Uniformitarianism also called comparative-historical method is an important paradigm in the process of geological research (Manning 2016; Windley 1993). We can deduce the conditions, processes and characteristics of ancient geological events by using the existing laws of geological action, through the geological phenomena and results left over by various geological events. 'The Present is the key to the Past' was the uniformitarian paradigm(Hutton 1788; Windley 1993) and nature is the best geological museum and laboratory, so field investigation is the premise and foundation of geological research and it is the traditional thinking method of geology. Therefore, a uniformitarian approach with the combination of modern scientific and technological is applied to improve the analysis of river blocking event.
For an ancient river blocking event, there are many nailing researches. A method called the ' trinity ' combination of residual landslide dams, upstream lacustrine sediments and downstream break-outburst sediments was proposed (Chen and Cui 2015; Li et al. 2020). In short, the fact that the river is blocked has basically reached a consensus. My main research purpose is to supplement the method and further study analysis method due to the mutual influence of these landslide damming of river events. And the method is reliably suitable for landslide dam classified by Costa and Schuster (1988).
3.1 Analysis of an independent landslide damming of river event
For one landslide damming of river event, some traces would be left near the location where the river was blocked. Considering a fact that there would be less key direct information left in some river blocking events, it is necessary to analyze and summarize from different dimensions and perspectives.
3.1.1 Evidence from Remote Sensing Interpretation
The occurrence of the river blocking event requires the joint action of two aspects. The first is the stream channel, the other is blocking dam. For the remaining Quaternary ancient river blocking event, the square is generally large. Therefore, we can search along the river to find the landslide dam body on both sides of the river on remote sensing images (Fig. 3), which is the potential evidence of blocking the river. Therefore, there will be greater changes in topography, including the change of bank slope morphology and the phenomenon of river diversion. In Google image, we can roughly circle the scope of the remaining dam block, find the source of blocking material. At the same time, we can also use DEM data for 3D modeling in GIS software such as ARCGIS to obtain some relative geometric parameters, such as the accumulation area, the accumulation length along the river, the accumulation width perpendicular to the river direction and the accumulation thickness of the dam body at the collapse.
3.1.2 Evidence from morphology
After determining the approximate location of river blocking, the phenomenon of river blocking can be more accurately identified by detailed field investigation. Taking on-site photos of Gangda as an example (Fig. 4), we can see that the dam bodies on both sides of the Jinsha River have geometric continuity and a good curve can be obviously observed by abstracting it into geometry. In the photograph taken, the geometric size of the river blocking body is well identified and recognized. At this time, important information such as the maximum thickness, average thickness of the accumulation body, and the geometric position and size of the breach are recorded, which are of great significance for the verification of the inversion results of the numerical simulation.
3.1.3 Evidences from geology
For river blocking dam that we can find source area, geological continuity usually is maintained. That is, the lithology of the dam is similar to the lithology of the material source area. Similarly, we can look for material sources on this basis. When the lithology of the bank slopes on both sides is inconsistent, and the lithology of the residual dam on both sides is consistent, we can infer the occurrence of river blockage and determine the source of material. However, in turn, when both sides of the bank slope has the same lithology, even if the internal lithology of the residual dam is consistent, it can not be inferred that the material comes from one bank, which requires further analysis combined with remote sensing interpretation.
3.1.4 Evidence from sedimentology
In the study section, there are a lot of fine-grained sediments (Fig. 5). These lacustrine sediments not only directly reflect the sedimentary environment, but also reflect the hydrodynamic conditions of the transport medium. In order to determine the grain size characteristics of the lake sediments, the samples were taken from the lake sediments during the field investigation. As the lacustrine sediments particles are small, they can be all brought back for grain size analysis to obtain cumulative curve of particle size. Then we can get particle size characteristic parameters (Table 1). Firstly, according to Eli law, the diameter of the bed load moving on the riverbed is proportional to the square of the flow velocity (Eq. 1). In this study, we selected the maximum d50 value in Table 1 as the calculation data. By the following assignment, d=d50max=2.6×10−5, rS = 2700, r=1000, g=9.8, we calculated V=1.73×10−2 m/s, which is far less than the normal velocity of Jinsha River. Secondly, according to Stokes formula, similarly, 0.026 mm is chosen as the calculation particle size, and the average temperature of Jinsha River is selected as 9.2°C (according to Batang Hydrological Station) to select the particle size calculation coefficient. The sedimentation velocity v = 8.146×10−7 m /s is calculated. and the setting time is about 568 days when the settling height is 40 m. Through the above rough calculation, it is concluded that a certain thickness of fine sediment layer on the upstream of landslide body must be formed in a stable still environment where the river is blocked. Therefore, the existence of the lacustrine deposits layer can effectively reveal the river blocking event.
