the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 02 Sep 2024
Delayed Stormflow Generation in a Semi-humid Forested Watershed Controlled by Soil Water Storage and Groundwater
Abstract. An analysis by Cui et al. (2024) of stormflow responses to rainfall in a mountainous forested watershed in the semi-humid regions of North China identified a distinct threshold for bimodal rainfall-runoff events, where delayed stormflow appeared to be influenced by shallow groundwater. This study further investigates the processes driving these bimodal events, focusing on the dynamics of soil water content (SWC) and groundwater level (GWL) during storm events. The results show that delayed stormflow is governed by the interplay between SWC and GWL. Delayed stormflow is initiated when SWC exceeds the soil’s water storage capacity, while its timing and volume are determined by GWL fluctuations. During rainfall, SWC increases rapidly; if it does not reach the soil's water-holding capacity, it stabilizes after the rainfall ends. Conversely, if SWC surpasses the soil's storage capacity, it decreases rapidly post-rainfall, with the excess rainwater infiltrating deeper to recharge groundwater, leading to a gradual rise in GWL. As GWL rises, increased hydraulic conductivity facilitates the movement of shallow groundwater into the stream channel, resulting in delayed stormflow. Simultaneously, the effective connection area between the stream channel and adjacent hillslopes expands vertically. At specific high GWL thresholds, GWL responses across the watershed converge, significantly increasing groundwater discharge and reducing lag times, often causing the delayed stormflow peak to merge with the direct stormflow peak. These findings enhance our understanding of delayed stormflow generation in similar regions and contribute to refining runoff generation theories.
- Preprint
(2031 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2177', Anonymous Referee #1, 08 Nov 2024
This is a timely manuscript on the bimodal response of semi-arid catchments. This work is novel in that there is very limited work looking into the bimodal response in catchments and the threshold behaviours of these responses. While the authors provide a great deal of information on the processes and do a good job, there a few areas that need some work. Once done, it should be an excellent contribution to the field. My comments are attached in a separate file. My recommendation is major revisions.
-
RC2: 'Comment on egusphere-2024-2177', Anonymous Referee #2, 14 Nov 2024
This manuscript examines the runoff generation processes leading to delayed peaks and proposes a conceptual model for the Xitaizi Experimental Watershed (XEW), North China. Overall, this paper could have the potential to contribute to the literature, but I think improvements in the way of presenting the research questions, the results and discussion would help to highlight the unique aspects of this work.
In my opinion, the manuscript is strongly linked to Cui et al (2024), which has recently been published in HESS and addresses the “characteristics and occurrence” of bi-modal events in the same catchment (https://doi.org/10.5194/hess-28-3613-2024). This is a personal opinion, but I do not understand the strategy of publishing two independent papers instead of summarising key results in one. I understand the same dataset has been used in both and methods descriptions are very similar (or the same). The authors mention that the data will be made available in Zenodo at the time of publication. Does this mean that the data is different than https://zenodo.org/records/12581739? If yes, it would have been nice to make the data available.
I think the introduction could present better the significant amount of literature where the role of soil water content and groundwater levels in the generation of delayed peaks (and its timing) has been explored, including catchments in Japan, central Europe, UK, USA, as well as Africa and New Zealand. The reader should be better informed about what is already known and why the presented work in needed. The reader could understand from the introduction that the mechanisms and thresholds have not been previously investigated. For instance, I disagree with the statement in lines 65-66.
I also had some problems to understand some of the methods. The XEW is a relatively small catchment (4.22 km2) with quick reaction times (e.g. Figure 11), why did you average the 5-min data (e.g. discharge) or 10-min data (soil water content) to hourly values? I assume by smoothing the data you might be losing significant information when looking at reaction times (which is a significant part of the presented work). Did you check that out? Did you loose two years of discharge data due to environmental challenges? Maybe there was another reason for this.
Soil water content probes were installed at two sites: “five sensors installed in Hillslope 1 and three near WS900 at 80 cm depth intervals”. At which depths? Installing them at “80 cm depth intervals” seems impossible if soils are 1.5 m depth. How were probes installed? Which was the data variability? How and why were the locations chosen and how do they represent what is happening in the catchment? Data was “aggregated to hourly intervals, and the arithmetic mean SWC across the profiles was used for analysis”. Why wasn’t the response of different layers investigated? This seems a lot of averaging to me and we have no clue of data variability across the 8 sites.
I also would like to have more information about how the groundwater data has been treated. The authors mention that the data of each well has been normalised using the Detty and McGuire (2010) method to normalise groundwater levels using an index (IG) calculated for each borehole. To my understanding Detty and McGuire did not use any index for normalisation: “For each well and event, we calculated the median height of the water table above the lowest recordable depth of each instrument and normalized that value to the total range of heights observed throughout the study period at each well (0 D minimum observed height or lowest recordable depth, 1 D maximum observed height, referred to hereafter as ‘normalized’).” I am not sure you are referring to this. Why were different hillslopes instrumented? Which is the logic behind the location of the equipment. The authors then calculated the arithmetic mean of the index to represent the overall groundwater level in the watershed. It is very difficult to address the implications of this as we do not know how the data looks like (I understand all plots show average data), but averaging data from all wells where there is water is a very simplistic approach and the authors should provide evidence that it is not.
