the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 May 2026
Sediment transport capacity in a large braided river: integrating substrate mapping with flow scenarios
Abstract. Large braided rivers offer a challenging field environment and while our ability to map these complex environments through remote sensing has advanced significantly, even detailed surveys represent only a moment in time. In this study we have linked the remotely-sensed observations of substrate, including deposited fine sediment, at reach-scales to a hydraulic model of the braided Rangitata River, New Zealand. Excess fine sediment can alter fluvial form, ecosystem health and groundwater recharge. The advent of large-scale mapping technologies has enabled reconstruction of fluvial substrate facies over broad scales, and now spurs an examination of what can be learned with the greater fidelity of data from these surveys compared to simpler renditions of the riverbed. Here we propose a hybrid methodology that links spatially-distributed models of bed substrate with river hydraulics reconstructed from a ‘model library’ of steady state flows to better infer the controls on sediment transfer and deposition in braided rivers.
In this paper, we map potential sediment transport and depositional environments over a range of naturalised and modified flows. Our findings show that spatial bed data is the key to longitudinal consistency in sediment transport. The ‘model library’ method allowed us to efficiently test the tendency and magnitude of deficit, equilibrium, or surplus sand capacity in any location and the sensitivity of the result to bed composition, hiding and exposure formulations. The results indicate that the bed is sensitive to contemporary changes in hydrologic regime, particularly in the side channels accessible during the common ‘moderate’ flows that occur for tens of days per year. Simulations comparing the river’s sand transport capacity under the present hydrologic regime with a naturalised hydrologic regime indicate that the impacts of the flow abstractions are comparable to a 10–15 % change in the bed sand fraction. Maps of potential sand deficit align well with observed depositional areas, and highlight the critical importance of the ‘moderate’ flows.
- Preprint
(4482 KB) - Metadata XML
-
Supplement
(1121 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2149', Anonymous Referee #1, 27 Aug 2026
-
RC2: 'Comment on egusphere-2026-2149', Anonymous Referee #2, 27 Sep 2026
This manuscript couples a remotely sensed, facies-based map of bed composition to a precomputed library of steady-state 2D hydraulic simulations to estimate bedload and suspended sand transport capacity over a 56 km reach of the braided Rangitata River, comparing present-day (abstracted) and naturalised flow regimes. This represents an impressive data-collection and analytical effort. The flow-library approach is validated against event simulations, and the 432-run sensitivity programme on bed parameterisation is exceptional. The central finding - that abstraction shifts the discharge band of effective sand transport and creates capacity deficits in side channels at intermediate flows, with an effect comparable to a 10–15% change in bed sand fraction - will interest both the modelling community and freshwater managers.
I think the substantive issues to resolve before publication are (1) the supporting points for the claim that predicted sand deficits "align well with observed depositional areas", (2) the asymmetric treatment of supply-side versus bed-side uncertainty, and (3) two equation errors (Eqns 2 and 19). There is a small block of misplaced text and several internal numerical inconsistencies are detailed below.
General comments
1. The abstract and conclusions state that "maps of potential sand deficit align well with observed depositional areas", but this comparison is never presented as a figure or statistic, and the claim is somewhat circular as framed: the "observed depositional areas" are the 2021 fines facies map (Fig. 3c), which also enters the model directly through the local sand fraction Fs, so some correspondence is expected by construction. I encourage the authors to make the test explicit, noting that this need not involve re-running the 2D hydraulics: the capacity calculation can be determined with the stored flow library, and the sensitivity suite already contains fixed-Fs runs (Table 2; Fs = 1–50%). A deficit map derived from one of these, compared against the mapped fines distribution, makes the observation independent of the input. Alternatively, compare the longitudinal profile of predicted deficit against the observed downstream trend in sand cover (Fig. 4), a scale at which the prediction no longer inherits the input Fs map cell by cell. This could also be compared against independent observations: the ESNZ pebble-count and bulk-sample data (L245–247), or the anecdotally reported deposition sites that motivate the study (L120), tabulated against predicted deficit zones. If none of these is feasible, the prominence of the deposition maps and the associated claims should be reduced. With validation, these maps would be among the most valuable products of the framework. This bears directly on the motivating research question and is the most important revision.2. The methods carefully distinguish transport capacity from actual transport (L295–298), but this framing changes slightly in places: the abstract and conclusions state that "the bed is sensitive to contemporary changes in hydrologic regime". The static-bed model demonstrates that capacity is sensitive; bed response is an inference. Please reword so the demonstrated result and the inferred consequence are clearly separated.
