Evaluation of the Rock Uplift Pattern in the Central Yunnan Subblock, SE Tibetan Plateau: Based on the Bedrock Channel Profile

The uplift pattern of the southeastern Tibetan Plateau is strongly related to the topographic evolution stemming from the India–Eurasia collision. However, whether strain is localized along major faults that bound large tectonic blocks or is accommodated across regions has been strongly debated. In this study, we used stream power incision models to obtain the distribution pattern of the channel steepness indices to understand the rock uplift pattern across the area, as increased channel steepness indices often correlate with the rock uplift rates. In this study, the river longitudinal profiles were analyzed to obtain the distribution of the channel steepness indices in the Central Yunnan subblock. The results suggested very weak correlations between the steepness indices and the lithology, precipitation, sediment flux, or channel concavity indices. Along the Xiaojiang strike–slip fault and the interior subblock, the uplift rate was slower, while the northern part had uplifted faster and was controlled by thrust fault systems. The channel steepness increased gradually from south to north. Thus, the distribution pattern of the normalized channel steepness, k sn, index within the Central Yunnan subblock provides notable support for the argument for the thrusting transformation-limited extrusion model of the Tibetan Plateau.


INTRODUCTION
The Tibetan Plateau is one of the most studied natural laboratories for examination of the relationship between continental collision and landscape evolution (Clark et al., 2005;Clark, 2012;Allen et al., 2013;Zheng et al., 2013;Liu et al., 2008;Molnar and England, 1990;Liu-Zeng et al., 2018;Wang et al., 2019). Although there has been much work done in the region, it is still poorly understood how the southeastern margin of the Tibetan Plateau responded to the India-Eurasia continental collision. Additionally, whether the strain is mainly localized along the boundary faults of large blocks (Tapponnier, 2001) or accommodated within the block (Houseman and England, 1993;Royden et al., 1997;Clark et al., 2005;Clark et al., 2004;Wang et al., 2017) has been strongly debated.
Research on the bedrock channel fluvial systems has examined the relationships between climate variations, surface processes, and tectonic evolution (Raymo and Ruddiman, 1992;Hartshorn et al., 2002;Dibiase and Whipple, 2011;Ferrier et al., 2013;Dibiase, 2014;Finnegan et al., 2014;Pan et al., 2015;Kirby and Ouimet, 2011;Kirby et al., 2007;Lave and Avouac, 2001) and has been used to interpret the relationships between topography, elevation, and denudation rates (Pritchard et al., 2009;Schwanghart and Scherler, 2020). In this study, we selected the Central Yunnan subblock as our study area ( Figure 1). We extracted longitudinal river profiles from 320  rivers and calculated the channel steepness indices and concavities to determine the uplift patterns in this area (e.g., Kirby and Whipple, 2001;Whipple, 2004;Wobus et al., 2006;Oskin et al., 2014;Su et al., 2016) based on the digital elevation model data of the 90-m Shuttle Radar Topography Mission (downloaded from http://srtm.csi.cgiar.org/) by ArcGIS 10.2 and MATLAB R2015b. Thereafter, the influence of lithology resistance, precipitation, and sediment flux on channel steepness indices was analyzed to obtain the distribution pattern of the rock uplift within the subblock.