$$d=\frac{rk}{2gf({r}_{s}-r)}·{V}^{2}$$
1
where V is the Velocity acting on the surface of sediment particles, m/s; d is the diameter of sediment particles, m;rS is the density of sediment particles, kg/m3; r is the density of water, kg/cm3; f is the coefficient of friction; g is the acceleration of gravity, 9.8 m/s2.
$$v=\frac{2}{9}·\frac{\left({\rho }_{S}-{\rho }_{W}\right)g}{\eta }·{r}^{2}$$
2
where v is the sedimentation velocity of soil particles, cm/s; r is the radius of soil particles, cm;ρS is the density of solid particles, g/cm3; ρw is the density of water, g/cm3; η is the coefficient of dynamic viscosity of water, Pa·s; g is the acceleration of gravity, 980 cm/s2.
Table 1
Characteristic Parameters of Cumulative Percentage Curve of lacustrine sediments
Sample | Effective size | Mean size | Control size | d30/mm |
d10/mm | d50/mm | d60/mm |
S1 | 0.0050 | 0.026 | 0.032 | 0.015 |
S2 | 0.0029 | 0.015 | 0.019 | 0.0064 |
S3 | 0.0090 | 0.03 | 0.036 | 0.021 |
S4 | 0.0016 | 0.01 | 0.014 | 0.0043 |
S5 | 0.0043 | 0.019 | 0.026 | 0.012 |
S6 | 0.0027 | 0.024 | 0.029 | 0.0048 |
S7 | 0.0042 | 0.017 | 0.018 | 0.0087 |
S8 | 0.0020 | 0.009 | 0.013 | 0.0045 |
S9 | 0.0015 | 0.006 | 0.0088 | 0.0019 |
S10 | 0.0023 | 0.011 | 0.014 | 0.0051 |
Note: Location of the sample is showed in Fig. 2 as Snumber |
3.1.5 Evidence from break-outburst sediments
The dam break-outburst sediment is also one record of landslide dammed lake and it is also an important way to understand the dam-break process, which is usually difficult to find in an old river blocking event. According to the particle size of dam break-outburst sediments, the flood parameters at that time can be obtained by back analysis (Chen and Cui 2015; Ma et al. 2018). Besides, reasonable analysis of dam break-outburst sediments can also be made to determine the sequence of river blocking events
To summarize, for a complete landslide blocking the river event, mainly from the beginning of landslide to the end of dam break, we can mainly investigate, describe and summarize from the above five aspects. Among them, lots of lacustrine deposits is the most critical and convincing evidence for long-term existence of river blocking.
3.2 Analysis of interdependence landslide damming of river events
The characteristic of the study reach is that there have been many river blocking events. Therefore, more data are needed to explain whether these river blocking events interact with each other, which may be inconsistent with or even contrary to the results obtained from a single analysis of river blocking. These problems will mainly affect the judgment of river blocking time and thus affect the order of river blocking events, so more means and evidence are needed to explain the overall process of river blocking events. For example, if the WDL dammed lake formed early and lasted long, then the dating age of the lacustrine sediments is likely to indicate WDL rather than other dams upstream. Besides, considering a long time existing dammed lake, the effect of water on the genesis of other landslide dams shall be considered in numerical simulation even in such dry and rainy areas
3.2.1 Elevation inference
In terms of elevation, there is a rule that Edam (the elevation at stable formation of the dammed lake) ≥ Elake (the highest elevation of the dammed lake) ≥ Elacustrine (the highest elevation of the lacustrine sediments). Since the ancient barrier lake has disappeared, we can obtain information that the present shape of the dam and the highest retention elevation of the lacustrine sediments. If the lacustrine sediments and landslide dam belong to the same river blocking event, then the highest elevation of the former cannot be higher than that of the latter. If not, the lacustrine sediments would not be formed by the dam. The principle is to rely on the geological boundary.
