Soil moisture and streamflow deficit anomaly index: an approach to quantify drought hazards by combining deficit and anomaly

Popat, Eklavyya; Döll, Petra

doi:https://doi.org/10.5194/nhess-21-1337-2021

Articles | Volume 21, issue 5

https://doi.org/10.5194/nhess-21-1337-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

https://doi.org/10.5194/nhess-21-1337-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 21, issue 5

Research article

|

03 May 2021

Research article |

| 03 May 2021

Soil moisture and streamflow deficit anomaly index: an approach to quantify drought hazards by combining deficit and anomaly

Eklavyya Popat and Petra Döll

Download

Final revised paper (published on 03 May 2021)
Supplement to the final revised paper
Preprint (discussion started on 19 Aug 2020)
Supplement to the preprint

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

SC1: 'Discussion', Alban Kuriqi, 19 Aug 2020
RC1: 'Review of the manuscript NHESS-2020-265', Anonymous Referee #1, 05 Nov 2020
- AC1: 'Answer to comments of Anonymous Referee #1', Eklavyya Popat, 17 Dec 2020
- AC3: 'Additional changes of manuscript', Eklavyya Popat, 17 Dec 2020
SC2: 'Opinion on manuscript', Fabio Micale, 05 Nov 2020
- AC4: 'Repy to opinion on manuscript', Eklavyya Popat, 17 Dec 2020
RC2: 'Review', Anonymous Referee #2, 12 Nov 2020
- AC2: 'Answer to comments of Anonymous Referee #2', Eklavyya Popat, 17 Dec 2020
- AC3: 'Additional changes of manuscript', Eklavyya Popat, 17 Dec 2020

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (further review by editor and referees) (19 Dec 2020) by Carmelo Cammalleri

AR by Eklavyya Popat on behalf of the Authors (21 Dec 2020) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (05 Jan 2021) by Carmelo Cammalleri

RR by Anonymous Referee #2 (15 Jan 2021)

RR by Anonymous Referee #1 (21 Jan 2021)

Suggestions for revision or reasons for rejection

General Comment

I would like to thank the authors for carefully considering my comments and revisiting the manuscript accordingly. Even if I found the new version significantly improved compared to the original version, I still have some major concerns on some details of the methodology, as well as on the way the results are presented, especially in the framework of a work on global scale drought.

The authors introduced some simplifications from the original formulation of the DSI, but in my opinion they fail in highlighting if/how these simplifications impact on the behavior of the index.
In my understanding, the two main differences are: 1) the d index is compute on the Smax, and 2) the computation of p is performed differently.
From the reported results I see a very high fraction of data with high d values (Fig. 2), and I am wondering how much of this is related to the adopted modification.
In term of p, the new Figure S1 show to me a quite different behavior compared to DSI (e.g. for F = 0.87 P is 0.1 and 0.4, respectively).
Similarly, Fig. 1 is supposed to show two contrasting examples, as done in Cammalleri et al. for DSI. However, these two cases do not seem opposite example at first glance. If you look at Fig. 3 in the DSI paper, in one case the DSI resembles d while in the other resembles p. In your analogous figure, both cases resemble p. This means that either you selected two cases that are too similar or that your simplifications do a disservice to the index. Please provide better examples or clarify.

Following my main concern in the first round of review, I still found the section on results quite lacking in the context of a global study. Too much emphasis is given to a specific arbitrarily-selected month (August 2003), and the addition of few figures in the supplementary materials does not alleviate the issue. I strongly suggest to the authors to reshape the approach adopted to show the results, in a way that better highlight the results during relevant droughts.
As you stated, most of the globe is likely to be in no drought during any given month, so it is meaningless to show the behavior of the index during such period.
I much prefer the approach adopted in Fig. 9 for the EFR, and I suggest to expand this event-based approach to the other analyses as well. This is valid for both indices.
Also, at the moment there is no clear distinction between the “results” and the “discussion” section, with both containing what can be called results. The discussion section should discuss the outcomes reported in the results section, not adding new outcomes. Please improve this structure, even by simply merging the two into a single results and discussion section and homogenize.

Overall, I see a lot of potentiality in this paper, and a large amount of valuable data that can really help the drought community in improving the understanding of both soil moisture and river droughts, but I think that these data need to be better presented to the readers to transfer the right message.