Regional Setting
The Sichuan-Yunnan rhombic block is located in the southeastern margin of the Tibetan Plateau. The Lijiang-Xiaojinhe Fault cuts the block into two parts: the Central Yunnan subblock in the south and the northwestern Sichuan subblock in the north   (Figure 1). In this paper, we mainly studied the Central Yunnan subblock.
The average elevation of the Central Yunnan subblock is about 2,000 m. The geotectonic location belongs to the southeastern margin of the Tibetan Plateau. It is adjacent to the Simao block in the south and the Caledonian fold belt in southeastern Yunnan to the east (Huangfu and Qin, 2006;He et al., 2009;Wang et al., 2015). For most parts of the area, annual rainfall ranges from 600 to 1,100 mm and gradually increases from north to south. Owing to the abundant rainfall, the study area has developed a dense river network, with many large rivers originating in the area (e.g., Longchuan River, Red River, Nanpan River, Pudu River, Anning River, and Xiaohedi River) (Figure 1).
The lithology of the central region of the subblock is relatively homogenous (consisting of Cretaceous-Jurassic sandstones and mudstones), while the lithology of the border region of the Central Yunnan subblock is complex (including Paleozoic shales, schists, and gneiss, Cambrian limestones and dolomites, and Triassic phyllites and mudstones). The subblock is In the Cenozoic, most of the fault zones and orogenic belts of the Tibetan Plateau were active due to the strong collision between India and Eurasia (Wang et al., 2012;Shen et al., 2016;Zhang et al., 2016;Zhang et al., 2017;Liu-Zeng et al., 2018). The Xianshuihe-Xiaojiang Fault is a very large left lateral strike-slip fault (e.g., Xiang et al., 2002;Li and Zhang, 2013;Yan and Lin, 2015;Xu et al., 2007). In the Late Cenozoic, the Xianshuihe-Xiaojiang Fault offset the NE-SW Longmenshan thrust fault by as much as 60 km, splitting it into the north Longmenshan (NLM) and south Longmenshan (SLM) thrust Frontiers in Earth Science | www.frontiersin.org February 2022 | Volume 10 | Article 821367 6 belts (Burchfiel et al., 1995;Wang et al., 1998). The SLM includes the Muli Fault, Lijiang-Xiaojinhe Fault, and Jinhe-Qihe Fault (Yin et al., 2020;Wang et al., 2012).
The uplift history and the expansion of the southeastern margin of the Tibetan Plateau have been studied intensely using several methods. Analysis of the 18 O isotopes of carbonate rocks in Cenozoic sedimentary basins Tang et al., 2017;Hoke, 2018) interpreted that the palaeo-elevation of the southeastern margin of the plateau reached its present height prior to the Eocene. In addition, episodic periods of uplifts have occurred in the region since the Late Cenozoic, determined through various thermochronometers (e.g., Xu and Kamp, 2000;Chen et al., 2006;Wilson and Fowler, 2011;Wang et al., 2012;Tian et al., 2013, Tian et al., 2014, Tian et al., 2015Shen et al., 2016;Zhang et al., 2016;Zhang et al., 2017;Liu-Zeng et al., 2018;Wu et al., 2020).

METHODS
The stream power incision model relates rock uplift and river incision over time along the longitudinal profile of the channel (Howard and Kerby, 1983;Howard et al., 1994), and it can be represented by Eq. 1: where z represents the channel elevation, x represents the upstream distance, t represents time, U is the rock uplift rate, and K is the bedrock erodibility and is related to the channel width, the water discharge, the sediment load, and the lithologic resistance (Howard and Kerby, 1983;Howard et al., 1994). m and n are constant exponents, and A is the drainage area.
For a steady state, channel elevation at a particular point does not change, and then zz/zt equals 0. From this, we can obtain Eq. 2.
where k s and θ are the channel steepness and concavity indices, respectively. Generally, θ oscillated between 0.3 and 0.6 (Snyder et al., 2000;Snyder et al., 2006;Whipple, 2002;Kirby et al., 2003;Wobus et al., 2003). K s is usually proportional to the rock uplift rates (Eq. 3). Although the longitudinal profile analysis was FIGURE 9 | Comparison of normalized channel steepness (k sn ) to annual precipitation in the Central Yunnan subblock. The precipitation data (timescale: from 1970 to 2019) were downloaded from http://data.cma.cn. The tributary catchments (white boundary) were selected randomly for correlation analysis between channel steepness with precipitation and drainage area. developed under steady-state assumptions, it has been extended to transient systems as well (e.g., Pan et al., 2015;Wang et al., 2017;Wang et al., 2019;Ma et al., 2020), and several studies (e.g., Snyder et al., 2000;Kirby et al., 2003;Wobus et al., 2003;Hu et al., 2010;Burbank and Anderson, 2011;Kirby and Whipple, 2012) showed that K s is a useful index for denoting rock uplift rates at transient states.
θ can be gained by a log transformation to Eq. 2 by Wobus et al. (2003).
The concavity and steepness indices are the slope and intercept of Eq. 5, respectively. K s can be gained using Eq. 6 (Perron and Royden, 2013).
where A 0 is the area factor and z b is the channel elevation at river outlets (x = 0). A 0 /A (x)' is dimensionless. Setting A 0 = 1 m 2 , the slope of Eq. 6 is k s . We first used the ArcGIS 10.2 and MATLAB 2015b scripts (www.geomorphtools.org) to derive the concavity indices ( Figure 2), steepness indices (Figure 3), and the longitudinal profiles (Figures 4-6). The steepness indices were determined using logS-logA plots according to the methods described by Wobus et al. (2003). Then, we interpolated the normalized channel steepness, k sn , values of the river basins in the Central Yunnan subblock and created three swath profiles of the elevation and local steepness (Figure 7) (the profiles are E/W-directed, 220-350 km long and 20 km wide), finally presenting a distribution pattern over the Central Yunnan subblock. Due to the strong dependence of k s on θ, a fixed reference concavity of 0. 45 (a typical value used in many studies, e.g., Wobus et al., 2003;Hu et al., 2010;Perron and Royden, 2013) was used to generate x-z plots and to calculate the k sn indices in order to compare the channels and channel segments of varying drainage areas (Whipple and Tucker, 1999). We resampled the elevation data at 20-m contour intervals with a 250-m smoothing window to decrease data noise (e.g., Wang et al., 2017).

