The paper entitled “Drainage rearrangement in an intra-continental mountain belt: A case study from the central South Tian Shan, Kyrgyzstan” by Gong and co-authors, presents a geomorphic analyze of the Saryjaz drainage basin localized in the Kyrgyz Tianshan. Based on a topographic analyze and the measurement of various geomorphic metrics, the authors identify knickpoints of different origins. They found several transient kick points in tributaries downstream of 180° direction change of the Saryjaz river. Authors argue these transient knickpoints were created by a large river capture which occurred 1.5-4.4 Ma ago and was driven by the overfill of a large intermontane basin.

Reconstructing the drainage evolution of a given river basin through time is often challenging because rivers, being very active and mostly destructive in intramountain area, leave few remains of their past. To tackle this issue, this study presents an interesting geomorphic approach which combine various measurements of several metrics mainly to identify knickpoints along river profiles. The main results of the paper, ie the identification of the transient knickpoints, is mostly robust. However, (1) the general motivation of the study is unclear, (2) the interpretation on the origin of the transient knickpoints (i.e river capture) should be better supported by a robust (statistical?) analysis, and (3) the scenario proposed to explain this supposed river capture (i.e. basin overfill and inverse river flowing) is speculative since not supported by any robust observations/data. More observations, analyses and quantifications are needed to better support this story.

These 3 main issues should be better addressed before final publication.

Motivation and goal of this study

The general motivation of this work and why authors have decided to study the Saryjaz drainage basin is unclear and based on a subjective assertion. It is said lines 52-54 “Within this tectonic and climatic context, the South Tian Shan exhibits a significant contrast between a longitudinal drainage pattern in the west and transverse drainage in the east (Fig. 1).”

Also L83-85: “Drainage basins in the east are mostly characterized by transverse streams, whereas to the west, they show longitudinal patterns. The transition zone between these two drainage patterns hosts both the highest topography and the Saryjaz River catchment (Figs. 1, 2).”

So, as far as I understood, the starting point of this study is a supposed contrast in river drainage pattern between East and West of the South Tianshan and the fact that the Saryjaz basin is located at the transition zone and is, therefore, of particular interest. First, what means East and West ? With respect of what? Longitudes should be given.  

More importantly, this assertion of a contrasted drainage pattern is qualitative and may not be supported by the data. Indeed, the only argument presented by the authors is a rapid interpretation of Fig. 1 which displays the main rivers around the Saryjaz basin. Yet, on this figure several large longitudinal rivers are also present to the East of the Saryjaz basin as well as transverse rivers to the West. Moreover, why focusing only on this part of the South Tianshan? This range is much larger. Farther East (longitude >E83° which is the limit of Fig.1) the range is dissected by a long longitudinal river that goes down to the Bayanbulak intramountainous basin.

In conclusion, the statement that the drainage pattern of the South Tianshan exhibits a longitudinal contrast is rather vague, subjective though important for the study. This needs to be strengthened for instance using a more quantitative approach. Below is a suggestion of a quantitative analyze of the flowing directions of rivers in the South Tianshan (see screen shot in Fig. A that show the region that I have selected). This analyze could likely be refined/improved.

Using GIS and 90m SRTM DEM I extracted the flow directions of each cells of the main rivers (defining an arbitrary accumulation threshold). The figure B shows the frequency distributions of the NS, EW, NW-SE and NE-SW directions along longitudes. The frequency values are normalized to the total number of cells found for each bins of the histograms. We see from this figure that the change of East-West flowing directions in South Tianshan is tenuous. May be, it increases a little from E79° to E77° of longitude?

NB: I restricted my analyze to the main reliefs and excluded the large Tarim flat foreland basins

The flow accumulation threshold should be better defined as the limit between the hillslope and fluvial processes (log plot of slope vs. drainage area).

Also, the strike of the range is not purely E-W while the flow directions were calculated based on cardinal points.

Authors also need to clarify their zone of interest. What means “South Tianshan”? Why not considering longitude >E83°?

Moreover, while in the introduction the authors describe an EW contrast of drainage pattern across the South Tianshan, most of the paper focuses only on a U-turn made by the main stem of the Saryjaz river.