For Edam, we obtain the profile chart according to the DEM, then the original dam shape is roughly outlined reference to the form of the Baige landslide occurred in the upstream according to Feng et al. (2019). Finally, the reasonable elevation value (Fig. 6) is deduced. For Elacustrine, we use lacustrine sediments elevation recorded on field investigation by comparing the relatively highest point (Table 2)
Table 2
elevation of dam crest and Maximum elevation of lacustrine deposits upstream the dam (m)
dam | Minimum elevation of dam crest (m) | Maximum elevation of lacustrine deposits upstream the dam (m) | dam | Minimum elevation of dam crest (m) | Maximum elevation of lacustrine deposits upstream the dam (m) |
WDL I | 2500 | 2426 | SWL | 2394 | 2430 |
WDL II | 2463 | SDX | 2423 | 2445 |
RCR | 2444 | 2442 | GD | 2455 | 2446 |
Based on the above results, we can analyze that WDL I, II and RCR river blocking events have a wide range of influence, and other river blocking events might have been affected. Therefore, the influence of each other must be considered in sampling analysis.
3.2.2 Dating
For several river blocking events, dating is a direct method to determine the sequence. Especially, multi-method dating campaigns enhance our understanding of the beginning and end of the river blocking event However, due to the limitations of objective conditions, such as the error of test methods, the lack of availability of dating samples, insufficient funds and the uncertainty of whether the obtained samples have been in the accumulation body or later mixed in, it is often necessary to analyze them from multiple aspects use different methods(Fan et al. 2021). Although more direct and high precision evidence is the 14C dating age of the dam material, 14C dating requires high wood charcoal samples. First, this section belongs to the dry and hot valley, and there is less vegetation on both sides of the river, so the sample collection is very difficult. Second, the source of samples can 't be guaranteed, so data may be deceptive. Therefore, the dating data of lacustrine sediments can often be used to assist the explanation. At present, the optically stimulated luminescence (OSL) dating method is widely used.
In a single river blocking event, it is reasonable to infer that the normal sequence is formed later in the upper than lower part, in other words, the bottom is older than the top. Through field investigation, there is no sequence inversion caused by tectonic movement. Besides, according to the stratigraphic relationships between the dam body and the lacustrine sediments, a relative age for the dam can be concluded.
However, for several river blocking events, due to the influence of river geomorphology (Fig. 7), the analysis of the dating results of the lacustrine sediments can be divided into the following situations:
Situations 1, the formation of lacustrine sediments in the same river blockage with the same bottom baseline (Fig. 8(a)). In general, through the detailed investigation on the site, the relative bottom and the relative top of the lacustrine sediments are found. Through the time difference between the top and bottom, we can roughly infer the duration of the river blocking.
Situations 2, in the same river blockage with the different bottom baseline (Fig. 8(b)). Through data processing, let the bottom at the same elevation, and then according to the situation 1 analysis.
Situations 3, in different periods of river blocking events with the same bottom baseline (Fig. 8(c), Dam II with Dam III). The age of sediments in the bottom is most likely difference. Data analysis can often form two series.
Situations 4, in different periods of river blocking events with the different bottom baseline (Fig. 8(c), Dam I with Dam II). First according to situations 2, then according to situations 3. The specific method will be described below.
Of course, these situations only consider the general case, not all.
We first assume that several river blocking events are independent of each other, and then we make the chart (Fig. 9) according to our dating data and others (Chen et al. 2013; Chen et al. 2018). The analysis results of these data are obviously contrary to the assumption preceding part of the text that there is a negative linear correlation between years and elevation. So these data are not independent and need to be processed further. According to the results of dating data and its error, the frequency statistics is carried out with 100 years interval, and four peaks are found (Fig. 10). Under the guidance of no clear experimental purpose, the results of random sampling are related to the distribution of samples, so it can be considered that these dating data roughly represent the four river blocking events.
Therefore, we processed the data as follows:
First, the data were classified by river sections according to their locations.
Then, in each category, the classification is further carried out according to the linear relationship.
Last, the classified data are reasonably segmented combined with the results of the time-frequency histogram, and the classification results are processed into the same baseline to obtain the final results (Fig. 11).
In summary, compared with the results of 14C dating data (Chen et al. 2013), the results of WDG II are very close, indicating that the analysis method is suitable. At the same time, by comparing the two results, the approximate time of river blocking in this reach can be obtained. The river blocking occurred in WDL I reach about 6300 years ago, and the duration of river blocking is about 1000 years; The river blocking occurred in WDL II reach about 1900 years ago, and the duration is about 400-840 years; The river blocking occurred in RCR reach around 1300 – 1400 years ago, and the duration is about 190 – 370 years; The river blocking occurred in SWL reach about 1370 ago, and the river blocking was relatively short; The river blocking occurred in SDX about 750 years ago (a time away from today) and 510 (a time close to today), and the river blocking duration was about 100-110 years; The GD river blocking time was about 900 years ago, and the river blocking duration was uncertain because of less data.