3. Every deficit or surplus statement is a comparison between capacity and supply, yet the two sides receive different treatment: the bed side is explored through 432 sensitivity runs, while the supply side rests on a two-segment rating curve, the Hicks et al. (2011) load estimate, and a single representative sand fraction of 40% drawn from observations spanning 27–50% with evident event dependence (L180–185). Even a simple envelope - 30% versus 50% suspended sand fraction, or rating-curve confidence bounds - propagated through Figs 8, 11 and 12 would indicate whether the 120–600 m3s-1 deficit band is robust to supply uncertainty, and would considerably strengthen the deficit conclusions.
4. Alternative mechanisms for the downstream increase in sand cover. Section 5.2’s conclusion that "the river’s ability to transport sand increases downstream because the bed fraction of sand also increases downstream" follows by construction - sand capacity scales with Fs in the transport formulation - so the substantive question is why Fs increases downstream. A brief canvas of complementary mechanisms - in-reach abrasion of greywacke gravels, sand inputs from terraces and legacy landforms, vegetation trapping - would sharpen the causal interpretation, even if only to rule them out qualitatively.
5. Definition of "moderate" flows. The pivotal discharge band is variously "moderate" flows occurring for tens of days per year (abstract), 100–500 m3s-1 (L188), 120–600 and 120–400 m3s-1 deficit bands (L506–510), and "intermediate" flows elsewhere. A single definition, used consistently, would help readers (and importantly practitioners) take away the operative flow range.
Specific comments
Eqn 2 (L308). With s defined as ρs/ρw ~2.65 (L309), the denominator should be (s-1)g, not sg: the Wilcock–Crowe dimensionless transport rate is defined with the submerged specific gravity.Eqn 19 (L381). I concur with RC1’s correction (the exponent should be Fs^(-1/5); as printed the logarithm’s argument is negative for any Fs < 1). On RC1’s question of whether the resulting large difference between Fs, GW and Fs is realistic: it is the intended behaviour of the Grams and Wilcock (2014) formulation - with the corrected sign, Eqns 18–19 reproduce the benchmark (L690: 90% of the pure-sand suspended rate at Fs ˜ 19%). Correct the equation, confirm the implementation, and note this behaviour explicitly.
Eqn 1 (L276). RC1 notes the inconsistency between Eqns 1 and 14; the resolution is that "log" in Eqn 1 must denote the natural logarithm for dimensional consistency with the C = 18 log10 form of Eqn 14. Please write "ln" explicitly.
L167–169. The post-abstraction mean (63.6 m3s-1) is below the median (67.8 m3s-1) - seems unlikely for a flood-skewed record - and a median reduction of only ~7 m3s-1 is difficult to reconcile with a consented take of ~27 m3s-1 at intermediate flows (L218). RC1 suggests dropping the median; either way the values need verifying, as the inconsistency suggests a possible error.
The caption describes only panels (a)–(d), with the letters misaligned (the caption’s (c) corresponds to panel (d); panel (c) is the t-ratio map). Here I disagree with RC1’s suggestion to drop panels (e)–(h): panels (e), (f) and (h) constitute the validation of the flow-library interpolation - the t-ratio distribution in (h) underpins the 91% agreement result that justifies the library approach on which the whole paper rests. I recommend retaining these, and describe them clearly in the caption.
The paragraphs at L600–613 ("Retaining the facies-based GSDs…", discussing Figs 10c–f) belong to the Section 4.1.3 sensitivity results but sit inside Section 4.2, interrupting the Figure 12 discussion (an editing artefact?). The block should follow L576. Sections 4.1 (L408) and 4.1.1 (L414) carry identical titles, and "Run 257" (L478) is used before Table 2 defines the run codes.