The motivation of the paper being unclear, the goals and the reasons why authors have chosen the methods they used are also unclear.

It is said L53-54 “However, it is not clear how or if the drainage pattern responded to Cenozoic structural reactivation and the uplift of individual ranges, a major change in climate, or the impacts of locally intense glacial erosion.”and  L55-59: “To unravel these complexities, we investigated the transition area between the regions of longitudinal and transverse drainage: the anomalously large Saryjaz catchment, which drains the highest part of the South Tian Shan, and includes two Neogene intermontane basins: the Ak Shyyrak and Saryjaz basins (Figs. 1, 2).

What means “unraveling these complexities”? Is the goal to explain the supposed East West difference in drainage pattern (assuming it is real)? If yes what are the hypotheses to explain this difference and how these hypotheses can be tested? For instance, what would be an impact of a Cenozoic reactivation? How would it differ from a change in climate and/or glacial erosion? Consequently, what are the methods to test these hypotheses, why, how do they help to test these hypothesis/the different scenari? Why in particular focusing on knickpoints?

Authors only said they combine series of topographic analyses and metrics without explaining in detail their goals and what each method will provide “We combine quantitative analysis of topography using fluvial metrics (i.e. slopes, channel steepness, integral proxy χ), with mapping and characterization of knickpoints throughout the catchment.”

The methodological part is very technical and does not provide much more information regarding the goals and what each method will provide to address the general question and hypotheses of the study, which are, I recall, unclear.

Interpretation of the knickpoints

First, it is said L244 that knickpoint at distance <250m from a fault are identified as “tectonic knickpoints” and hence disregarded. But why 250m? please justify this value. Moreover, could tectonic knickpoints migrate upstream and hence be at a distance >250m from the fault they originated from? For instance, in the Apenin in Italy, tectonic knickpoints are localized several kilometers upstream of the faults (see Whitaker et al., 2008). This point should be better discussed/argued.

More importantly, it is said L410: “our observation are clearly more consistent with transient knickpoint migration triggered by drainage capture”.

To support this conclusion, according to authors (see Fig. 3a), the knickpoints should “clearly” have the same Incision Depth and have initiated all at the same time. First and so far, the Onset time vs. Distance from outlet is not shown anywhere. This is surprising.

I understand that the Onset time are shown in Fig. 7b, where, indeed they seems similar in all studied basins (whatever the glacial model is). But Fig. 7b is a “Caltech plot” while an indisputable demonstration would be to show the onset time as function of the Distance from outlet as suggested in Fig 3.

Yet, I wanted to plot the Onset time vs. Distance from outlet to check authors conclusions but I couldn’t because the data are not provided. The onset time of knickpoint migration should be given in Table 1 and not only shown in a “Caltech” plot in Fig. 7b.

Second and similarly, in Fig. 6b, giving the scattering of the data, the fact that incision depths are similar in all basins is far from being obvious and “clear” as argued by authors.

To support their conclusion about the origin of the transient knickpoints, authors could better statistically analyze their data (Online Isoplot R, F-Test ??).

Scenario to explain the supposed drainage capture

But let’s assume that the drainage capture is robust and supported by a better statistical analyze of the data. The scenario that is proposed to explain this capture also questions and may request more quantitative analyzes/observations.

The mechanism that is proposed is “overtopping of the divide: during the Pliocene-Pleistocene period, the Ak Shyyrak Basin gradually filled with sediment, until river aggradation caused the west-flowing channel to eventually reach and overtop the drainage divide.” To support this mechanism the authors claim that “the sedimentary remnants, inferred from satellite imagery” occur “east of and high above the Saryjaz river in the vicinity of the “U-turn” (Fig. S6). It is very challenging to see sediments on the Google Earth capture presented in Fig S6. I zoomed on google Earth on the crest where authors claim sediments remain (see Fig. C). I don’t see any evidence for sediments here. The crest is darker but I think it is just a shade and a misfit between the image and the DEM. And we can see the bedrock everywhere on both flanks. To better support the presence of sediment high above the river, do authors have any other observations? Like field photo? Geological map?

