In some cases, the author report correlation coefficients (r) and in some cases the coefficient of determination (R²). I would recommend a consistent use of one of the two metrics.

Line 51-55: It remains unclear how the analysis of water-use efficiency contributes to the main goals of this study.

Line 77-84: Here, listing of the main objectives would help framing the study. As it is written now, it emphasises the “exploration” of various datasets, but I think the logical links between these analyses need to be explained.

Line 92: I do not think that the “minimum 30-min average” is a good metric to support the statement that winters are mild.

Line 120: How were these limits determined?

Line 224-225: This statement should be supported by observations. What is the contribution of February to total winter snowfall events?

Figure 2: How was NDVI treated when ground was snow covered? This analysis would be strengthened if statistical analyses of differences between land cover types would be presented.

3.2 Flux dynamics in the two footprints: The authors analyse two footprints separately, which is a reasonable approach if underlying land cover composition is very different. However, the authors find very similar NEE and ET dynamics and it remains unclear what the added value of this analysis is for the study. A better explanation of the separate treatment of the two footprints would be useful.

Line 328: Perhaps rephrase: “in order to maximize leaf area”

Line 372: Gap-filling algorithms like Marginal Distribution Sampling (MDS) do not prescribe functional relationships.

Pirk and others explore the response of a Norwegian alpine tundra ecosystem to a year with anomalously late snowmel. The Introduction bounced around a bit between different topics including carbon flux, plant succession, global change, hydrology, remote sensing, and more. All of these things are interconnected of course, and the Introduction was very nicely cited, but the topics could be linked more clearly to point toward the particular topic of this study. As a consequence it wasn't entirely clear why the landsat, sentinel, and modis observations were used when modis measures more frequently at coarser scales and landsat and sentinel measure less frequently at finer scales, and how these observations fit together. Was MODIS for historical melt out dates and how were melt out dates characterized for the 16 day landsat overpass? The results are interesting but I had a difficult time understanding how everything fit together. 

Are fig. 5 c and d on log scales?

The eddy covariance measurements were discussed nicely and the gapfilling approach was well suited to the site. 

All in all with some restructuring and focus on a consistent narrative the manuscript will be publishable as it makes some interesting points.