Review Reports

Round 1

Reviewer 1 Report

Reconsider after major revision, comments in file attached

Comments for author File: Comments.pdf

Author Response

We appreciate the constructive comments provided by the reviewer. The manuscript has been thoroughly revised to address the reviewer’s suggestions. The detail of my edits is provided below each comment from the reviewer.

Author Response File: Author Response.docx

Reviewer 2 Report

 

GENERAL COMMENTS  

This paper compares several different agricultural strategies, in terms of their potential profitability and implications for water availability, in four watersheds in western Nepal in the next three decades in a changing climate. A central finding is that there are potentially more profitable alternatives for the current standard wet season rice – dry season wheat strategy, but that such an agricultural intensification is only hydrologically sustainable when both streamflow and groundwater are used for irrigation. 

A very considerable research effort has been put in this paper. Overall, most of the modelling work and data analysis appears to have been well done. The quality of the presentation is also reasonable, except for some minor problems in the English language. 

Unfortunately, there are severe limitations and apparent flaws in the climate change analysis. Point 2 below is especially serious. 

1)      Only one regional climate model simulation, using the NOAA_RegCM4 regional climate model, has been used to drive the hydrological model. This hampers the robustness of the results, since short-term projections of temperature and (in particular) precipitation change are heavily affected by internal climate variability. Furthermore, the resulting uncertainty is hidden when only one climate change scenario is used. This limitation should be expressed more clearly: the climate change scenario used in the study is just one of many plausible possibilities. 

2)      There is an apparent discontinuity in the used climatic time series between the “current” (2000-2021) and the two future (2022-2035 and 2036-2050) periods. This is most evident from Figure 4: during the dry season from October to April, the precipitation in the two later periods is several times larger than in 2000-2021, so that even the wettest months in 2000-2021 are drier (or nearly so) than the driest of the same calendar months during the later periods. Such a difference is far too large to result from the relatively modest forced climate change between 2000-2021 and the later periods, and it is also virtually impossible to get so large difference from random internal variability. The only viable explanation is that the data for 2000-2021 and the two later periods are methodologically inconsistent.  

Because the manuscript does not discuss in all detail how the climatic time series before and after 2021 were obtained, I cannot pinpoint the exact cause of the problem. However, it strongly suggests that the bias corrections that the authors are using do not work well on the monthly time scale, or that they are not applied correctly. One way to get this kind of results would be to use observations for the earlier period (2000-2021) but imperfectly bias-corrected RCM data for the later two periods. Another alternative is that the authors used the same RCM both before and after 2021, but the boundary data for the RCM were obtained from different sources (first from a reanalysis data set, but after 2021 from a global climate model). 

Whatever the precise cause of the problem, it invalidates those parts of the study where the results for 2000-2021 and the two future periods are compared.    

3)      There is an inconsistency between the annual Tmax and Tmin results reported in Figure 5 and the monthly values shown in Figure 6. The latter indicates systematically warmer temperatures in 2036-2050 than 2022-2035 in all months of the year (except for no change in February), whereas the annual time series in Fig. 5 suggest a surprising cooling trend from 2020 to 2050. Furthermore, even in the warmest summer months, the absolute values of Tmax and Tmin in Figure 6 are lower than the (annual average?) values reported in Figure 5. 

Unfortunately, problem 2 affects all those parts of the study (e.g., calculations of water availability and plant production) where comparison is made between 2000-2021 and the two later periods. The only good option would be to use methodologically consistent temperature and precipitation time series throughout the period 2000-2050. In practice, this means using a NOAA_RegCM4 simulation with boundary conditions taken from the same global climate model throughout the 2000-2050 period. If this is not feasible, an emergency solution could be to just use the trend information after 2021: (1) fit linear trends to all time series between 2022 and 2050; (2) report the best-estimate values that result from this trend line for the beginning and the end of the period. Clearly, however, such short-term trends may be affected heavily by internal climate variability. 