Moreover, those supposed sediments lie at >3200m more than 950m above the modern river. What is the elevation of sediments in the Ak Shyyrak basin, is it consistent with such elevation downstream above the U-turn ? Do you have any Geological map of Ak Shyyrak basin ?

I understand it is difficult to do field work in this region but I think a better geomorphic analyze of the different valleys/basins and their sediment remains/filling is still possible using high resolution satellites images and DEM. It is important to better support the proposed scenario for instance focusing on recent alluvial deposits that could be found in the different valleys/basins. What are their elevation?  Do they support sediment overfill of the basin and a major incision since 2-4Ma?

Also, in the proposed scenario “in Miocene time, a west-flowing river connected two intermontane basins, Saryjaz and Ak Shyyrak, and likely continued westward to join the current Naryn River.” Then “The capture event would also have reversed the Ak Shyyrak river to flow east into the Saryjaz catchment rather than west into the Naryn catchment” . I think a more quantitative analyze of the implication of this inverse flowing is requested. The modern water divide between the river that flows West toward the Naryn basin and the Ak Shyyrak river that flows East lies around 3500m. In the proposed scenario the elevation of the divide should not have changed (assuming negligible uplift and sediment filling?). If we assume that the Miocene East flowing Ak Shyyrak river had a similar slope (<0.5-1%) than today in the Ak Shyyrak basin, then the elevation of this river at the U-turn location (90 km east from the modern divide) should be >600m above 3500m. The valley bottom being now at 2300m at the U-turn, this would require >1800m (3500+600-2300) of incision in 2-3 Ma (>0.6-0.9 mm/a), and even more upstream. Is it realistic for the Tianshan? How does this value compare with other region where drainage basin captures have been documented? Is it consistent with the order of magnitude of incision depth observed with the transient knickpoints? Any observations upstream of the U-turn to support more than 1800m of incision? A figure with the paleo-profiles vs modern profiles to see the implication in terms of incision would be helpful.

Other comments

Part 2.3 should go in the method (3.3) and not along the Geological/climate background of the Tianshan. This would avoid several useless repetitions and would make the reading more straightforward and easier.

The calibration of the erodibility coefficient is limited to 8 unpublished 10Be derived basin average denudation rates located around the Naryn basin, >200km west of the studied area. Why not using the largest 10Be dataset of Charreau et al. 2023 which includes much more larger drainage basins in the Eastern Tianshan? Lithology, climate history etc is likely similar in Eastern Tianshan.

By the way, the reference of Kudriavtseva et al., even if in review, is not provided in the list of publication at the end of the paper. Instead of this unpublished paper I would rather quote the PhD thesis.

L112: In the Eastern part the Tianshan is bounded by the Junggar basin to the north. Not the Kazakh plateform

L116: what about the Kazakh plateform mentioned before?

L142: those rates are derived from GPS. This should be said.

L142-145: several studies have also constrained the Quaternary deformation rates across the South Tianshan using geomorphic markers.

L145-148: these values are for the entire Eastern Tianshan, not only for the South range.

L213: some explanation in sup info of how fig S1 has been made would be helpful. Not everyone is a specialist of the power law approach.

L245: again why 1000000m2 ? justify this choice

L308-309: it is very hard from Fig 2b to see this gradian. May be a plot of slope vs distance along the basin and/or a zoomed map would be helpful (could be given in sup info)

L313-314: how can we see the change in Ksn in the Ak Shyyrak basin from fig 2b while it shows the slope ?

L348-349: please provide a figure (map, plot) that shows this “slight” increase.

L381-382: useless repetition. This was already said in the method part.

L315-316: what is the width of the swath profile? This is important since it could bias the data and derived interpretation. The distinction between the so-called low-relief vs high-relief region is arbitrary and qualitative. The change in reliefs along the main stem is in general very gradual and I don’t see two zones marked by a sharp change between them. Only the region around the U-turn shows a rapid change in relief but if we overlook this area, reliefs gradually increase from 130km to 60km, then remain high from 60 to 20km and then decrease slowly up to the outlet of the basin

For instance, in Fig6a, reliefs are relatively high from 90 to 100km (>3000m) while authors call this region low-relief. Similarly, reliefs are low near the outlet (<1000m) while called high-relief.