RESULTS
In the Central Yunnan subblock, the 320 extracted river profiles had a mean concavity of 0.28 ± 0.063 (1σ), with a total range of 0.014-0.63 ( Figure 2). In this section, however, we just described the most representative stream longitudinal profiles in the subblock (Figures 3-7). The detailed results are depicted in the Supplementary Material (Supplementary Figure S1 and Supplementary Table S1). Furthermore, the k sn and θ values are also listed in the Supplementary Material (Supplementary Figure S1 and Supplementary Table S1).
In the northern areas controlled by a series of NE-SW thrust faults (such as the Jiulong and Muli thrust fault belts and the Jinhe-Qinghe thrust fault zone), these rivers had higher k sn values (Figure 3), which can also be found from profile 1 ( Figure 7B). The upper reaches of the Anning River also had high k sn values, which may stem from their close proximity to areas of thrust faults (Muli Fault and Jinhe-Qinghe Fault) (Figure 3).
However, in the south areas controlled by the strike-slip faults and within the block, the rivers had lower k sn values (Figures 3,   (Figures 7C, D). Based on the distribution pattern of the k sn indices along the Central Yunnan subblock ( Figure 7A), an approximate trend of change has been outlined, in which the average k sn indices showed a general and gradual decrease from north to south.

DISCUSSION
The K sn values were higher in the northern region controlled by the thrust faults and on both sides of the Jinsha River and were lower within the block and near strike-slip faults. The results suggest that the longitudinal river profiles correlated strongly with the underlying tectonic regime and may be used to make inferences regarding the tectonic setting or structure. However, if we want to attribute the variation of k sn to the rock uplift rates, the influence of the lithology, precipitation, and sediment flux should be excluded.

Lithology
Channel bedrock erodibility is largely controlled by the lithologic resistance (Stock and Montgomery, 1999;Palumbo et al., 2010;Wang, 2014;Wang et al., 2018). As river incision through more resistant rocks (low K) requires greater stream power compared to less resistant rocks (high K), knickpoints are usually observed near the contact between rocks, with higher k sn values in areas underlain by harder units (Duvall et al., 2004). Therefore, we analyzed the relationship between nine representative rivers with different lithologies (Figures 5, 6) and their k sn values to assess the influence of lithologic resistance.
Although the knickpoints observed along the Longchuan River (large-scale river, basin area > 10 4 km 2 ), C23 and Lz9 (middle-scale river, 10 3 < basin area < 10 4 ), are located near lithologic boundaries (Figures 5, 6), the influences of variable lithologies on the channel profiles were weak. Along these rivers, the extracted k sn values did not match the mapped lithologies ( Figure 8). For example, in the lower reaches of the Anning River, although the lithology varied, we did not find an obvious knickpoint ( Figure 5). But an obvious knickpoint was extracted from H20 ( Figure 6), where the bedrock lithology of these channels is uniform. Therefore, regional lithology does not appear to be a major influence on the channel profile of the subblock.

