Hydro-stochastic interpolation coupling with the Budyko approach for prediction of mean annual runoff

Qiu, Ning; Chen, Xi; Hu, Qi; Liu, Jintao; Huang, Richao; Gao, Man

doi:https://doi.org/10.5194/hess-22-2891-2018

Articles | Volume 22, issue 5

https://doi.org/10.5194/hess-22-2891-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-22-2891-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 22, issue 5

Research article

|

16 May 2018

Research article |

| 16 May 2018

Hydro-stochastic interpolation coupling with the Budyko approach for prediction of mean annual runoff

Ning Qiu, Xi Chen, Qi Hu, Jintao Liu, Richao Huang, and Man Gao

Download

Final revised paper (published on 16 May 2018)
Preprint (discussion started on 11 Sep 2017)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Review on "Hydro-stochastic interpolation coupling with Budyko approach for spatial predictions of mean annual runoff" by Ning Qui et al.', Mirko Mälicke, 27 Oct 2017
- AC1: 'Response to RC1 by Mirko Mälicke', Xi Chen, 07 Dec 2017
RC2: 'Hydro-stochastic interpolation coupling with Budyko approach for spatial prediction of mean annual runoff', Jon Olav Skøien, 07 Dec 2017
- AC2: 'Response to RC2 by Jon Olav Skøien', Xi Chen, 14 Jan 2018

Peer-review completion

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

ED: Reconsider after major revisions (further review by editor and referees) (18 Dec 2017) by Erwin Zehe

AR by Xi Chen on behalf of the Authors (14 Jan 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (17 Jan 2018) by Erwin Zehe

RR by Mirko Mälicke (13 Feb 2018)

RR by Jon Olav Skøien (13 Mar 2018)

Suggestions for revision or reasons for rejection

This is just a short review, focusing on one concern I have with this version of the manuscript. As I have been in touch with the authors directly (with the editor in copy), I know that they have already prepared a new and improved version, which they will upload as soon as this review round is finished. This review will therefore only address this concern, and I will later give a more detailed review report based on the new version.

In this revision, the authors included the observations in Figure 7 and made sure that the color keys were the same for all figures. I was then very puzzled by the poor results of the hydro-stochastic interpolation alone (Fig 7b). Having worked quite a bit with this type of interpolation myself, I think the method should be able to achieve better results than what can be seen in this figure, and that the difference between the methods might be too large.

I am not questioning that a combined method can be able to achieve better results than the two models individually. However, when both methods give rather poor results (as in Fig 7a and 7b) then it is surprising to see such a large improvement as in Fig 7c, particularly as the two methods seem to overestimate and underestimate at the same locations. I think the interpolation worked better for the residuals, but it would be good to know if this is because the residuals are actually easier to interpolate, or if this is because the interpolation of raw data failed for some other reason. If it is the first case, this should be better explained in the manuscript, as it would point out a weakness with these type of interpolation methods. If it is the second case, then the interpolation of raw data would get a better result, but still most likely not as good as the combined method, and most of the manuscript can be left as it is.

Hide

ED: Publish subject to revisions (further review by editor and referees) (13 Mar 2018) by Erwin Zehe

AR by Xi Chen on behalf of the Authors (14 Mar 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (15 Mar 2018) by Erwin Zehe

RR by Jon Olav Skøien (06 Apr 2018)

Suggestions for revision or reasons for rejection

This is a new version of a manuscript describing a coupling of the Budyko approach and hydro-stochastic interpolation. The manuscript has been improved, but there are still a few issues. I also still have some questions regarding the hydro-stochastic interpolation. I am suggesting minor revision here, as there might be good answers to my questions, in that case the editor can accept the manuscript after receiving these. If he is not satisfied with the answers, I’m happy to have another look at a new version.

I am happy to see that the new version gives better and more sensible results for the hydro-stochastic interpolation. However, in the answer to the previous review, the authors added the weights for one of the stations. These weights puzzled me, and needs some explanation, by checking closer how the weights could occur.
The weights are for the HWH catchment, which is in the bottom part of the domain. The two closest catchments (BLY and a catchment where I cannot see the name) are not even mentioned in the list of weights. QL and HNZ are quite highly correlated, but still get large weights both of them. This is rather weird, even if hydro-stochastic interpolation is depending on the configuration of the observations and their spatial supports in addition to the distance between them.

Does the software have a correction for negative weights? That could be an explanation, but then it needs to be mentioned. Still it is more common to see the observations "in the shade" getting negative weights. By that I mean that I would expect a relatively large weight for BLY and the unnamed catchments, and negative weights for QL and/or HNZ. If negative weights is the cause, it is necessary to check how large they are before rescaling. Just deleting them and rescaling the rest might not be the best solution, it might be better to pick fewer neighbours for the interpolation in the first place.

Minor
Section 2.2 is still too long – 3 pages is not necessary for methodologies that have been published before.

In P16, it is mentioned that the area is divided into a 40x50 grid. I find this relatively course resolution for many of the smaller sub-basins, which are likely to get rather few grid points for the calculation of the Ghosh-distance (not geostatistical distance), if I understand correct.

Section 5 is very much a repetition of the summary from 4.4. Some more discussion would instead be useful, such as how does the results in this paper compare to similar studies before.

Edits
There are still many grammatical errors – often related to articles and plural/singular. Already in the start of the abstract, it should either read “A Hydro-stochastic interpolation method…” or maybe better: “Hydro-stochastic interpolation methodS based on traditional block kriging HAVE often … A caveat in such methodS ARE that…” There are several more examples in the manuscript, such as P5L84 – such AN approach. P6 L114-119 “THE semi-empirical …” “into A deterministic trend” “calculated as THE difference …”

P4L63 Blöschl
P4L65-66 I find it unclear what this sentence really says. It should anyway be constrains.
P4L83 What is an “integrated course”? Rephrase
P5L80 I think there are more important reasons for the deviations than the influence FROM heterogeneous rainfall.
P5L85 “method FOR describing complex runoff patterns suffers FROM AN inevitable …”
P5L90-93 KED is not recent – maybe rather something like “A method that combines both deterministic patterns and stochastic variability is kriging … ” And then “It takes deterministic patterns of spatial variables into account and incorporates these as a local trend, a smoothly varying secondary variable, instead of a function of spatial coordinates.”
P6L117-118 comma after residuals, remove “both of”
P13L266 “millions of tons of water” – this is not very precise, and actually not a particularly large number for a region of this size. As an example, 10 million m3/year equals 0.3 m3/second.

Hide

ED: Publish subject to minor revisions (review by editor) (09 Apr 2018) by Erwin Zehe

AR by Xi Chen on behalf of the Authors (18 Apr 2018) Author's response Manuscript

ED: Publish as is (24 Apr 2018) by Erwin Zehe

AR by Xi Chen on behalf of the Authors (27 Apr 2018)

Short summary

The spatial runoff is decomposed into a deterministic trend and deviations from it caused by stochastic fluctuations which are described by Budyko method and stochastic interpolation. This coupled method is applied to spatially interpolate runoff in the Huaihe River basin of China. Results show that the coupled method reduces the error in overestimating low runoff and underestimating high runoff suffered by the other two methods, so it improves the prediction accuracy of the mean annual runoff.