L474-476: this is wrong. 10Be derived paleo-denudation rates reconstructed in the Eatsern Tianshan increase from 9 to 4Ma and then remained steady (see fig 11 of Puchol et al., 2017). Moreover, Kudriavtseva et al. worked in western Kyrgyz Tianshan, not in the eastern region.

Figures

Figre 2c: it is very hard to see the Ksn “lines” and their values

Figure 3: the definition of the Onset time is unclear. Is it an absolute time? A duration with respect to the onset of base level drop ? or the time since knickpoint started to migrate (ie time before present)?

Figure 6: as said before, what is the width of the swath profile? How were defined the low- vs high-relief zones? Reliefs >2000m can be found in the so-called low-relief zone while reliefs <500m are present in the “high-relief” zone.

Figs. 6a-c: horizontal scales should be the same for all these 3 figures and the same than figure 6a. Moreover, I would put those figures below each other and aligned to figure 6a. It is unclear which type of knickpoints is shown here. I guess transient ones but please clarify this in the caption.

I see only 3 “upstream knickpoints” in figure 6a while 5 are shown on figs 6b and c.

Review of manuscript by Gong, Ling Xiao et al. submitted to Earth Surface Dynamics (ESurf) and titled “Drainage rearrangement in an intra-continental mountain belt: A case study from the central South Tian Shan, Kyrgyzstan”

Authors: Lingxiao Gong, Peter van der Beek, Taylor F. Schildgen, Edward R. Sobel, Simone Racano, Apolline Mariotti

Gong et al observe transient knickpoints in tributaries downstream of a sharp 180ᵒ bend in the main stem of the Saryjaz River in the south flank of the Tian Shan mountains. The analysis of the knickpoint distribution show that their elevations decrease downstream, whereas incision depth, χ values of knickpoints (measured from the trunk river tributary junctions) and steepness index (ksn) values and ratios are constant among tributaries. They interpret ~500 m of incision as driven “top-down” by a large-magnitude river-capture event, and that late Cenozoic tectonic rock and surface uplift did not trigger the capture.

Main comment

That is a very interesting study that should help gain insight into the interactions between mountain building and the dynamics of surface processes. Disentangle the relative effect of external factor and intrinsic drainage dynamics on river profile evolution in tectonically active settings is not an easy task. However, in its current form the manuscript lacks some analysis to support the conclusions, and I recommend the authors to add analysis following my comments below, before publication. In general, the text is clear and concise, well written and the figures are of good quality. References to previous work is also good.

The authors base their interpretation on diagnostic features for river capture extracted mainly from three papers cited in the manuscript (Giachetta and Willett, 2018; Whipple et al., 2017 and Rohrmann et al., 2023). However, my main concern is that the geomorphic evidence highlighted in their analysis of the topography do not exactly fit with such diagnostic features for river captures. The geomorphic evidence presented in the manuscript is:

1) in the tributaries of the Saryjaz River, the increase of elevation of knickpoints (interpreted as transient features) parallel the modern trunk river profile, and

2) their location at similar chi values (5.4+-30%) measured from the trunk river.

The authors analyze the amount of incision in tributaries of the downstream reach below the capture point inferred to be located at a prominent “U-turn” in map view of the Saryjaz River. The amount of tributaries incision in the downstream reach of the Saryjaz River does not increases upstream. However, I would have expected to see a graph where knickpoint elevation increases upstream, jointly with an increasing amount of incision. Upstream of the capture point, knickpoints in tributaries of the Saryjaz trunk river would be expected to lie at the same elevation, jointly with a decreasing amount of incision headward, which is what is theoretically expected in the case of a gain in drainage area of a trunk river. Because these behaviors are not recorded in the tributaries, I conclude the authors do not show an analysis that support the transient knickpoints to result from an increase in erosion rate downstream of a capture point. Rather, the slight decrease in the amount of incision from km40 to km80 (Figure 6c) seems to disprove their interpretation.