Specific Comments

L8-9. remove “the condition of”
L11. Please add full reference to the DSI in the abstract.
L12. Too many details for an abstract. Please reword “… is based on… mapping function” as something in the line “…as a simplified version of the DSI (ref)”.
L39. I think that “can be used” is not necessary here.
L98. outputs.
L100. Please add “:” after “five sectors”.
L109-110. I suggest to reword this sentence. At first glance, it seems that “source of water abstraction” is not accounted, and not that this is done by a different module.
L131-133. It is important to mention specific validations made on low-flow/drought (if any). It is well-known that performances on the lower spectrum of the flow regime may differ quite significantly from the average/high regimes. In absence of specific validations on drought, it is worth mentioning this other possible source of error.
L135-136. This sentence needs some rewording in my opinion.
L144. Here I see a major difference between this approach and the one proposed in DSI. If I understand correctly, in DSI the critical water content is used (50% of filed capacity) rather than the field capacity itself. Indeed, the absence of water stress starts in many cases well before field capacity in reached. I suggest to make this difference much more clear (see also discussion 4.1) and to clarify the impact on the high d values observed over most of the globe in fig. 1.
L211-216. This is a quite key point, that needs to be stressed more. It would be interesting to have some more insight on the differences with anomalies computed on the deficit, even if no detailed analysis is provided.
L245. It would be more consistent to use the ECDF for all the cells, especially since you are not using the mode as reference (which needs to be derived from the theoretical distribution), even if I guess that there is not much difference in the case of a good fitting. I also suggest to integrate figure S4 here rather than as supplementary material.
L252. The relationship between dsoil and what? Please rephrase.
L264. “very high soil moisture saturation”. Please reword.
L286. Still, very surprising that most of Canada during December 1999 is in high water stress. The high fraction of the world with dsoil > 0.75 is very surprising in general. This need a clear explanation (it seems to happen also in December 1999). Also, I do not think that the analysis of a single month (out of 30+ years) is enough here. You need to come up with a synthetic map that summarize the whole period (or the major events for different areas), not just a single case (see major comments).
L301-302. Is it realistic to call this soil moisture drought, when most of the deficit is due to snow fall rather than liquid precipitation (hence, the soil is covered in snow)? I see this more of a problem for the indicator rather than a desired behavior. Example: How is a good thing that Australia and Canada behave the same?
L308-310. this is an unnecessary introduction.
L337. The effect of EFR in defining a drought is quite important and needs to be better highlighted in my opinion. At the moment, relegating this analysis to a supplementary material does not give justice to a really key point of transferring this concept from soil moisture to stream flow.
L337-340. I am not following this argument. It seems to me that QDAI is rather similar to p in the cell over US, and I do not see any major clear differences between the two sites in term of “strength” of the droughts. Please elaborate better this concept.
L352. “…is mostly smaller than less than…”
L354. Cells with recurring zero-flow should be treaded differently, since the deficit (how low is Q compared to the historical data) cannot be used as reliable quantity for drought. The length of dry spells is considered a much better proxy here. I suggest to better clarify that such areas need to be masked from the analysis (as successively discussed in L365-366). Please also use a different color for these areas, since the current color is too similar to the dark red used for extreme drought (it is the case also for Fig. 6, see int vs. high frequency).
L361. “…no-drought conditions according to QDAI (Figure 6)…”
L399. A clear definition of (semi)arid and humid is needed in the methodology. Also, what about the other climates? Where all the cells classified as either of the two, and how?
L407. I do not really see any major differences between the two months, which is also kind of expected if you do a global average. Differences can be related only to the fact that there is more land in the northern hemisphere compared to the southern. Again, I see more useful an analysis on the full dataset (or specific drought events) rather than 2 randomly selected months that give very similar outcomes.
L449. These two examples need to be better highlighted in Figure 10. For what I can see, even SSFI has no drought at the end of 2005, or maybe you are referring to end of 2004. What I see is two drought periods in SSFI across the 2004 and 2005 lines. Also, what about the other two cases? No examples where QDAI improves on SSFI?
4.4 I miss the role of this section, which again focuses only on a single point and a single case. Even on the specific case, what is the message that you are trying to pass here on this known phenomenon?
Fig. 4. There are some inconsistencies in the plots. The legend of the plots on the bottom line seems to be off, also according to the text (d should be in green and p in blue). Also, in the top-left panel Qant seems to be always below Qant_mean, which suggest to me that the opposite occurs before 2000. This can be related to a trend in the data, which should be highlighted and discussed.