More detailed, mostly minor comments on the manuscript follow below. 

DETAILED COMMENTS / SUBSTANCE  

1. L17-18. This decrease in temperature seems surprising; see also General comment 3 above. In any case, temperature changes should be reported in degrees C, not in per cent.
2. L209. Which global climate model was used to give the boundary conditions for NOAA_RegCM4?
3. L211-213. A single climate model simulation gives only of many plausible realizations of climate change, with no uncertainty information. Only studying the results of a single model may therefore lead to unjustified confidence in the results.
4. L251-253. Is this trend statistically significant? It seems to be very small compared with the interannual variability.
5.  L276-277. The cooling shown in Fig. 5 is inconsistent with Fig. 6. Furthermore, even if Fig. 5 is correct, this cooling would be unexpected in the context of the ongoing global warming. In that case, its most plausible explanation is an unusual realization of internal variability in the model simulation. But as internal climate variability is largely unpredictable, the apparent cooling has no predictive value even if it is simulated by a climate model.

6. L278. If this cooling of maximum temperatures really occurs (which seems less likely than not!), this would reduce rather than increase problems associated with warm extremes.
7. Unlike Figure 5, Figure 6 suggest a warming from 2022-2035 to 2036-2050 in all months except for February. Furthermore, the average maximum and minimum temperatures given in Figure 5 are much lower than those in Figure 6, even in summer. Please check your calculations.
8. The numbers given on L325-328 are an order of magnitude smaller than the annual averages in Fig. 7. Why?
9. Table 2. Why are the yields much lower in 2021-2035 and 2036-2050 than in 2000-2020? This most likely has something to do with the apparent climate discontinuity discussed in General comment 2.
10. L442-443. Is this decrease real (cf. Fig. 6)?
11. L508-509. See the earlier comments. The purported decrease in maximum and minimum temperatures is an unlikely scenario.

DETAILED COMMENTS / LANGUAGE AND PRESENTATION

1. L13. and reducing / eliminating poverty?
2. L15. the Soil and Water Assessment Tool
3. L15-16. the climate change scenario
4. L37. months of June to September
5. L42-43. Since neither surface nor groundwater ...
6. L67-68. almost two thirds
7. L97. efficiently?
8. L102. over long periods, they are mostly used?
9. L106-107. Hydrologic models … have been applied
10. L122-123. slower compared with what? RCP8.5? The RCP4.5 scenario can be seen as a middle-of-line world, far from RCP8.5 but also well above RCP2.6 and other low scenarios that are compatible with the Paris agreement.
11. L131. irrigation schemes, and have frequently experienced
12. L141. India border, with one third of its drainage area in Nepal and two thirds in India?
13. L154-155. 3% to 10% of the total land area (or of the agricultural area)?
14. L165. months of June to September
15. L192. parameters such as
16. L197. Omit 65 from the end of the sentence.
17. L198. in [1]?
18. L201. [1] in the end of the sentence?
19. L203. a questionnaire survey
20. L216-217. I would omit "representing slower global warming and limited climate change". "a medium emission scenario" tells enough.
21. L219. the Climate Model data for hydrologic modelling (CMHyd) tool
22. L239. For each selected scenario
23. L271. 2000-2021, not 2020-2021.
24. L277. Please give the temperature trends in degrees C, not in %.
25. Figure 5. The scale for minimum temperature (18 to 30C) is three times as wide as that for the maximum temperature (26 to 30C). This gives the misleading impression that the variation in minimum temperatures is very small. Please use a scale of (e.g.) 18 to 22C for minimum temperature to eliminate this problem.
26. L288-290. Please give the temperature changes in degrees C, not in per cent. Furthermore, which months does this sentence refer to? Based on Figure 6, there is no change in February, but warming in all other months of the year.
27. Figure 6. Because a significant number of people suffer from red-green color blindness, I would recommend replacing the green lines with blue lines.
28. L312. Please also mention the current period (2000-2021) in the figure caption.
29. L324. The average streamflow during the current period (2000-2021)?
30. L337. if the surplus is conserved?
31. L385. Should be Figure 8.
32. L392. legume crops ... improve soil fertility
33. L398-399. only a negligible
34. L410-413. Please specify whether the results are for a certain watershed or averaged over all relevant watersheds.
35. L416. yield of vegetables
36. L420. Are these values averages over all watersheds or only one of them?
37. L428. lentils require / lentil requires
38. L460-461. comparison ... was also performed
39. L465. higher local price of lentil?
40. L466. by 118% relative to rice-wheat?
41. L486. The legend in the bottom-left panel of Fig. 9 is “rice-wheat-potato” rather than “rice-wheat-mung”. Which is correct?
42. L488. Assumptions and limitations
43. L518. resulted from temperature stress?