However, they refer in the discussion to geological evidence that support a capture event around the “U-turn” in the Saryjaz River. The contradiction between geomorphic and geological data should be investigated. The problem may lie in the location of the capture point which may not be located at the “U-turn” in the modern river path, as proposed by the authors. I think the effect of drainage integration of the upper Saryjaz catchment on the chi-elev plot should be quantified to predict the amount of incision such a capture would produce at the capture point and downstream, like the test of capture gain effect on river profiles in Giachetta and Willett (2018).

Maybe a comparison of the longitudinal river profiles of the transverse rivers sourced in the South Flank of the Tian Shan and that emerge in the Tarim Basin, with the Saryjaz trunk river and its tributaries would help convince the reader that a capture event produced the main transient knickpoints. See Giachetta and Willett (JGR 2018) for the diagnostic features suggesting capture events (Scenario 2). Please consider also to add a plot of knickpoint distance from the Saryjaz outlet vs chi (calculated from the trunk river outlet) to compare with the theoretical prediction of the scenario 3 in Giachetta and Willett (JGR 2018).

Specific comments

L171 [… the amount of incision recorded below the knickpoint will be similar for all Tributaries…], what data or model support this assertion? that’s not what Giachetta and Willett, 2018 and Rohrmann et al., 2023 show in their studies. They argue for a downstream decrease in the amount of incision in the downstream portion located below a capture point, after a gain in stream power of a capturing river. Channel gradients typically tend to relax downstream of a capture due to the amplified stream power, whereas upstream of a capture, a knickzone tends to gradually form and expand.

The authors decided to show only a subset of the tributaries and the associated knickpoints they studied. I think the reader needs to see the raw data in a synthetic figure in the main text. Please add a chi-elevation plots of each tributary subcatchment with all their tributaries (not a stack of S4 subplots), together with the localization of the knickpoints used in the inversion. this would help support the interpretations. I think the ‘representative’ χ – z plot of transient slope-break knickpoints in figure 4b does not give enough information of the general geometry of the tributary profiles of the Saryjaz River. Also, a companion figure of fig 4a with the geological map would help to assess any possible tectonic/lithologic control on the spatial distribution of transient knickpoint.

l411-412: “infer that the capture position is marked by the “U-turn” in the Saryjaz River” models predict deeper incision at capture point! This is not what fig6c shows.

L412“Ak Shyyrak River corresponds to a paleo-downstream reach of the upper Saryjaz,” in this interpretation, but the authors don’t show new data (e.g. paleoflow directions) to support that. Maybe, better say “…would correspond…”

L416-418: that’s not what models show… maximum amounts of incision are expected at and around a capture point.

L442-444: “The divide lies within Neogene sediments, providing a minimum elevation reached by the fill of the Ak Shyyrak basin prior to the capture and supporting a scenario in which capture was driven by overtopping of the Ak Shyyrak basin.” Please add the position of the sediment remnants on the topographic profile in figure 6a. This would help visualize the degree of overfilling in the previously closed longitudinal valleys in the interior of the Tian Shan.

L449-452: “Our analysis of the Saryjaz catchment demonstrates that over a long period of time after 450 the capture event, the impacts of drainage capture will migrate from the trunk to the tributaries, producing transient ksn anomalies, and eventually reshaping the whole river profile into a new equilibrium state.” If true, why do only ~200m of incision is observed around the U-turn (fig 6c)? I would expect a deeper incision after 2.8 ± 1.3 Ma (l459) of upstream propagation of the capture-induced wave of incision. Actually, models predict the maximum of capture-induced incision at the capture point which is not what is observed. Please discuss this point.

l480-482 “our study here do not see transient knickpoints associated with this reactivation from Saryjaz catchment, which might indicate that the change in uplift rates was rather low, or that it actually occurred earlier and no longer visible as a transient signal in river profiles” please explain why the knickpoints in the Saryjaz River especially the ones close to the South Tian Shan Fault may not be tectonically-induced. I guess the inversion of river profile with chi values calculated from the outlet of the Saryjaz River would give ages similar to l477-478 “[…] sharp change in provenance from a mixed Tian Shan-Pamir source to local source between 6 and 3.5 Ma (Rittner et al., 2016; Richter et al., 2022).” If true, rock uplift may explain some of the observed transient knickpoint, while others could have been driven by a river capture event. If not, this point would reinforce authors’ interpretation.

