|The manuscript has been widely improved in the structure, important information has been added about the model, and the scenarios have been more clearly described and interpreted. The addition of uncertainty estimates is much appreciated. |
I still have a few comments and questions which are detailed below, key points are:
-Checking the symbols used for the model variables and parameters which are sometimes varying in the text itself and between the text and the conceptual schemes
-Moderating the conclusions about impacts of the tested scenario on nitrate load considering the processes not taken into account by the model.
-I am surprised that the model does not use input data, especially for nitrate which is OK in the currently tested scenario which focused on P load reduction, but it limits the range of scenarios for which can be tested using the model. So nitrate is more used here for constraining the deep flow path than to properly test any management strategies about nitrogen (as explained by the authors this would need simulation greater than a decade)
-Adding some discussion about the non-simulated peak events on Phosphorus (PP mainly): it is ok that the model does not aim at representing all the variability and will neglect some processes. I fully agree with the authors that a daily time step model won’t be able to catch precisely all the variability of storm events, however since events are an important contributor of PP loads, as highlighted by the HFD time series, it is worth to discuss this point as a limitation of the model.
Abstract line 3: “AND which focusses”? (Suggestion)
p.2, paragraph 1, line 4: Just to point out that if the meso-scale may be the relevant scale for policy making in the UK, and that it is a relevant scale regarding the physical system, it is not always the relevant scale in terms of economic and human systems.
p. 2, paragr 2, line 7: “complex model simulations are prone to high uncertainty” Ok but I am not sure that it is not the case as well for simpler models.
p. 3 paragr 4, lines 8-10: I do not understand the last sentence of the introduction
p. 4, (1): Doesn’t it depend on the management issue?
p. 4, (2): “how nutrients are lost”, what do you mean? which pathway lost which nutrient?
p. 4 paragr. 1, line 5: (suggestion) “meso-scale diffuse MULTIPLE pollution”
Section 1. 3.
p. 9, paragr. 2, line 5-6: weekly time series of nitrate concentrations have also been shown to be sufficient for load estimates. The sentences “The correlation between C and Q … Cassidy and Jordan (2011)” is confusing, it seems to refer to SRP concentration in the middle of the nitrate paragraph. At the end of this paragprah: I assume indeed that nitrate is more concentrated in groundwater flow than in overland flows but is it probable that sewage effluent are rich in nitrate as well?
p. 9, paragr. 3, lines: 3-4: It is frustrating not having weekly PP data, especially because you argue to use the HFD time series to select a daily time step as relevant, while PP is the most sensitive variable to the monitoring frequency. Also, when I looked at the Figure 2 when it is cited for the first time (p. 8) I was wondering why there were no LFD for the TP, maybe adding the precision of missing data p.8 in the paragr. (1) so that it is known before (just suggestion).
p. 9, paragr.(i) Type “3” events ma be associated with decoupling in PP and SRP transfers… as they are used in the following, it may be worth to properly define “Type 1”, “2”…
p. 9, paragr. (iii) However, type “5” events on Fig. 2 seems to be associated with small peak of Q.
p. 12 reference to Figure 2: the model does not reproduce the background NO3 concentration but it does not reproduce any dynamic in base flow at all!! I am not sure that depicting this curve is really useful. In revanche when the authors say that “However analysis also shows the advantage of a constant leachate concentration” I do not see such analysis (or reference to the relevant paper)? What do you mean by constant leachate? Constant within a year or over the entire period?
p. 12, paragr. 2 “Type 4” events represent 75 % of the 12 identified events, meaning that 25% of the events will not be reproduced by the model. As events are an important contributor of PP export, computed reduction of P load in the mitigation scenario are probably overestimated.
p. 13, paragr. 1, line 3: “If the mode is ABLE (?) to capture”
p. 14, paragr. 2, line 8 (ii) : subsurface component is often supposed to be faster than the deep component due to difference of the hydrological properties in the material, or e.g. a decrease of transmissivity with depth.
p. 14, paragr. 2, (2.2.2) flux rates unit should be m.day-1 (point or separation between m and day)
p. 15 between eqs. 4 and 5: “into the subsurface DS and DG stores” should be “SS and Dg stores” isn’t it?
p. 15 between eqs. 6 and 7: the flow (QSUB) should be (QSS)?
p. 15 Eq. 7: idem than above and S(t-1) should be SSS(t-1) isn’t it?
p.16. After Eq. 10: What are the observed data used to calibrate K(N)? Do you have any observed concentration and flow data for the overland flow component?
p. 16, Before Eq. 11 “The constant concentrationS In the dynamic”
p. 17, Eq. 13: Add parentheses.
p. 19, paragr. 1: So I understand that the hydrological criteria values did not change (80% and 10% for Nash and VE) otherwise it should be added in Table 3.
p. 19: Management scenario description is much clearer. Is there any reason to have chosen this scenario? Is it to have a significant response in term of modelled stream water quality? Or is it something actually discussed by the managers?
p. 21, 3.3.3 paragr. 1, “The load from DG... After storm events” What do you mean?
p. 21 last paragr., Addition of the uncertainty is appreciated!
p. 22, 3.3.2 Fig. 5 seems to show a small dephasing for SRP , could this be explained by some lack in the phasinf of modelled QSS?
p. 22, paragr. 3 : modelled event load is half of the one estimated from HFD. Could this be due to the non-simulation of event type different from 4? Or due to the time step?
p. 24, paragr. 2, line 13: “aN optimizing”?
p. 25, paragr. 1, lines 3-4-5: Be careful here. The model does not take into account the variability of the inputs and the transfer from one store to the other. In particular it does not take into account the memory effect in the subsoil due to the microporosity. It won’t be able to reproduce the past scenario with rising nitrate concentrations since the 1940s.
p. 25, paragr 1, Line 10-11 “The MI scenario…” It is because in the scenario, inputs are supposed to be reduced!!! With same inputs, reduction of overland flow could lead to increase leaching and enrichment of the SS S and DG stores, so should their concentrations increase as well (which is not the case in the model). Maybe it is not the case on this particular catchment as precipitation are high enough to flush all the surface store in a year, but under more limited infiltration context, this could accentuate the leaching of nutrient in deep stores.
Table 2: KSSF should be KSS?
Figure 4: the scheme still needs some improvements:
-inputs are represented while they are not injected in the model.
-outputs representation is a bit strange; concentrations would be more relevant from my point of view.