High-Resolution Spatio-Temporal Estimation of Net Ecosystem Exchange in Ice-Wedge Polygon Tundra Using In Situ Sensors and Remote Sensing Data

Round 1

Reviewer 1 Report

Dear Authors,

I have gone through the article and found that the manuscript is well structured, the language is correct and clear and the title and abstract clearly describe the content of the manuscript. The study suggests the importance of considering microtopographic features and their spatial coverage in computing spatially aggregated carbon exchange. In my opinion the manuscript is almost ready to be published after minor correction below: 

1) In my opinion the name of the country must be included in title. 

2) spatiotemporal should be written like spatio-temporal

3) Introduction is written very lengthy, I suggest the authors should simplify these parts.

4) Resolution of figure 2 need to be increase 

Author Response

Please see the attached document. 

Author Response File: Author Response.docx

Reviewer 2 Report

The study is good with great application potential for future studies in the Arctic area.

Author Response

Thank you for the comments. We have corrected typos and grammatical issues. 

Reviewer 3 Report

Review of “High-resolution Spatiotemporal Estimation of Net Ecosystem Exchange in Ice-Wedge Polygon Tundra Using In Situ Sensors and Remote Sensing Data”

Haruko M. Wainwright et al. integrate remotely sensed greenness index and NDVI, LiDAR derived DEM, and CO2 flux data from an EC tower and sample chambers along an automated tram to estimate (daytime?) NEE over ice wedge polygon tundra ecosystem. They develop and use a Kalman filter approach to integrate datasets. This is an interesting and relevant research question as efforts are currently focused on understanding Arctic ecosystem C cycling. While the overall approach is sound, there is some missing information that would be helpful to readers, and some further analyses that should be completed to demonstrate the utility of this approach.

First, the authors use NEE but are vague as to the actual timing that NEE is measured and simulated. This is NEE during the daytime and should be designated as NEEday. NEEday during the growing season is typically not solely a function of leaf area (or its surrogate NDVI as estimated here). How does PPFD interact with LAI? In most terrestrial systems, lowest NEE day occurs at maximum solar irradiance. Authors should consider providing either a trace or integrated PPFD values for the five DOY periods that they sampled. The authors have measured PPFD (or related irradiance) on the tram, presumably, but it is unclear why they only use the relationship in Figure 6a to model fluxes.

Further, their analyses seem to be only part of the story; what happens during the nighttime? These are no doubt minimal amounts of time that nighttime occurs during the growing season at this site, but an estimate of how many hours the NEE values account for would be helpful to readers. Then, it is then confusing in Tables 1 and 2 as to what these NEE values represent, and which time periods they cover. Also, the average NEE and NDVI values in this table need error terms because there is variability associated with each value. A key question that is not addressed here is how does air (or ambient) temperature affect the spatial distribution of NEE in these systems?

In the Methods section, there is not much information about the duration of the sampling during the 8 measurement periods; the authors report 5 of these DOY periods. How long is their sampling period, and are they midday, throughout the day, or…? This would be helpful information for the reader. The authors are also not clear about the EC sampling, providing few details as to how they processed the NEE data. How are missing data gap-filled, or are these periods ignored? Do they include a CO2 storage term? How was the system calibrated? Although the footprint information is interesting, how do they really use this information to evaluate the NEE from the chamber measurements?  

Authors are using datasets from different years; does the NDVI greenness product line up with the NEE data? It is not always clear when each dataset was collected. For example, in Figure 2, panels a and b are relatively insensitive to the timing of these images, however, c is the greenness index and this is time sensitive. Can the authors be more specific as to the date of this image; is this peak greenness/NDVI for the growing season, or?

Figure 5: The bottom two panels for soil moisture content and NEE should have error bars as these are presumably replicated measurements. Also, the solid lines between values are a bit misleading, because these are point estimates. The x axis is labeled as distance (m) which is distance along the tram, but there may be a better way to label this to provide more relevant ecological information?

Figure 6a. NEE vs. NDVI. This figure is the main relationship used to scale NEEday, but the authors provide little statistical information about how well this predictive equation performs. Authors should provide a r2 and F and p values either on the figure or as a separate table.

Figure 6b. This is odd because it shows a clear mismatch of Ig and NDVI during some sampling periods. Would it make more sense to plot Ig vs. NDVI for the time period that is closest to when Ig was measured?

Figure 7, and/or the photo in Figure 3b. It would be helpful to plot the position of the eddy flux tower as the authors have done with the tram and indicate an averaged daytime flux footprint around it. The authors need to be clear what NEE value they are reporting on Figure 7. Is this average, 24-hour NEE, or peak NEE?

Figure 8. The authors present a time series of daily fluxes (presumably 24-hour NEE, or is this only NEEday?) derived from the EC data and scaled NEE from the tram system. A much more effective way to show this relationship would be an x, y plot with correlation statistics (r2, F and p values; these are independent estimates but may be autocorrelated time series). This figure should also designate the timing of snow melt and then first significant snow fall. That way the entire growing season is presented, if possible.

Tables 1 and 2 should have error terms reported for the averaged values of NDVI and NEE. The authors could also test for significant difference among values, because if there are no significant differences then the approach is still valid, but the distribution of NEE is relatively invariant across this landscape. It’s still important to know this.

Specific comments:

Some of the Figures contain legends or labeling that is too small and thus difficult to read.

Line 477: The authors have not demonstrated how closely the NEE tower data is to the scaled NEE sample chamber data, so that “successful integration” may be a bit premature, given comments on Figure 8.

Lines 496-516: This is a nice summary paragraph about the importance of this study, and it interprets the findings of the study well.

Lines 518-519: How would repeated Ig images improve this approach? This and a few other changes to improve this study could be mentioned here, such as integrating nighttime data for complete NEE data over the 24-hour period.

Line 526: Where in the results do the authors show the Spearman’s rank coefficient data for the relationship between NDVI and NEE? This should appear in a Table or otherwise be included in the Results section.

Line 529-530: What do the authors mean by “can compensate for each other…”? This is an important sentence summarizing the approach at peak values, and should be clear.

Lines 541 and 584: “cancel out” may not be the correct term here. Maybe “variation is reduced when integrated” or equivalent?  

At the end of the Discussion, it would be interesting if the authors would suggest a few improvements to their study so that complete growing season NEE could be partitioned, as they have done with NEEday. Also, how do the authors anticipate changing air temperatures and longer growing seasons would affect their results, and specifically each of the types of landforms in the Tundra polygons?

Author Response

Please see the attached document. 

Author Response File: Author Response.docx