L486-496. In the last paragraph the authors suggest a 3 phases evolution, with 1) uplift of a new south topographic barrier, 2) the infill of closed basins in the center of the Tian Shan and 3) the opening of the closed basins by overtopping of sediments at the origin of a capture event. The bottom Paleozoic bedrock of the longitudinal valleys in the central Tian Shan, where late Cenozoic clastic sediments aggraded, is more than 1000 m above the northern margin of the Tarim Basin. If such a difference would have existed at the time of disconnection/closure of the Ak Shyyrak and Saryjasz Basins in the center of the orogen, one may have expected the erosion by the Saryjaz River to have balanced the uplift rate. The authors claim this did not append and that the Saryjaz River have been defeated by rock uplift, which in turn would imply that in Miocene times(?) the Saryjaz River did not have yet its present high potential energy toward the South. A corollary is that the center of the Tian Shan should have been surface uplifted by 1000 m with respect to the Tarim Basin before event 3. This would support the view that a delay exists between orogen building and drainage reorganization, as observed in many geological settings (e.g. Babault et al., J of Asian Earth Sci 2018, Rohrmann et al., Science Advances 2023). Maybe the authors may discuss this reasoning in their model of mountain building and drainage evolution.

Specific comments

Please add a cross-section in the geological settings of the South Tian Shan that passes close to the transverse reach of the Saryjaz River.

I did not find a reference to the method that has been used to calculate the steepness index values.

It may help the reader to see the Figure S3 in the main text, also add the location of the U-turn in that figure and highlight the location of the transient knickpoints upstream of the U-turn, I can’t see them.

L406 “tributaries upstream of the “U-turn”, especially within the intermontane basins, show lower slopes (mostly < 30°) and generally lack slope-break knickpoints.” Please give a comment/explanation for this feature which seems significant to understand the drainage evolution of the Saryjaz River.

l409-410: ‘Considering theoretical predictions of the differences in patterns of knickpoint elevation and incision depth for knickpoints triggered by drainage capture versus base-level fall (Fig. 3),’ there is a difference between theoretical predictions and the trends in figure 3. This should be corrected, with increasing incision upstream in the tributaries of the downstream reach of the Saryjaz River, see comments above.

L410-412: “our observations are clearly more consistent with transient knickpoint migration being triggered by drainage capture. We infer that the capture position is marked by the “U-turn” in the Saryjaz River” to help the reader, it may be worth to recall here which are the features you observed that you take as diagnostic for river capture.

Other comments

L52-53 and L83-84 “…transverse drainage in the east…” true in the south flank only! Should be explained with more details

L168: “…up tributary valleys” specify downstream of the capture point

L169 “…constant vertical velocity” specify, true if n=1 in the detachment-limited stream power incision model.

L172-173: “…all tributary knickpoints will have retreated a similar distance from the tributary junctions because they were all initiated around the same time” assuming detachment-limited stream power model and n=1. I would expect this is not true if n ≠ 1. Horizontal and vertical migrations of knickpoint are function of the erosion rate of the propagating slope patch raised to the power of 1/n, and of the background uplift rate also raised to the power of 1/n (Royden and Perron 2013). If you have a look the figure 3b in Giachetta and Willett (2018), after a capture event, knickpoints in the lower tributaries have lower slopes than tributaries just below a capture point. The difference of migration rates might be small but it would be interesting to quantify it.

Figure 6d please reverse axis and plot distance from outlet vs chi as in Giachetta and Willett JGR 2018. Although I understand that the plots 6b and 6c show distance from outlet in the x axis.

Please check you specify/add everywhere in the ms. i.e., in the text and in the figures (not only in the figure captions of the plots) if chi is measured from the trunk stream or from the outlet (of the trunk river) in the Tarim Basin. Because the studies you refer to values are plotted against both variables.