The preferred model is Run 257 (L478, L700; Fig. 8 caption), but Figures 13 and 14 use Run 256 as baseline - make these consistent or justify the change.
L611–613: reducing ξmax from 20 to 10 is reported to increase transport by 5%, but to 1 by only 3% - non-monotonic where monotonic behaviour is expected; check whether "respectively" is reversed. L237: "1010 m2 of riverbed" (the braidplain is of order 108 m²) simplified into "~3 × 103 grid cells" (Fig. 5 gives Ngrid = 500,000) - both need correction. L481: thalweg "assumed bed Fs of 5%" versus Fs = 10% in the Fig. 7a header. Healey (1997) is quoted as 42–138 kt/yr at L194 but 50–100 kt/yr at L663. L513–515: 273 => 240 m3s-1 is a 12% reduction, stated as 11%, and Fig. 12c uses 271 m3s-1 where the text gives 273. L766: the cross-reference to Fig. 8a should apparently be Fig. 13b.The abstract (and L717, L799) states that the effect of flow abstraction is "comparable to a 10–15% change in the bed sand fraction". This reads two ways: Fs changing by 10–15 percentage points (from 20% to 30–35%, say), or by 10–15% of its value (20% to 22–23%) - a substantial difference. The sensitivity runs (Fs = 1–50%) suggest the former. Please state the intended meaning, ideally with a worked example, and the direction of the equivalence (transport lost to abstraction = lowering Fs by that amount).
Technical corrections
L57: "variation in sediment facies govern" ? governs.
L120: "Anecdotal reports … provides" ? provide.
L129: "Gaeumann et al. (2005)" in text versus "Gaeuman" in the reference list.
L195–196: sentence ends without a full stop.
L267: "Schubert (2015)" => Schubert et al. (2015).
L285: "shear stress (t)" => (tau).
L340, L698: "McCarron (2019)" ? McCarron et al. (2019).
L378: "if defined as" => "is defined as".
L487: "Point 3 2". L529: "GSDs sand fractions" - missing "and".
L532: "then testing spatially D50" => spatially varying D50.
L606: "after Fs~=15" - formatting.
Fig. 4 caption: "prevalence expressed a width" => "as a width"; "(shaded)" appears twice.
Figs 13/14 captions: stray parenthesis after "12b". Spelling: "naturalised/naturalized" and "parameterised/parameterized" both appear - please standardise.
"Rangitata | Rakitata" appears once (L143) and nowhere else - adopt a consistent usage.
References: Ferguson (1986) and Nava et al. (2019), both cited at L178, are missing from the list; Altenau et al. (2017) has unknown author "and Presented, A. T."; Redolfi et al. (2016) and Adams (1980) have malformed DOIs; Kuhnle et al. (2016) conflates conference and journal details; Gibb and Adams (1982): "oamaru and banks peninsula" lowercase; L206: "Rogers et al., (2025)" - remove the comma.Concluding remark
In closing, I would say I think this work is an important step in modelling of river sedimentary systems. A vital question for the next generation of reduced-complexity and 1D morphodynamic models is whether bed condition can be carried and evolved robustly rather than fixed as a snapshot. The authors’ stated ambition (L777–779, L795) to distill the spatial information captured here into a simpler framework that can consider change over time addresses exactly that question. The paper suggests that spatially resolved substrate data is now the binding constraint, rather than hydraulics or transport formulation. With the revisions above, this will make an excellent contribution.Citation: https://doi.org/10.5194/egusphere-2026-2149-RC2
Data sets
Substrate and hydrodynamic scenarios for "Sediment transport capacity in a large braided river: integrating substrate mapping with flow scenarios" J. M. Rogers https://doi.org/10.6084/m9.figshare.32030595
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 311 | 108 | 32 | 451 | 46 | 48 | 37 |
- HTML: 311
- PDF: 108
- XML: 32
- Total: 451
- Supplement: 46
- BibTeX: 48
- EndNote: 37
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
see attached file