Formation of meandering streams in a young floodplain within the Yarlung Tsangpo Grand Canyon area in the Tibetan Plateau

Meandering is one of the most common platforms of alluvial rivers that characterizes sinuous and winding courses (da Silva & Ebrahimi 2017; Li et al. 2017; Nyberg et al. 2023). Despite its prevalence in fluvial systems, the natural phenomenon of meandering also occurs in the meltwater of Arctic glaciers, the Gulf Stream in the deep sea, and channels on intertidal mudflat and bedrocks (da Silva 2005; Seminara 2006; Howard 2009; Kleinhans et al. 2009).

Regarding the formative causes of the meandering river, numerous scholars offered a plethora of explanations grounded in diverse theoretical frameworks, however, a general consensus is still lacking (see Seminara (2006); Pittaluga & Seminara (2011); Güneralp et al. (2012); da Silva & Ebrahimi (2017); Kleinhans et al. (2024) for reviews). One of the more accepted views is that the intrinsic instability within the dynamic system of water, sediment and river boundary arises from periodic large-scale bedforms, known as alternate bars (Colombini et al. 1987; Nelson 1990). The formation of alternate bars culminates in sinuous thalwegs navigating in-between bars and banks that eventually trigger the planimetric instability of rivers, leading to the convective growth of meander bends (Ikeda et al. 1981; Lanzoni & Seminara 2006; Seminara 2006).

Laboratory experiments and field investigations also highlight the critical role of floodplains in sustaining meandering river patterns (Howard 2009; Kleinhans et al. 2024). First, no successful laboratory attempts have been made to establish a steady and high-sinuosity meandering river in a cohesionless environment; rather, meandering only occurs as a transitional state between an initially straight river to eventually a braided river (e.g., Federici & Paola 2003). Later, the introduction of the cohesive material or riparian vegetation proves to be critical in sustaining a laboratory meandering river (Schumm & Khan 1972; Smith 1998; Gran & Paola 2001). Naturally, meandering rivers often associate abundant riparian vegetation and cohesive banks that provide proper resistance against the braiding tendency of rivers (Parker 1998; Ielpi & Lapôtre 2020; Ielpi et al. 2022; Finotello et al. 2024).

Other explanations include the adjustments of the valley slope (Leopold & Wolman 1957), path of the highest probability (‘Random Walk’ theory by von Schelling (1967)), minimum variance of the downstream curvature gradient (Langbein & Leopold 1966; Yalin 1992), dynamic flow properties that facilitates structured transport of sediment (e.g., da Silva & El-Tahawy 2008; Blanckaert 2010), and slump blocks protection on outer banks (Liu et al. 2024).

The Cuoka sub-basin (29°52.58′ N, 95°33.54′ E) is a contributing area of the Parlung Tsangpo River located in the northeastern part of the Yarlung Tsangpo watershed near the Yarlung Tsangpo Grand Canyon of the southern Tibetan Plateau (Figure 1(b)).

The water source of the Cuoka sub-basin primarily originates from the melting waters of the Cuolama Glacier (Figure 1(b)) that flows from south to north into the famous Guxiang section of the Parlung Tsangpo River (Figure 1(c)). The sub-basin has a length of 900 m and a width of 350 m, with a depressed center and elevated sides, culminating in a basin-like formation (Figure 1(d)). The exit of the floodplain to the Parlung Tsangpo River is dammed in the northeast, leaving an outlet in the northwest (Figure 1(e)).

Due to the great Chayu earthquake (8.6 on the Richter scale) in 1950, massive ice avalanches and landslides occurred in the upper and middle reaches of the Guxiang Gully, accumulating large amounts of loose debris that blocked the gully. Later in 1953, a massive debris flow occurred, transporting a huge amount of sediment to the lower reaches of the gully, forming the Guxiang alluvial fan (Liu et al. 2015; Yu et al. 2020). In the meantime, the Parlung Tsangpo River forms a dammed lake at the Guxiang section during flood seasons. During the flood season, the rising level of the Parlung Tsangpo River backwaters the Cuoka floodplain, resulting in a significant water rise that overfloods the entire sub-basin forming a shallow and grassed lake (Figure 1(e)). This facilitates the sedimentation of fine grain and clay within the floodplain. Also, note that the main part of the Cuoka floodplain is exempt from influences of the Parlung Tsangpo River which is ladened with sediment. The evidence is shown in Figure 1(e) where the waterbody of the Cuoka sub-basin remains clear with slight sediment, contrasting with the turbid sediment-laden flow of the Parlung Tsangpo River.

