Dear Authors,
I have now read your responses and found some of them absolutely satisfying, especially the arguments that you put forward defending the mechanistic model approach. And I recommend to mention this reasoning explicitly and visibly in the text (e.g. in an own paragraph in the discussion).

Before you start working on your resubmission please consider the following changes:

1. It is good publishing practice to keep the text as short as possible and strictly related to the novel scientific content. This is valid both for the main text and for the supplemental material. The supplemental material of this manuscript must be heavily shortened. You can't use the supplemental material to publish a comprehensive model documentation. For this I recommend the EGU Journal GMD. Please follow closely the respective recommendations of Reviewer 1.

2. A weakness of the manuscript is lack of quantitative model performance. I completely agree with Reviewer 1 that showing some time series and discussing them is neither representative nor comprehensive. If I understand it correctly, you have now expanded the respective figures and include all data, which is good but will probably make the figures even less readable. Time series are not useful to get a clear impression of the model performance. If one would like to display the model performance in a figure, simple scatter plots (e.g. measured = y axis versus modelled = x axis - not as time series) were much more useful. However these graphs omit the temporal development and therefore cannot simply replace the time series. Please revisit the literature, there are many quantitative indicators for model performance (e.g. linear regression parameters, slope and offset and their significance and uncertainty, correlation analysis, and a number of more specific model performance parameters) that should be used in and can easily be presented in a table.

3. You did not comment on the lack of uncertainty analysis. Reviewer 1 mentioned “The absence of any uncertainty analysis, and the focus on just a few of the many emission events observed, and the use of data from just one site, also make it hard to evaluate the work.” Please provide an uncertainty estimation in the manuscript. You claim that the quantitative relationships that you use in the model have not been fitted from the site but arise from other studies or theory. You could, e.g., use the uncertainty that is connected with these relationships and estimate the consequence for the uncertainty of the predicted N2O fluxes with a state of the art Monte Carlo approach.

With kind regards,
Andreas Ibrom, Associated Editor

Dear Authors,
thank you very much for the consideration of my suggestions.

It is very good that you made a sensitivity analysis for two key model parameters. An uncertainty analysis would now calculate the uncertainty of the N2O flux simulations based the uncertainty of the parameters, which you could have easily estimated from the spread of different cited literature values, times the model sensitivity that you have estimated in this very simple sensitivity analysis approach. A complete uncertainty analysis would however use all relevant model parameters and would estimate the corresponding model uncertainty in all reasonable parameter value combinations (e.g. Monte Carlo simulation). It is exactly this complication that complex models have to cope with. Maybe you will consider this for future work.

The new table 3 with regression coefficients shows honestly that the model performance is not exceptionally high, for reasons that are however very well explained in the corresponding text and discussion. I was wondering whether you could have given an additional measure on the model performed when considering time integrated values, e.g. daily, weekly or monthly data. Here the exact timing for the emission pulses will have less influence but you would on the contrary show, whether or not the seasonal N2O flux dynamics were captured well by the model.

You submit version 3 of the manuscript. The reviewers have only seen version 1. To ease following the changes we ask authors to submit a marked-up manuscript version (track changes in Word, latexdiff in LaTeX) converted into a *.pdf including the author's response must be submitted (see http://www.biogeosciences.net/for_authors/manuscript_preparation.html authors response). I have seen that you changed more than you mentioned in your latest author comments (bg-2015-621-author_response-version1.pdf), e.g. a new figure 1 (which, by the way, needs a change from a slide type to a publishable figure). Please provide me with the requested document that marks up all changes relative to the submitted manuscript (version 1).

I’m sorry for the inconvenience and hope for your understanding.

With kind regards,
Andreas Ibrom, Associated Editor.

Dear Authors

I'm pleased to inform you that your excellent work and manuscript is now accepted for publication in Biogeosciences.
I thank you for your patient and fast collaboration.

Please change on page 55 "This rises were similar" to "These rises were similar".

With kind regards,
Andreas Ibrom, Associated Editor