|The Authors addressed both Reviewer’s main methodological concerns and have performed an extra extended analysis on the dataset according to temperature, wind direction, weighted footprint function (α) and time of day. The results of this analysis are provided in the supplementary material. These extra figures show interesting trends in the dataset according to temperature, footprint function and time of day, while the directionality of the data seems to be negligible. |
The Authors have added some paragraphs of discussion on these trends and on the respective uncertainties of their method in the manuscript. However, they have not taken the new data insights into account for the calculation of their results, keeping the results of the paper unchanged.
I agree with the reasoning of the Authors on the negligible effects of the directional variability on Fnatural, but not about the effects of temperature and time-of-day variability (Lines 272 – 277). There are clear trends according to temperature and time-of-day for all studied species (Figures S2, S3). These trends indicate that Fnatural cannot be kept constant in Eq. 2. Moreover, since the few data used for Fvilage estimation is almost exclusively between 9 – 15 h (Lines 317 – 318) and the temperatures range mainly from -1.5 – 10.5 (Figure S5), the medians used for Fnatural (24h, temperature range: -10,5 – 13.5) are not representative. From Figures S2 and S3 it can be derived that Fnatural for N2O and CO are generally underestimated and for CO2 overestimated in the applied method.
Ideally, the method should not keep Fnatural constant, but variable according to the actual environmental conditions at the time of the measurement. I assume that if the Authors would reprocess their data and extract Fvillage taking into account the actual temperature and time-of-day statistics (or even incoming radiation statistics, if available), then their results would be significantly different. Possibly the huge difference between TD and BU N2O emissions would be decreased and for CO2 would be increased. Also, the time-of-day variability of Fvillage in Figure S5 would be different, since it “carries” the effects of Fnatural variabilities.
I therefore ask the Authors to redo their analysis according to the above reasoning and update the manuscript accordingly. If I am mistaken and the differences from the current Fvillage estimates are negligible even when the Fnatural is variable, then I would ask the Authors to at least provide the statistics in their answer and not change the manuscript significantly.
Lines 215-216: Please add in parenthesis that to derive α, the integral concerns all grid cells appearing in Fig. 3.
Figure 5 Caption: Shouldn’t the 4277 be 1120?