Author Response

We appreciate the constructive review and detailed substance/editorial comments provided by the reviewer. The manuscript has been thoroughly revised according to the reviewer’s suggestions. The detail of my edits is provided below each comment from the reviewer.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Accept in present form

Author Response

We would like to appreciate the kindness of the Reviewer for accepting the revised manuscript in its present form. The constructive critical comments from the reviewer have really helped in improving the quality of the manuscript.

Reviewer 2 Report

 GENERAL COMMENTS

I appreciate the effort that the authors have made in revising their manuscript. Nevertheless, I still feel that their climate change scenario is problematic, and the associated uncertainty should be discussed more explicitly.

First, as discussed in my first review, there is an abrupt change in the dry-season precipitation climate between the years 2000-2021 and 2022-2050. Based on the explanation that the authors give in their response, this is most likely because observations were used before 2021 and bias-corrected (but still to some extent biased) RCM data thereafter. For analysing the impacts of climate change, it would have been better to use bias-corrected RCM data throughout the 2000-2050 period. While this would have to some extent deteriorated the climate data in 2000-2021, it would have avoided a major artefact in the climate change signal.

Second, the cooling (or, at least, lack of warming) that the authors report and cite as a cause of increased temperature stress for crops differs from other climate model studies. While I have found nothing that would be exactly comparable with this study in terms of the time periods, the scientific consensus is that climate will become warmer (rather than cooler) in South Asia. This text is from the IPCC Working group sixth Assessment Report (available from https://www.ipcc.ch/report/ar6/wg1/), on p. 1981 in the Atlas chapter: 

“It is likely that surface temperatures over South Asia will increase more than the global average and more so over TIB, with projected increases of 4.6°C (3.4°C–6.0°C) during 2081–2100 compared with 1995–2014 under SSP5-8.5 and 1.3°C (0.7°C– 2.0°C) under SSP1-2.6 (Interactive Atlas). Summer monsoon precipitation in South Asia is likely to increase by the end of the 21st century while winter monsoons are projected to be drier. Over the same time periods CMIP6 models project an increase in annual precipitation in the range 14–36% under SSP5-8.5 and 0.4–16% under SSP1-2.6 (medium confidence).”

As another example, some results are shown for the CORDEX simulations in this article: 

https://www.sciencedirect.com/science/article/pii/S1674927817300436

(see, in particular, the climate changes from 1976-2005 to 2036-2065 in their Fig. 6). 

Thus, whatever the reason of the cooling (or lack of warming) in your climate scenario, it should be made clear that this result is unusual when compared to other research.