Within the Cuoka sub-basin are the two streams which exhibit a meandering pattern in their middle section. From the view of Figure 1(d), the mainstream of the sub-basin is the one on the right, with sinuosity = 1.52 (considering only the meandering path in the middle section). The total stream length is 1,200 m, and the average bank full width B is 31.5 m. On the left of Figure 1(d), the tributary stream manifests also a slight meandering course in the middle (sinuosity = 1.24). The total length of the tributary stream is 800 m, and the average bank full width B is 9 m.

Seven field surveys of the Cuoka floodplain were conducted (as of 14 December 2020; 8 March 2021; 25 May 2021; 25 December 2021; 30 May 2022; and 22 April and 23 December 2023), covering both dry and flood seasons (Figure 1(d) and 1(e)). Details of the field investigation are provided in Supplementary Figures S1 and S2 and Text 1.

The soil samples of the riverbank are further analyzed, as shown by the four sampling points (S1–S4) in Figure 3(a), in which S1 lies 20 cm below the bank surface and S4 is at the bank toe. The particle sizes of the first three samples (S1, S2, and S3) were analyzed using a Malvern MS2000 laser particle size analyzer. Sieving analysis was used for the sample at the bank toe (S4) as larger grains were found in the vicinity of the riverbed. Volume fraction (%) distributions against the particle sizes are presented in Figure 3(c) and 3(d) for all samples. For the first three samples above the riverbed (S1–S3, Figure 3(c)), the volume fraction curves are essentially the same, indicating a continuous process of sedimentation after the 1950s. The characteristic grain sizes D₅₀ (50% of the accumulative volume fraction) are 108.8, 112.1, and 121.5 μm for samples of S1, S2, and S3, respectively. At the bank toe (S4, Figure 3(d)), double peaks of the volume fraction curve (corresponding to particle sizes of 0.5 and 10 mm) are observed, highlighting dual influences from both the original basin and the subsequent sedimentation process after 1950.

The talwegs for the meandering part of the main and tributary streams are presented in Figure 4(b) (here, the diamond symbol marks the apexes). This measures 14.0 and 20.6 for the normalized wavelength (Λ_M/B) of main and tributary streams. Consider the wavelength equation by Parker (1976), i.e., Λ_M = 5c(Bh)^1/2 (here c is the dimensionless Chézy resistance factor that can be approximated by c = 6.82(h/D₅₀)^1/6 according to Yalin & da Silva (2001)), it yields Λ_M/B = 13.9. Here, the predicted value of Λ_M/B is very close to the actual wavelength of the mainstream but deviates significantly from that of the tributary stream. Because Parker's equation (1976) stems from one of the earliest analyses of bed instability (alternate bars), the alignment of predicted and measured mainstream wavelengths reinforces the link between alternate bars and the formation of river meandering in the mainstream, as first illustrated in Figure 4(a).

This reveals critical geometric parameters of the channel regarding the Kinoshita function, i.e., deflection angle at the crossover θ₀ = 70°, flatness factor J_f = 0.12, and skewness factor J_s = 0. Therefore, it shows that the mainstream is in the middle stage of its evolution (θ₀ = 70°), noticeably flattened (J_f = 0.12) but without any skewness. On the other hand, the tributary stream shows more randomness, attributing possibly to the local discontinuity of the floodplain (e.g., the local hard points, Guo et al. 2019).

For the mainstream (Figure 5(d)), the upstream part exhibits mainly longitudinal migration (lateral migration is nearly 0) that maximizes near the upstream crossover. Downstream, the river exhibits both longitudinal and lateral displacement in the same order, both maximizing just before downstream crossover. Differences between the maximum and minimum lateral migration rates highlight the lateral expansion for the meander loop (=0.042 B/year = 0.89 m/year). Associating the lateral amplitude of the meander loop (lateral distance between the two apexes, 105 m) and the lateral loop expansion rate, it estimates approximately 107 years for the chronology of the mainstream meander. Despite an overestimation of 1.45 times, the estimated lateral chronology of the meandering development still falls within the recent century. Nevertheless, the overestimation results from an equivalent migration rate assumption at the earlier stages of the floodplain when the bank height is significantly lower. Figure 5(e) is the migration pattern of the tributary stream, showing predominate lateral displacement. Also, the lateral migration tendency replicates that of the mainstream, yielding approximately 83.3 years for developing chronology.

The migration rates of the two meandering streams within the Cuoka floodplain (0.7–0.8 m/year) are consistent with other large-scale natural meandering rivers. For instance, the migration rates of meandering rivers in the Upper Mississippi River Basin are reported to be around 0.5–2 m/year (Hudson & Kesel 2000), while in the Amazon Basin, migration rates can range from 0.1 to 1.5 m/year (Latrubesse & Franzinelli 2002).