3.2.3 Interpretation of geological phenomena
Good analysis results should be able to reasonably explain the observed phenomenon. We explain the field investigation on this basis that the above results are correct.
First, considering landslide dam, GD and SDX residual dam body are relatively complete, the distance between the right bank dam and left bank is short, and the collapse occurs at the cross section, which indicates that the dam body exists for a short time.
Second, considering the characteristics of dam break -outburst sediments. In the upstream of SWL and SDX, the sedimentary layers under different hydrodynamic environments are found, and the maximum number of accumulation layers in SWL (Fig. 12) is more than that in SDX. This is because the formation of SWL is earlier than that of SDX and GD, so it is affected by the two river blocking events.
Third, from the point of view of lacustrine sediments, the stratification of lacustrine sediments should be more obvious and nearly horizontal in the general long-term stable water environment, while the stratification of lacustrine sediments found in SWL, SDX and GD deposits is not obvious, indicating that the water environment is not long-term stable. It was also found that the bedding of the lacustrine sediments was inclined, indicating that the sediments formed before landslides (Fig. 13). The horizontal continuous lacustrine sediments in the WDL-SWL reach up to hundreds of meters are splendidly more obvious than that in the SWL-GD reach. Therefore, according to the results of 3.2.2, the sequence of each river blocking is relatively reasonable, which is consistent with the results of field investigation.
3.2.4 Knickpoint
Fluvial response causing by landslide dam may theoretically influence sediment yield, channel planform, cross-section, gradient, or bed configuration(Korup et al. 2006). Among these potential response variables, we are interested in long-term fluvial response, especially in channel gradient (Fig. 14). By finding the turning point between the gentle gradient and the steep gradient, knickpoints can be recognized in the river long profile(Safran et al. 2015).
It can be seen from the Fig. 14 that the river blocking event not only generates knickpoints at the dam site, but also generates knickpoint s at the upstream. Moreover, the results of the crack point analysis are basically consistent with the above dating data. The blocking age of WDL I and WDL II is long, and the change of river channel is obvious; The SWL blocking event lasted for a small time and had little effect on the river. In terms of elevation, the height fitted by the profile shape of the dam body seems to be conservative, such as the height of the WDL I may initially reach 2600 m, and there are multiple dam breaking events.
3.2.5 Deformation analysis
Usually, the longer the accumulation body exists, the more tectonic activities it experiences, causing relative instability, vulnerable to erosion and gradually disappear. This explains why the longer the time the river blocking event is, the harder we find the remaining dam.
In this paper, the surface deformation of the residual dam body was analyzed through the SARscape module of ENVI 5.3 using remote sensing image from January 2018 to December 2020. Then the average deformation rate and the response to the flood from Baige barrier lake were obtained of both the whole dam body and part of the whole area along the river (Fig. 15). By analyzing Fig. 15(a) and (b), we could draw such a conclusion that the dam bodies are in denudation state, but the denudation rate is small, which is consistent with the situation that the accumulation dam can exist for thousands of years. Comparing GD, SWL, WDL I and WDL II in the figure, the older the age, the greater the deformation rate; SDX is abnormal in Fig. 15(a) because the construction camp of Suwalong Hydropower Station is built on it, and it has been artificially transformed. While in Fig. 15(b) of the control, the part of the whole dam along the river is slightly uplifted mainly due to the uplift of the region with around 5 mm/yr rate(Zhan et al. 2018). The RCR river section is relatively small because this reach may be a relatively "sedimentary area" in the whole study reach. As can be seen from Fig. 15(c) and (d), flood is not the main reason for the disappearance of the dam. After the flood, the dam body is in a state of accumulation rather than erosion, and the accumulation thickness of RCR is the largest in the whole river section of the study area, which may be related to the existence of a large number of continuous lacustrine sediments in the upstream of RCR; Moreover, comparing Fig. 15 (a) and (b), it can be drawn that the erosion rate along the river section of RCR is lower than the whole erosion rate, because the material on the upper part of the dam body is eroded and stripped from the original position and then deposited in the lower part of the dam body along the river.
If the above deformation can only explain the result of a period of time, the landform of the accumulation dam body is the result of long-term evolution. Through the rough measurement of two relatively distant points on both sides of the residual dam body by Google map, the erosion section length perpendicular to the river of WDL I is about 1100 m, WDL II is about 500 m, RCR is about 350 m, SDX is about 190m, and GD is about 200 m, these data are positively correlated with the ages of the accumulation dams.
In conclusion, the evidence of deformation analysis can indirectly prove that the age of the accumulation dam body is reliable.