Dear Editor

Along with my greetings, the comments are attached. I think the article's idea is good but is not well organized. Additionally, one equation requires revision. 

Greetings

General comment

Payraudeau et al. present a novel modelling study advancing the field of simulation of pesticide and isotope ratios at catchment scale. Studies as this one are rarely conducted but very important to understand more on pesticide fate at catchment scale. I have two main concerns that I strongly recommend that the authors take into consideration. These are elaborated below.

Main comments

1. A key novelty of this study is the application of a spatially distributed catchment model to simulate pesticide concentrations and isotope ratios (only in topsoil unfortunately, see main comment #2), as stated also on line 92. My first main critique to this study is that the added benefit of a spatially distributed model for this purpose is barely demonstrated nor discussed, albeit expectations are raised in lines 87-89. In that sense the added benefit of this distributed model approach compared to the earlier lumped (not spatially-distributed) model study by Lutz et al. (2017) -  applied to the same catchment but with an earlier dataset limited to catchment outlet (but no topsoil) data -  does not become clear. The focus in the model is more on linking degradation rates to temporal variations in temperature and soil moisture but that can also be simulated with a lumped model approach. There must be spatial heterogeneity in the catchment like in pesticide application at plot level, but also in spatial variation in soil texture and soil moisture content (Figure 1 nicely shows the spatial variations in crop types; and the descriptive text mentions various soil types). The added benefit of simulating such spatial variations much better in a distributed model than in a lumped model is not discussed. I recommend that the authors show and discuss this key advantage better as that seems the main novelty of this study. Or could a lumped model not have resulted in an equally good model fit as page 5 states that the spatial variation in soil parameters was in fact low in this (specific) catchment?
2. My second main critique is that the authors did not make use of the pesticide concentrations and isotope ratios at the catchment outlet (besides the topsoil data) as also reported by Alvarez-Zaldivar et al. (2018). I strongly recommend that the authors add the simulation of the catchment outlet data as reported by Alvarez-Zaldivar et al. (2018) to this work. I see this as a pre-requisite to allow to draw conclusions on pesticide degradation at the catchment scale. The model is now only calibrated to observations on the first cm of soil in the catchment. The (degradation) model is not calibrated to anything deeper than this top one cm. Calibrating the model also to the available catchment outlet data enables to draw conclusions on pesticide degradation at the whole catchment scale, and would make this a really unique study. All conclusions drawn on catchment scale are in the current version based on extrapolation of the topsoil data. Also include a discussion on the pros/cons of calibrating such models with only topsoil data (as done in this current work) vs. outlet data (as done by Lutz et al. 2017), or both. See later “other comments” that elaborate on this main comment #2.

Additional comments:

1. Title: why “models”? Plural? It seems only one model has been applied?
2. Introduction:
   1. Line 72: Explain the concept of CSIA better such that the reader can understand how information on degradation can be derived from isotope ratios.
   2. Line 78: some publications have shown that sorption can lead to significant isotope fractionation effects in the spreading direction of pollution which might be relevant in catchment studies as well. Why not test this with this model, as this seems straightforward?
   3. Lines 77-83: I feel that neglecting isotope fractionation effects of non-transformation processes can be more convincingly elaborated.
3. Methods:
   1. Line 177: explain what a mixing topsoil layer is. Ploughed?
   2. Some textual errors on lines 168 and 177.
   3. Line 181: I think that not the various isotopologues are simulated in the model but the bulk heavy and light carbon in the pesticide?
   4. Line 187: “numerical diffusion”? Do they authors mean numerical dispersion?
   5. Line 188: explain what the export coefficient is.
   6. Line 192-193: does this then not prohibit the simulation of hot-moments?
   7. Model approach general: I do not understand why this model has not been applied to simulate data at the catchment outlet as it seems to be designed to this end. Could not a simpler model been applied to simulate only the data in the topsoil?
   8. Lines 205-206: what is meant with the depth of the groundwater layers varying constantly from upstream to downstream?
   9. Page 9: as sorption was simulated in the model, why not test / show that sorption isotope fractionation has limited/negligible effects on the simulated isotope ratios in topsoil (and especially at the outlet?)?
   10. Page 10: add bit of explanation to argue why it was chosen for to simulate only the bioavailable fraction? How would it influence model outcomes compared to the more common model formulation that simulates biodegradation of the total dissolved fraction? Is this in line with earlier catchment degradation model studies?
   11. Line 265: “depth-dependent degradation”. The argument not to include this in the model seems rather weak (because there are no data available from deeper soil layers; but there are in fact catchment outlet data available). If that reasoning is followed the authors should also only draw conclusions on what happens in the top soil layer and not in the catchment as a whole. Lutz et al. 2017 included “depth-dependent degradation” to enable fitting the concentration and isotope ratios of pesticides at the catchment outlet. But the current model is not constrained with such outlet data and therefore there will be high uncertainty on what happens at the catchment scale. In fact as the model is only calibrated with topsoil data (and discharge at the outlet but that is not that relevant for catchment scale degradation extent), the model is therefore only informative on what happens in the upper 1 cm of the catchment. See also main comment #2.
   12. The three applications as shown in Fig. 3 are not described in the methods section (or I could not find it): how much pesticide was applied when and where? Exactly this heterogeneity in source zone variation calls for a model like this one but the advantage of using a distributed model has not been illustrated.
4. Results & Discussion:
   1. Generally: explain why concentrations and isotope ratios of pesticide at the catchment outlet were not used. I assumed this was done when starting reading the paper but only in the R&D section is became clear that only data from the upper 1cm of the catchment skin were measured and simulated. How then can conclusions be drawn on overall degradation at the catchment body when these simulations are not constrained with outlet data? Thus relatedly, what is the relevance of the simulations of the upper 1 cm in the soil as presented in the key result Figure 3 on what happens at catchment scale? See also main comment #2.
   2. The three levels of sampling resolution as presented in Figure 4: composite, transects, plots are unclear, and not clearly described and explained in the method section. This would help also on improving on main comment #1.
   3. Lines 440-441: in my words: “more detailed spatial soil sampling did not help to better constraint the model”. Why then was a spatially distributed model needed?
   4. Lines 443-447: the difference in fractionation factor -2.7 vs. -1.5 is quite large and needs further explanation. Also compare and discuss the outcome with the fractionation factor calibrated by Lutz et al (2017): - 1.3 permil.
   5. Line 458+459: “measure”, change into “measurements”
   6. Figure 5: what is the relevance if these results at catchment scale when the model is only calibrated on topsoil data but not on outlet pesticide data?
5. Discussion is largely lacking. Add a larger discussion section to include for example:
   1. Comparison with findings Lutz et al. (2017) who applied a similar model to the same catchment but then calibrated to earlier data at the catchment outlet instead of the topsoil. Add some discussion to discuss why this study did find that having isotope data let to better model results, whereas this effect was not clearly found by Lutz et al. (2017). Note that Lutz et al (2027) had to include depth-dependent degradation to enable to simulate the outlet pesticide and isotope ratio data.
   2. Pros/cons topsoil vs. catchment outlet data.
   3. Line 493: I thought the conclusion was that there are no “hot-spots” in this catchment, as there is limited spatial variation among the topsoil samples? Therefore, add some discussion what the added benefit of a spatial distributed model was in this case.
6. Hot-spot and hot-moments. The paper does not discuss hot-spots in the catchment. The advantage of this model is that it enables to account for spatial heterogeneity but it seems that all model parameters were taken spatially homogeneously except with depth. Therefore the added benefit to the model applied by Lutz et al. 2017 remains unexplored.
7. Line 510-511: see earlier comments: as pesticide concentrations and isotope ratios were only measured on the topsoil one cannot simply extrapolate to the catchment scale. Degradation rates are likely much higher in the topsoil than in the deeper soil layers leading to overestimation of a catchment to degrade pesticide. The data to constrain the model also at catchment scale are available. Why not use them?