The authors present a study where they test an ANN to upscale instantaneous remote sensing observations (Rsi, RsiTOA, RSdTOA, theta_z, and L_D) to daily Rsd estimates from where they estimate daily ET. These results are also compared to two other methods for converting instantaneous observations to daily ET estimates. The paper is well written and easy to read. Two reviewers were mainly positive, and the 3rd reviewer expressed some concerns on the validity/usefulness of the method during (partly) overcast days. I think the authors correctly replied to comments of the 3 reviewers and the proposed changes are OK. However, all 3 reviewers commented on the selection of the FLUXNET sites. How representative are the sites for different climates, biomes and time of the year/seasonality?? Although the authors replied to Reviewer #1 that they will elaborate on it, but that they already showed that it does not influence the training of the ANN, I think the study will benefit from a proof of this claim. Especially, since the main objective of the paper is to show the use of ANN for upscaling from instantaneous to daily. Therefore, I agree with the suggestion of Reviewer #1 to do a 'sensitivity' analysis for the selection of the sites in place, time and biome.

Minor comments:
P3L2 and L12: what is the need of using E_T and ETd? Are they not the same?
P3L3: all symbols in text in italic (throughout manuscript)
P4L11: "... variables (e.g., dialy..." (add comma)
P4L11: theta is not explained.
P5L2: better: ".. predict Rsd based on Rsi satellite observations"
P5L2-4: Objectives 2 and 3 are not really objectives of this paper (since this is already done in the past). It's more that the results of the ANN are used to apply one method to upscale instantaneous observations to ETd and that these outcomes are then compared to two other upscaling techniques.
P6L7-10: What is the use of having ET with the units MJ/m2/d and Rsd in W/m2? Please use one of the two for both.
P7L12: add space between (Rye et al, 2012) and mainly.
P7L18: What is PURELIN? Please briefly explain.
P8L23-25:I would link here to figure 1 and use the same letters for the 3 method. Thus a=Rs-method, b= RsdTOA-method and c=EF-method.
Eq5: combine into 1 equation
Eq6: combine into 1 equation
P9L13-16: Use here the same order as the order of Eq8-12
P10L10: Unclear/vague sentence. Please rewrite.
P10L13: suggestion: use time-of-day instead of time-of-daytime.
P11L2: The categories of Tau are not explained in the text. When is something belonging to Tau_1 and when to Tau_4? (like explained in the caption of fig 5.)
Table 2: Maybe make it more clear that "Rs, RsTOA and EF" are the 3 methods and not that e.g., the R2 refers to the performance of an estimation of Rs.
Table 1, 2,3: Maybe convert these tables into similar graphs like figure 11.
Fig 1: What is the difference between Rsd_pred and Rsd?
Fig 5-caption: "..between Rsd_obs versus Rsd_pred...". Furthermore, also explain the transmissivity classes in the main manuscript text.
Fig 10: this figure is hard to read. Improve quality.