Chen & Tang (2012) incorporated the analytical solution of the meandering streams' flow field (Johannesson & Parker 1989) and the bank erosion and retreat model so as to compute the channel migration. Starting with an infinitesimal initial angle, the meandering centerline grows in amplitude while maintaining the sine-generated path at the early stages. However, at the latter stages, the growth of the loop sinuosity diminishes, and the centerline skews and flattens which transfers into a Kinoshita curve. The Kinoshita configuration of the Cuoka mainstream confirms the analytical predictions, suggesting a latter stage of channel development. However, the noticeably flattened (J_f = 0.12) but without any skewness of the channel indicates a new migrating pattern for the stream resulting from the complex interplay between local factors.

The first stage is that of Figure 6(a), elucidating the period after the formation of the original lateral moraine basin to just before the damming of the Parlung Tsangpo River at the Guxiang Gully in the 1950s. The main and tributary streams are expected to manifest as incised mountain rivers (as they are before entering the floodplain). Noteworthy, the elevated sides and depressed center of the Cuoka sub-basin were possibly formed during the downward movement of Quaternary glaciers, in which the basin's inner side was subjected to the scraping and grinding during the advancing of the ice mass while the lateral sides experienced shearing forces.

The second stage of the floodplain formation is the initial alluvial process post the damming of the Parlung Tsangpo River as depicted in Figure 6(b). Here, the stage's ending time in 1990 is only a rough estimation due to a lack of high-resolution images. In such period, the elevated water surface of the Parlung Tsangpo River backwatered the Cuoka sub-basin necessitated the sedimentation of fine particles and clay. The process smoothed out the original valley slope that eventually formed the composite for the current Cuoka floodplain. The damming of the floodplain's northeast outlet was also likely to occur in such a stage, resulting from the sediment deposition supplied by the turbid flow of the Parlung Tsangpo River. The sinuous paths of the two streams are expected to grow downstream due to boundary constraints. The ambient sedimentary environment, along with an appropriate width-to-depth ratio, facilitated the formation of alternate bars in the mainstream.

The last stage is when the floodplain stabilizes, i.e., ≈1990 to the first investigation during 2020, as shown in Figure 6(c). First, the stratified layer of silty clay and root-soil composite contracted the river into a single-thread channel. Meandering paths of the main and tributary streams as initiated by alternate bars or local discontinuities developed. The mainstream manifests prominent inner bank deposition and outer bank erosion that leads to the migration and expansion of the meander loop.

The occurrence of the meandering streams in the Cuoka sub-basin reveals the following critical conditions in the formation of the meandering streams: (1) formation of the floodplain that establishes the material base; (2) proper bank strength (or floodplain resistance) provided by the floodplain that leads to the river contraction into a single-threaded channel; (3) occurrences of alternate bars (due to fluvial instability) or local discontinuity (e.g., hard points) that initiate river meandering; (4) the leveling of the river valley slop, without which a braided river would likely occur due to high flow velocity and significant sediment transport capability.

Notably, the prolonged flooding of the Cuoka floodplain due to backwater from the Parlung Tsangpo River facilitates the growth of herbaceous plants while preventing the establishment of trees. Conversely, in other areas where forests proliferate, the excessive bank strength provided by the trees' root systems inhibits the lateral development of streams and ultimately prevents the formation of meandering channels.

A case study on the unique discovery of two meandering streams within a young floodplain is presented. The floodplain, Cuoka, is located near the Yarlung Tsangpo Grand Canyon of the southern Tibetan Plateau – a region characterized by bedrock and braided rivers. The discovery facilitates discussion and elucidation of the formative causes and morphological features of the incipient meandering rivers. It is found that:

• (i) Alluvial sedimentation of the Cuoka floodplain can be associated with the damming of its mainstream, the Parlung Tsangpo River in the 1950s. The subsequent backwater effects overflooded the sub-basin which initiated a continuous sedimentation process of fine grains and clay.

• (ii) Adequate floodplain resistance provided by a stratified layer of silty clay and root-soil composite contracts the river into a single-thread channel, providing a necessary condition for the meandering stream formation.

• (iii) The mainstream meander path follows the Kinoshita curve with noticeable flatness and no skewness. The annual migration rate of the two streams is consistent with other large-scale natural meandering rivers. Congruences and disparities with the analytical model by Chen & Tang (2012) highlight the complex interplay of local factors in shaping meandering processes, offering valuable insights into both the unique characteristics of the Cuoka streams and the broader principles governing meander formation.

The present work highlights the formation of the meandering river in the region with rapid climatic and geological changes such as the Yarlung Tsangpo River basin. The principles of meander formation through alternate bars and channel migration, as well as the role of riparian vegetation, can be generalized to other alluvial river systems worldwide.

This study was supported by the National Natural Science Foundation of China (42371015, 51979012) and the Fundamental Research Funds for the Central Universities.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.