DETAILED COMMENTS 

1. L125. by a CORDEX RCM
2. L157. source in Nepal
3. L200. administrative regions
4. L318-319. ... winter season affects / may affect ....
5. L331-332. Which "average" maximum and minimum temperatures does Figure 5 represent? The maximum temperatures are far too high to be yearly averages of daily maxima, and the minimum temperatures are far too low to be yearly averages of daily minima.
6. L333. Repeated from the previous review: why would this will have an adverse effect crop yield? A decrease in the maximum temperatures combined with a slight increase in minimum temperatures would make the climate less extreme, which should rather improve than deteriorate the growing conditions.
7. L342. decrease by 1.1C - between which periods of time?
8. L360-361. The quoted numbers suggest a fairly stable temperature climate, with warming in some and cooling in other months. Does this really lead to a significant increase in temperature stress, and if so, is the problem caused by warming or cooling?
9. L576-585. I don't think the stability of the bias correction is a major issue, considering the relatively near-term climate futures studied in this work. However, the use of observations to drive the modelling system in the current period (2001-2021) but bias-corrected RCM data afterward does create a discontinuity, if and when the bias-corrected model output during the current period still differs from observations. For the analysis of climate change impacts, it would have been much better to drive SWAT with bias-corrected climate model data throughout the whole 2000-2050 period, even if this slightly deteriorates the results for the current period. Running SWAT first with observations for 2000-2021 and with RCM output thereafter makes it impossible to distinguish between genuine climate change effects and the artefacts that result from the change in the source of data.

Author Response

Response to Reviewer 2 comments (Round 2)

 

GENERAL COMMENTS

 

I appreciate the effort that the authors have made in revising their manuscript. Nevertheless, I still feel that their climate change scenario is problematic, and the associated uncertainty should be discussed more explicitly.

First, as discussed in my first review, there is an abrupt change in the dry-season precipitation climate between the years 2000-2021 and 2022-2050. Based on the explanation that the authors give in their response, this is most likely because observations were used before 2021 and bias-corrected (but still to some extent biased) RCM data thereafter. For analysing the impacts of climate change, it would have been better to use bias-corrected RCM data throughout the 2000-2050 period. While this would have to some extent deteriorated the climate data in 2000-2021, it would have avoided a major artefact in the climate change signal.

 

Response: We really appreciate the detailed review and constructive comments provided by the reviewer. This would clearly improve the quality of our manuscript. In the present study, however, we were unable to run SWAT in the suggested way. We opted to use observed weather data for 2000-2021 and bias-corrected RCM data from the NOAA_RegCM4 model for 2022-2050 since it was one of the most widely used model in the region. SWAT was run throughout 2000-2050 by combining observed and RCM output. The reasons why we used this approach include use of calibrated SWAT that used observed weather data, and reliability of observed data over historical RCM data obtained from indirect measurements. As we agree with the Reviewer’s concerns, we have added further detail to the ‘Assumptions and Limitations’ section of the manuscript in which we suggest making use of bias-corrected RCM data throughout the simulation period for future predictions.

 

Second, the cooling (or, at least, lack of warming) that the authors report and cite as a cause of increased temperature stress for crops differs from other climate model studies. While I have found nothing that would be exactly comparable with this study in terms of the time periods, the scientific consensus is that the climate will become warmer (rather than cooler) in South Asia. This text is from the IPCC Working group sixth Assessment Report (available from https://www.ipcc.ch/report/ar6/wg1/), on p. 1981 in the Atlas chapter: 

“It is likely that surface temperatures over South Asia will increase more than the global average and more so over TIB, with projected increases of 4.6°C (3.4°C–6.0°C) during 2081–2100 compared with 1995–2014 under SSP5-8.5 and 1.3°C (0.7°C– 2.0°C) under SSP1-2.6 (Interactive Atlas). Summer monsoon precipitation in South Asia is likely to increase by the end of the 21st century while winter monsoons are projected to be drier. Over the same time periods CMIP6 models project an increase in annual precipitation in the range 14–36% under SSP5-8.5 and 0.4–16% under SSP1-2.6 (medium confidence).”

As another example, some results are shown for the CORDEX simulations in this article: 

https://www.sciencedirect.com/science/article/pii/S1674927817300436

(see, in particular, the climate changes from 1976-2005 to 2036-2065 in their Fig. 6). 

