CH₄ Fluxes Derived from Assimilation of TROPOMI XCH₄ in CarbonTracker Europe-CH₄: Evaluation of Seasonality and Spatial Distribution in the Northern High Latitudes

Round 1

Reviewer 1 Report

The Manuscript "Sensitivity Analysis of CH4 Fluxes Derived from Assimilation of TROPOMI XCH4 in CarbonTracker Europe-CH4: Evaluation of Seasonality and Spatial Distribution in the Northern High Latitudes" attempts to use recent TROPOMI-CH4 data in an inversion system dedicated to Northern High Latitude. There are very few inversion studies using TROPOMI-CH4 data as of today, and to my knowledge, this is the very first attempt to use such data for Northern High Latitude, making the manuscript relevant for publications after concerns are addressed.

General concerns

Scope of the study

The manuscript is generally well written and clear, but it is hard to know what objectives it tries to reach.

There is some ambiguity on whether the study is a pure academic exercise (with flux products being commented simply regarding the potential impact of satellite observations), or the conclusion have some value and could be compared to other NHL CH4 studies.

Data quality and filtering

It is a known fact the passive sounding in high latitudes is a very challenging endeavor. Very little comment is made on the data quality, on the filtering that is applied, etc. What is the impact of the albedo and the zenith angle on data quality? High latitudes satellite data are often discarded altogether and further justification should be given on how relevant and reliable the data used in the present study is.

The authors should give further information on these aspects, and in particular, they should show some maps of data availability per month  

Inversion set-up

Only little comment is given on the transport model resolution. Indeed, even though super-observations are computed by 1x1° boxes, this is still much smaller than the TM5 resolution outside of the zoom region. Such difference between the transport resolution and the observations is probably responsible on a large part for the discrepancies between the model and observations.

There is no comment on the spin-up used for the inversion. What initial conditions are used in TM5? And how long is the spin-up period if any?

Regarding errors chosen in the set-up, how was defined the transport error for satellite retrievals? It seems very small compared to the overall uncertainties. Moreover, the R matrix is chosen as diagonal (as classically done), whereas it should not be with satellite data. This choice is reasonable in the community for technical reason, but non diagonal terms should be compensated by higher standard deviations in the diagonal approximation.

Specific comments

- The averaging kernel formula are detailed. However, no specific comment is given on the vertical interpolation from the model to the averaging kernel pressure level. What approach has been chosen when the surface pressure differs between TM5 and the TROPOMI product?

- Figure 2 is hard to read. In particular, it would be necessary to be able to identify individual stations in the figure. A table with the station names and metrics should be added at least in supplement.

- The authors should consider adding a map with the TM5 resolution and the observation with their names.

- The observation error is taken as the direct sum of the transport error and the instrument error. It should be the square root of the sum of the squares. 

- TCCON bias is very strong. Is it a know behavior with TM5? Could it be explained? 

- The bias at surface sites is still very high in the posterior, even when in the invSURF set-up. What is the explanation for that? The inversion seems to have a limited impact on concentrations; is it a general feature of CTE-TM5?

 

 

 

Author Response

Please see the attachment for replies to the comments. For the draft paper by Lindqvist et al., please ask the Editor.

Author Response File: Author Response.pdf

Reviewer 2 Report

In this paper, a comprehensive analysis on changes of CH4 from remote sensing products and field measurements is presented. Topic is interesting, but there are some comments to improve its quality.

·         Sensitivity analysis (SA): Please let reader know what is sensitivity analysis? Did you analyze the effect of changes of input parameters on outputs? After reading the manuscript, "SA" is not clear for me.

·         Please present a workflow to describe steps of the proposed method. It is necessary, since there are some steps.

·         Please present a subsection about validation and another one for SA.

·         Line 229-232: it is a description about the method not results, please refer it in the methodology section.

·         Table1: based on RMSE values, I think a relative measure can present a better insight. Please provide absolute and relative measures, for example RMSE and (RMSE/mean_of_parameter).

·         Please discuss error sources in a part. It would be interesting to discuss error sources regarding RMSE or other measures.

·         Please provide recommendations for future works in the conclusion section.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper is focused on the determination of the CH₄ concentrations and fluxes in the Northern Hemisphere. This is the second greenhouse gas after the CO₂. The impact of the paper lies on its global scope, since this research reaches latitudes above 45º N. The temporal extension of this study ranges during 2018. Measurements are contrasted with values obtained from satellite observations. The annual cycle of concentrations is successfully described. However, the paper is mainly centred on the CH₄ fluxes, where varied calculation procedures are compared. Moreover, the CH₄ profile in the troposphere is presented. Three main regions are considered from the continental distribution, emission magnitude and seasonal cycle amplitude. The handled references are varied to determine the current state of the research about this subject. Consequently, the paper is quite complete, although some minor changes should be introduced prior to its publication in Remote Sensing.

The geographical extension of the paper is quite varied, since some results correspond to the Northern Hemisphere above 45º, other are divided into three regions: Canada, Eurasia+Fennoscandia, and Central+Eastern Europe, Fennoscandia is considered in Figure 6, and isolated sites are used in Figure 3 and Table 1. The authors should introduce a short explanation about the reasons to use this separated spatial distribution.

The authors should indicate the procedure to calculate the correlation between measured and calculated values. If the Pearson correlation coefficient is used, the authors should consider that a good correlation could not indicate a good agreement between both data series. A better statistic could be the index of agreement, since it considers the difference between measured and calculated values.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I thank the authors for their replies to my comments.

They improved the manuscript accordingly.

The manuscript can be published with minor corrections, although I recommend that the companion paper Lindqvist et al. is published at the same time as the present one, with proper cross references to link the two studies.

The minor corrections would mainly apply to the conclusion that should further highlights limitations of the study, in particular with regards to the unexplained bias with surface measurements.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf