The manuscript „AIUB-GRACE gravity field solutions for G3P: processing strategies and instrument parametrizations“ summarizes the analysis decisions taken at the Astronomical Institute of University of Bern (AIUB) for deriving monthly-mean gravity fields from sensor data collected by the GRACE and GRACE-FO satellite missions. Calculating gravity fields is a complex task involving the careful consideration of data from a range of different sensors (i.e., range-rate tracking, accelerometers, star cameras, GNSS navigation signals, etc.). Besides common standards & conventions, a number of choices are available to the processing centers like AIUB for individual aspects of the data pre-processing, screening, and integration. Documenting those choices as done in the current manuscript is important to foster international collaboration in order to identify the most promising ways to calculate GRACE-based gravity fields. The manuscript is generally well written and certainly fits the scope of the journal. Nevertheless, a number of comments might be considered before eventually accepting the article.

More details are needed on the outlier handling (P6L3). It seems to me that complete daily arcs are flagged and ignored, is that true at this point already? Do you have an idea about how much (potentially valid) data is removed by this step? What about options to calculate arcs that span over shorter periods of time, or that are shifted by 6 (or 12) hours? 

The question above is also related to your statements on p10, where the elimination of whole days is described as the last stage of data screening. From the text, it is not immediately clear which approach exactly is taken for the most recent G3P release from AIUB. AIUB-RL01 does not need to be discussed again at this stage. If there are no changes with respect to AIUB-RL02, this should be stated so. 

Results presented in Figure 7 need further analysis, in particular with respect to the months that appear to be worse than in previous releases. I suggest to explicitly map (i) the monthly-mean background model, (ii) the mean from Peter et al. (2022) and the residual obtained with both AIUB releases leading to the RMS values presented in the time-series for those months in question. Please exclude continental signals and choose a color range & spatial smoothing that amplifies signals in the oceans.

P1L5: RL06 is not the latest release of AOD1B anymore, since release 07 is already available (https://doi.org/10.1093/gji/ggad119). I suggest to skip the word „latest“ and mention only that RL06 is now incorporated instead of RL05, which makes this processing choice consistent with the RL06 gravity fields of both the SDS and many other contributors to Cost-G. 

P3L1: „non-unit“ diagonal elements sounds odd. Please revise.

P4L1: we can not see the star camera data in eq. 1 -> star camera data is not explicitly identified in eq. 1

P21L1: Author contributions of AJ and UM might be elaborated a little further in view of, e.g., software heritage used for this work.

The Astronomical Institute, University of Bern is processing the GRACE/GRACE-FO data for many years and generates GRACE/GRACE-FO based temporal gravity field solutions and orbit products. Recently, AIUB published a new release of its gravity field time series called AIUB-G3P. This paper now describes the processing strategy for this new release, especially compared to that for the previous release AIUB-RL02. I appreciate this paper since it gives important hints and background information about AIUB’s new gravity field product. Nevertheless, I’m not satisfied with some items, especially in the evaluation part, see my comments below. Therefore, I vote for major revision.

I have the following specific comments:

P6L3: You write: “Our new strategy for AIUB-RL03 is inspecting the GRACE L1B data product itself, and whenever we find out an outlier, we throw out the epoch of the outlier …” My recommendation: You should mention already here in this sentence in which kind of data you are looking for outliers. I suggest to write “Our new strategy for AIUB-RL03 is inspecting the GRACE L1B data product itself (KBR1B and ACC1B), and whenever we find out an outlier, we throw out the epoch of the outlier …”

P7L20: I suggest modification of this wording “a much looser outlier threshold”  into  “a much larger outlier threshold”

P8L1: Please correct: “3 shows” into “Figure 3 shows”

P10L4 and Figure 4: Your write “Then by plotting and 5 comparing the n gravity field solutions in terms of difference degree amplitudes of geoid heights …”.  Please add here that you compare with the monthly gravity field solutions ISTG-GRACE2018 and explain why you have chosen this TU Graz model for this comparison.

Please add the corresponding plot for Iteration 1 in figure 4.

P13L1-3 and Figure 6:   Yes, noise reduction of AIUB-G3P compared to AIUB-RL02 is obvious. But I do not really understand the “mean model”. You state this model is simply the average of monthly gravity field solutions of several processing centres and you refer to Peter et al. 2022. But I’m sorry, I cannot find a description of such an average model in this paper. Or do you mean COST-G combined models? Please specify and explain also why you chose the "mean model".

P14L5 and Figure 7: “Figure 7 shows 5 the noise over the ocean”: Please describe in detail what you mean with “noise” resp. with “Weighted STD over the oceans”, please specify what you computed. Please explain furthermore, why you did this computation over the oceans and not over land.

P14L8 and Figure 8: “Figure 8 shows the RMS over the years spatially”: Please describe also here what you computed. The caption of figure 8 also unclear for me: What are “RMS anomalies”? Has is something to do with gravity anomaly? In contrast, the color bar is indicated with “Temporal RMS in geoid height”. What is “temporal RMS”?

P15L3-6: The content of these sentences has already been explained previously in section 2

P15L7-14 and Figure 9: Please describe also here what you mean with “noise over the oceans”. I suppose this is the same functional as for figure 7.

P17L3ff and figure 11:  This evaluation part is nice, but it’s about gravity field solutions from JPL and TUG and has therefore no relation to the topic of your paper. Please remove this part.