Hide

ED: Reconsider after major revisions (further review by editor and referees) (21 Jan 2021) by Carmelo Cammalleri

AR by Eklavyya Popat on behalf of the Authors (02 Mar 2021) Author's response Manuscript

ED: Publish subject to minor revisions (review by editor) (07 Mar 2021) by Carmelo Cammalleri

Dear Authors

The reviewers are overall satisfied by your replies. Please include the following minor corrections to the manuscript: I will consider the publication after these comments are integrated.

L11. Please add reference to the DSI. The abstract should stand independently of the rest of the text (see guideline of NHESS).

L25 (BoN, 2018) should be outside the quotation marks.

L55-56. “…it indicates…(Dai et al., 2004)”. Please reword the sentence, which is not well structured at the moment.

L72. Differently…

L80. (That which should…) Please rephrase.

L91. I suggest to invert (a) and (b), since this is the order in which the two sub-section are then discussed successively.

L98-108. Please clarify which component is “static”, if any (i.e. do not account for inter-annual or intra-annual fluctuations).

L149. Cammalleri et al. (2016) scale soil moisture between “wilting point” and “critical point”, not saturation. Please clarify.

L151. Please remove “saturation”.

L153. “…while at high saturation they are higher”. This is surprising to me, since Cammalleri et al. define d=1 at critical point, which is much lower than saturation. I would expect to have much less d=1 in you approach as well (similar to zeros). Please clarify.

L211-L214. Please simplify and remove eq. (5). This can read as simple as “EFR is calculated for each calendar month as80% of mean monthly streamflow…, assuming…”

L258. Please add few lines specifying the locations where the ECDF is used (i.e. Australia, south Africa, etc.). Also, looking at figure S4, it seams to me that ECDF is used in more than 1/3 of the areas. Please check the values reported in the text.

L274. Please remove “saturation”. You can use soil water content instead of soil moisture.

L283. can be further explored.

L284. This high percentage needs to be discussed. Possible reasons? It is also worth to highlight that SMDAI resemble psoil in this case only because dsoil is quite high everywhere. This is not always the case.

L321. “…on more individual variable…”. Please add “(i.e…..)”. I thin it is worth a further reminder for the readers on the variables that play a role.
L342. I suggest to replace “sensible” to ‘reasonable”.

L348-L353. This paragraph seems a little out of place to me, since a dedicated section to this topic is successively presented. I suggest to shrink this paragraph, and reference the successive analyses.

L378. “if for many….”. Please check this sentence, which is currently unclear.

L396-397. Please rephrase and remove the incorrect term “more amount of soil moisture…”.

L397-398. “Differences are mostly small…”. Is this a sign that d is not that sensitive the absolute changes in Smax? So, for higher Smax also S will be higher? If this is the case, I think it is worth to be highlighted.

L408. I would say “defined” rather than “computed”.

L445-455. This description needs to be moved to the methodology section. Here you should report only the result of the comparison, but not the description of the SSFI.

L474. As already highlighted in the previous review rounds, this analysis is a little lacking and not fundamental for the goal of the paper. I suggest to remove this section, and correct the text (e.g. abstract) accordingly.

L508. outputs.

L509. …but also on the concurrent deficit….

L520. Please expand a little on these “additional indicators”. Are you referring to more detailed information on water use?

L528. data are available. Please, also add a sentence of the availability of the outputs of this study (SMDAI and QDAI).

Hide

AR by Eklavyya Popat on behalf of the Authors (15 Mar 2021) Author's response Author's tracked changes Manuscript

ED: Publish subject to technical corrections (23 Mar 2021) by Carmelo Cammalleri

AR by Eklavyya Popat on behalf of the Authors (29 Mar 2021) Author's response Manuscript

Download

Article (7856 KB)
Full-text XML

Short summary

Two drought hazard indices are presented that combine drought deficit and anomaly aspects: one for soil moisture drought (SMDAI) where we simplified the DSI and the other for streamflow drought (QDAI), which is to our knowledge the first ever deficit anomaly drought index including surface water demand. Both indices are tested at the global scale with WaterGAP 2.2d outputs, providing more differentiated spatial and temporal patterns distinguishing the actual degree of respective drought hazard.