The authors conclude that “delayed stormflow is initiated when soil water content reaches field capacity”. However, if I am not mistaken, there is no prediction of field capacity in the manuscript. This leads me to conclude that one of the ‘key poitns’ of the paper is not supported by data. I agree that the concept of field capacity, by definition, is not a static physical soil property. It also varies with depth. It can be determined in many ways, but it would have appreciated to have seen this addressed.
The authors selected events using an algorithm described by Tian et al (2012) – maybe a bit more information could be given. Separation seems to be exclusively based on rainfall patterns. My experience is that this type of algorithms can detect first peaks, but that they are not suited to investigate delayed flows. This because after a given event, other events can happen while baseflow is rising or falling (what would be delayed flow). I understanding that the authors identified 14 events when after an event there was not other events, resulting in nicely drawn delayed peaks. I do not see a problem with this, but there is no explanation about how the single events have been separated from the events with delayed peaks, what poses a fundamental problem for me to understand what has been done. Also, while reading the paper I kept wondering how the events would look like. I really miss hydrological data in the paper – as all the figures show processed/averaged data, or schematic figures (e.g. figures 6 and 7). I saw afterwards that there is an Appendix. This could have been mentioned (was it?).
The HYSEP program is used to separate baseflow from stormflow, with “manual verification and adjustment based on straight line separation methods”. Do you mean the constant slope method of Hewlett and Hibbert (1967)? I am not sure this data is used in the catchment and how does it compare to the tracer-based hydrograph separation carried out in Cui et al (2024). When you refer to event’s stormflow along the manuscript, do you refer to the discharge minus baseflow? This should be clarified.
The authors define thresholds in a very arbitrary way. For instance, the 0.20 threshold described in Figure 5 (lines 207-2012). Is this only a visual exploration? Was there a statistical way to define this threshold?
The structure of the manuscript is puzzling. There are three sections in the results, which include discussion and comparison with the literature (what should be moved to the discussion section). On the other hand, new results are presented in the discussion section.
Too little is said about the thick regolith, I think more information is needed here and it what would be very useful to understand the behaviour of the catchment. For instance, soils are described as “brown earth and cinnamon types”. A bit more information would be appreciated here. Also, at some point the authors argue that different groundwater dynamics in different hillslopes are due to specific hillslope’s geological structures. This should be further explored in the discussion (not the results section).
MINOR COMMENTS
- Line 29: not clear what you mean by ‘expands vertically’.
- Line 39: the catchment is 4.22 km2: what do you mean by flooding? I would use another term.
- Line 68: this sentence repeats the same as line 57.
- Line 69-70: give some examples of studies where they fail to do so and the reasons. I am not sure I agree with this.
- Line 75: the authors
- Line 76: low.
- Line 76-78: It is stated that analysis of 15 bi modal events collected during a decade have already been analysed and contributed to the advancement of runoff generation studies. Maybe it would be nice to summaries this in the introduction. Or do you refer to the work presented in the manuscript?
- Line 103: I would indicate there are 5 stations also here in text.
- Line 112: data covering two complete years?
- Line 112: data was lost during 2 years because of ‘environmental reasons’? This is not clear.
- Line 117: why did you aggregate the data?
- Line 127: I think the approach should be shortly described here.
- Line 188: “among these” reads confusing as you are not refering to the previous sentence.
- Line 226-228: I think this is rather an opinion and should be discussed in the discussion section.
- Lines 251-252: I would remove as a summary of previous section should not be needed.
- Line 286: why HS3 compared to HS1 and HS2.
- Line 295: replacing c?
- Figure 1. The exact same figure is used in Che et al. (2024, HESS). I wonder if this allowed without referring t the first figure published. It is difficult to see the location of the weather stations. Where are the five soil water profiles located? Are this indicated as “research hillslopes”? or what are research hillslopes? The authors refer to Hillslope 1 in line 116 - but not to the others. An explanation is missing.
- Lines 31-33. The authors conclude that their fundings “enhance our understanding of delayed stormflow generation in similar regions”. I think it would be nice to better explain this. Where? Why?
- I understand section 3.3 refers to the 14 selected events, is that right?
- Figure 8. It would be nice to have a little map displaying the location of the wells.
- Figure 9 is nice but difficult to understand with the little information we have about the catchment.
- Figure 10: I understand there are two points per event in that graph. Would be nice to know which points refer to Ts1-ts3 and which to ts2-ts3. I wonder if it is correct to use these two points per event to draw a regression line. The x axis indicates that there is 10 days difference between the reaction in one well and another. I do not understand this and I think the paper do not provide enough evidence to the reader to show what is going on. Why the others wells were not included in the analysis?
Citation: https://doi.org/10.5194/egusphere-2024-2177-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 133 | 46 | 14 | 193 | 1 | 2 |
- HTML: 133
- PDF: 46
- XML: 14
- Total: 193
- BibTeX: 1
- EndNote: 2
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1