Precipitation
The effect of climate (mainly precipitation) on the channel profiles must be examined, as increased rates of rainfall often lead to increased river discharge and erosion of the bedrock beneath the channel. Ultimately, increased rainfall rates will decrease channel steepness, as channel steepness is often inversely proportional to the erodibility (Eq. 3). Here, the monthly precipitation data of 15 meteorological stations were collected from the China Meteorological Data Network (timescale: 1970-2019; downloaded from http://data. cma.cn) (Figure 9). These data were interpolated to derive the spatial distribution of average precipitation across the Central Yunnan subblock. The annual rainfall in the study area decreased from >1,200 mm (close to the Red River fault zone in the south) to <500 mm (close to the Lijiang-Xiaojinhe fault zone in the north) (Figure 9). This may be related to monsoons from the Indian Ocean being blocked by the Himalayas, promoting increased southeastern compared to northwestern rainfall. Increased runoff can be seen in the numerous river channels that have developed in Southwest China, such as the Yarlung Zangbo River Grand Canyon, the Lancang River Valley, the Jinsha River Valley, and other south-to-north valleys (Nie, 2018;Li, 1999).
From these data, there were two reasons to exclude precipitation as having a key role in controlling the k sn values in the subblock. Firstly, by extracting the k sn values from a small watershed in the study area, we found that the k sn values in the northern part of the study area were higher than those in the southern part of the subblock (Figure 7). However, the rainfall in the southern part of the study area is higher than that in the northern part, which is inconsistent with our observation results of k sn values ( Figure 9). Secondly, we selected catchment basins with approximately 200 tributaries to quantify the relationship between rainfall and k sn values. Rainfall in these basins ranges from 600 to 1,100 mm ( Figure 10A), and the k sn values had a wide range from 10.5 to 207 m 0.9 . Figure 10A shows that the correlation between the k sn values and precipitation was very weak (R 2 = 0.0719). As such, precipitation did not seem to control the distribution of k sn values.

Sediment Flux
Abrupt changes in drainage area, water discharge, and sediment flux at tributary junctions might trigger knickpoint initiation. This behavior has been suggested in theoretical river incision models that included a dual sediment flux dependence (e.g., Gasparini and Brandon, 2011). In these models, the efficiency of bedrock incision increased with additional sediment flux up until the point when the increasing sediment no longer functioned as a tool to abrade the bed, but rather covered and armored the bed from further incision. The sediment flux in rivers mainly depends on the interplay between the actual total sediment flux (ATSF) and the river sediment capacity (RSC) (Willgoose, 1994;Tucker and Slingerland, 1996;Slingerland et al., 1997;Sklar and Dietrich, 2001;DiBiase et al., 2010). If ATSF < RSC, the materials carried by the river will erode the base of the river and increase the erosion rates (Wang et al., 2019). If ATSF > RSC, the material carried by the river will be deposited on the river bottom, thus protecting the riverbed from further erosion (Whipple, 2002) and lessening the rate at which the knickpoints move headwaters (Wang et al., 2019). As a result, the channel will become flatter and have a low concave.
Knickpoints along the Anning River (Figure 8) are located at the transition from the base of the channel, being characterized by Frontiers in Earth Science | www.frontiersin.org February 2022 | Volume 10 | Article 821367 sediment characterized by bedrock. However, the k sn values across the Central Yunnan subblock are not controlled by sediment flux.
We compared the k sn and concavity values extracted for each class of lithology through a statistical analysis derived from Figure 2. Comparison of the concavity and k sn values showed no linear relationship. In a normalized height-distance coordinate system, the concavity indices are close to a constant and represent similar channel shapes (Whipple and Tucker, 1999;Duvall et al., 2004;Kirby and Whipple, 2012).
The records of sediment load and drainage area were from Qin et al. (2019) and Pan (1997). Channel steepness values are shown in Figure 6.
We also found no systematic change of the k sn values with varying sediment concentrations and the basin area. We obtained the sediment load data (including bed load, suspended load, and solute) from hydrological stations and hydroelectric stations of the Central Yunnan subblock and calculated the k sn values of the channel portions downstream of the knickpoints (Table 1) (Pan, 1997;Qin et al., 2019). The soil erosion modulus of the Jinsha River basin in Yunnan is 1,530 t/km 2 (Huang et al., 2006), and the calculated sediment transport capacity is far greater than the actual sediment transport capacity. We found no obvious relevance between the watershed area and the k sn values in the tributary catchments (R 2 = 0.0182) ( Figure 10B). The calculated sediment transport capacity is far greater than the actual sediment transport. Therefore, k sn should increase with the increase of the mean sediment load, but the trend cannot be seen from Figure 10C. Despite limited data, the sediment loads increased with the watershed area (R 2 = 0.4793) ( Figure 10D).
Therefore, sediment flux did not appear to have dominant control on the k sn value in the channel profile.

