Dear authors,

The reviewers of the manuscript “Tree-grass phenology information improves light use efficiency modelling of gross primary productivity for an Australian tropical savanna” recognized the work as relevant for the audience of Biogeosciences, so do I. The Reviewers’ suggested a series of additional analysis and editorial modifications to improve both the readability and the robustness of the analysis. In the Interactive Comment the authors nicely discuss how the reviewers’ comments can be incorporated in the revised manuscript. Therefore, the manuscript should reconsidered after the major revisions suggested by the Reviewers’ will be included.
One contrasting comments from the reviewers was about the trade off between length of the paper and methodology details provided. I agree with the authors consideration that, although the paper is lengthy, it needs to include a range of measurement that need to be properly described for sake of repeatability. The use of supplementary materials can be an option.
During the review I suggest the authors to address the issue raised by Reviewer 1 about the calculation of APAR, which can also help to address the comments of Reviewer 2 and 3 about the time lag between GCC, APAR, and GPP, as well as the differences between GCC and APAR.
Also, the authors to discuss more deeply the issue raised by Reviewer 1 about the estimation of the peak LUE. I consider valuable the idea of the Reviewer to use running means instead of the average by months. Another option for the authors is to optimize the peak LUE value.
The reviewers suggested also a valuable analysis to assess the robustness of the indices extracted from the digital camera data.
Last but not least the Reviewers pointed out a series of modification to Tables and Figures that need to be included.
Looking forward for reading the revised version.
Best Regards,
Mirco Migliavacca

Dear authors,

in the revised manuscript you addressed properly all the concerns of the reviewers.
The manuscript reads quite well and the analysis robust, and it will help to support further analysis on the phenology of Savanna and in general tree/grass mixed ecosystems.

The manuscript can be published after some minor revision

>P6 L1 Please describe very shortly the partitioning method used. Was it based >on a light
>response curve or night time NEE-temperature curves?
>We used a u* filter and artificial neural network approach, with soil water, soil >temperature,
>air temperature and EVI as the main model drivers, to determine respiration >(R)..

Can you please clarify which EVI was used? EVI MODIS or from field spectrometers? how was smoothed to produce daily time series (if it was)? a clarification could be useful for the readers (and at page 6 line 1 I haven't found the information).


Discussion with Reviewer #2 about automatic white balance
> However, white balance was set to zero in our analysis, which is a limitation in that
>it increases the scene illumination noise in our images.

If I understand correctly the white balance was kept fixed. In this case please clarify this aspect and the comment of the reviewer is amended. Of course the use of midday data also help as stated by the authors. The use of automatic white balance can have important impact on the digital number (DN) in particular of the blue channel impacting not only the noise but the seasonality on the GCC time series.

Sincerely,
Mirco Migliavacca