Thus, whatever the reason of the cooling (or lack of warming) in your climate scenario, it should be made clear that this result is unusual when compared to another research.

 

Response: We appreciate the constructive comments provided by the reviewer and we agree with the reviewer’s point that surface temperature is likely to increase in future. The model we selected for this study however showed a slight decrease in future maximum temperature (Line 284-285, Figure 5). The use of a single climate is one of the limitations of this study and this has been discussed in the revised ‘Assumption and Limitations’ section of the manuscript as: “Previous studies have for example shown that surface temperatures over South Asia are projected to increase by more than the global average [66,67]. Nonetheless, the climate model we selected for this study showed a decrease in temperature. This is unusual when compared to previous research [66,67]”.

The result of the study would have been different if we had used ensemble mean of bias-corrected RCMs for the entire simulation period.” In our future climate change studies, we will consider using these.

 

DETAILED COMMENTS 

1. L125. by a CORDEX RCM

Response: Thank you. The manuscript has been revised to address the reviewer’s suggestion.

 

1. L157. source in Nepal

 

Response: Thank you. The manuscript has been revised to address the reviewer’s suggestion.

 

1. L200. administrative regions

 

Response: Thank you. The manuscript has been revised to address the reviewer’s suggestion.

 

1. L318-319. ... winter season affects / may affect ....

 

Response: Thank you. The manuscript has been revised to address the reviewer’s suggestion.

 

1. L331-332. Which "average" maximum and minimum temperatures does Figure 5 represent? The maximum temperatures are far too high to be yearly averages of daily maxima, and the minimum temperatures are far too low to be yearly averages of daily minima.

 

Response: Thank you. The word average has been removed to make the sentence clearer.

 

1. L333. Repeated from the previous review: why would this have an adverse effect crop yield? A decrease in the maximum temperatures combined with a slight increase in minimum temperatures would make the climate less extreme, which should rather improve than deteriorate the growing conditions.

 

Response: Thank you. The sentence regarding the adverse effect of temperature change on crop production has been removed from the manuscript.

 

1. L342. decrease by 1.1C - between which periods of time?

 

Response: Thank you. “The average monthly mean temperature decreased by 1.1 ˚C in February and increased by 0.5 ˚C in October during mid-term future compared to near-term future.”

 

1. L360-361. The quoted numbers suggest a fairly stable temperature climate, with warming in some and cooling in other months. Does this really lead to a significant increase in temperature stress, and if so, is the problem caused by warming or cooling?

 

Response: Thank you. The sentence regarding the adverse effect of temperature change on crop production has been removed from the manuscript.

 

1. L576-585. I don't think the stability of the bias correction is a major issue, considering the relatively near-term climate futures studied in this work. However, the use of observations to drive the modelling system in the current period (2001-2021) but bias-corrected RCM data afterward does create a discontinuity, if and when the bias-corrected model output during the current period still differs from observations. For the analysis of climate change impacts, it would have been much better to drive SWAT with bias-corrected climate model data throughout the whole 2000-2050 period, even if this slightly deteriorates the results for the current period. Running SWAT first with observations for 2000-2021 and with RCM output thereafter makes it impossible to distinguish between genuine climate change effects and the artefacts that result from the change in the source of data.

 

Response: Thank you. In the present study, SWAT was not run separately for 2000-2021 using observed climate data and 2022-2050 using bias-corrected RCM data. SWAT was run throughout 2000-2050 by combining observed and RCM output. The reasons why we used this approach include the use of calibrated SWAT that used observed weather data and the reliability of observed data over historical RCM data obtained from indirect measurements. As we agree with the Reviewer’s concerns, we have added further detail to the ‘Assumptions and Limitations’ section of the manuscript in which we suggest making use of bias-corrected RCM data throughout the simulation period for future predictions.

 

 

 

Author Response File: Author Response.docx