Patterns of Rock Uplift
Non-tectonic factors (e.g., lithologic resistance, precipitation, and sediment flux) could not fully explain the k sn distribution. The most active areas of the interpreted uplift are scattered in the region controlled by the NE/SW-trending thrust faults in the north of the Central Yunnan subblock. Therefore, the distribution of the k sn values is likely tectonically controlled.
The Xianshuihe-Xiaojiang Fault accommodates the shortage from the India-Eurasia collision by the left lateral strike-slip motion (e.g., Molnar and Tapponnier, 1975;Molnar et al., 1987;Deng et al., 2002;Xu et al., 2003;Wen et al., 2003). Since 13 Ma, the maximum slip rate of the north part of the Xiaojiang fault system has reached up to 10 mm/annum , and this rate has decreased progressively southward. At the south end of the Xiaojiang Fault, no obvious slip can be found (Wang et al., 1998). In the growth model for the Tibetan Plateau of Tapponnier et al. (1982) and Tapponnier (2001), the strain was mainly localized along the boundary faults of large blocks (Tapponnier, 2001). Fast erosion rates in the adjacent zone of the strike-slip fault belt can be found 2021); however, the k sn values were low on both sides of the strike-slip faults and exhibited little change across the area ( Figure 7A). This may be due to the limited vertical slip and fast strike-slip rate (Molnar and Tapponnier, 1975;Molnar et al., 1987;Deng et al., 2002;Xu et al., 2003) in the strike-slip fault zone (Zhang, 2008;Wu et al., 2014).
Along the transition between the Xianshuihe Fault and Xiaojiang Fault, some active thrust faults with a NE trend were cut by this Xianshuihe-Xiaojiang Fault and propagated southwestward ( Figure 1). Slip vector analyses argued that the sinistral slip rates from west to east across these thrust faults have decreased . Their loss has been considered to be transformed into local crustal shortening perpendicular to the active thrust faults. As a result, the distribution pattern of k sn along the Xiaojiang Fault, obtained in this study, is in good agreement with the thrusting transformation-limited extrusion model of the Tibetan Plateau.
The slope-area regression of the longitudinal profiles revealed that the rock uplift in the interior of the block is lower compared to the north part controlled by a series of NE-NW thrust faults (such as the Jiulong, Muli, and Jinhe-Qinghe thrust faults) where rivers have high k sn values and contain several knickpoints along their profiles, suggesting that the tectonic signals have propagated upstream in the landscape by knickpoint recession. This process resulted in higher indices in the downstream, along trunk and tributary channels Zhang et al., 2020). The lower part of the Jinsha River trunk also showed high channel steepness, and the region east of the Xianshuihe Fault also had high channel steepness, although distant from the thrust faults, which appears to be the products of the knickpoint recession (Zhang et al., 2022).
We postulated that, in the interior of the block, the rivers experienced successive base-level uplifts that have migrated upslope, reaching the headwater (e.g., Clark et al., 2005;Schoenbohm et al., 2006;Royden et al., 2008;Wang et al., 2017). The distribution of lower k sn values at the headwaters, the unperturbed reaches with a tight slope-area regression with low k sn values, suggest that, in the growth model for the Tibetan Plateau of Tapponnier et al. (1982) and Tapponnier (2001), the great crustal shortening between the Indian and Eurasian plates was mainly absorbed by a series of strike-slip fault systems and the related thrust faults.

CONCLUSION
In this study, we calculated the distribution pattern of normalized channel steepness by the stream power incision model of 300 rivers in the Central Yunnan subblock. The distribution pattern has been clearly observed to represent a general increase in value from north to south in this subblock. Higher k sn values were associated with the NE/SW-trending Cenozoic thrust faults (including the Muli Fault, Lijiang-Xiaojinhe Fault, and Jinhe-Qihe Fault). Lower k sn values were found on both sides of the Xiaojiang fault zone and the Red River fault zone. The influences of non-tectonic factors were discussed and excluded. We suggested that very weak correlations were found between the steepness indices and the lithology, precipitation, sediment flux, or channel concavity indices. Along the Xiaojiang strike-slip fault and the interior subblock, the uplift rate was slower, while the northern part had uplifted faster and was controlled by thrust fault systems. The channel steepness increased gradually from south to north. Thus, the distribution pattern of the K sn index within the Central Yunnan subblock provides notable support for the argument for the thrusting transformation-limited extrusion model of the Tibetan Plateau.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
LY and YD conceived of the presented idea. YD and WZ developed the theoretical framework. LY, DZ, and DW developed the theory, analyzed the data, and performed the computations. HY, YR, and JL verified the analytical methods. All authors discussed the results and contributed to the final manuscript.

FUNDING
This study is supported by the National Natural Science Foundation of China (nos. 41802215 and 41762017).