3.1 Paleoflood deposits
SWDs are fine-grained sediments carried in suspension in high energy floods that are deposited in areas of low velocity, such as embayments, alcoves, and tributary junctions, and are commonly used as paleo-stage indicators (PSIs) 18,19. One set of SWDs (N30°8'38.67", E95°0'51.15") was found in the lee of a river bend 8.3 km downstream of the 2000 AD dam near Shuangyu village (Fig. 1C, Fig. 2A). The SWDs underly a 1.4 m thick landslide deposit at 52 m above the river level. The sequence is nearly horizontally bedded and comprises four units (Fig. 2B-C): (i) c. 20 cm thick mixture of coarse sand, fine gravel, and diamicton, possibly reworked; (ii) 160 cm of uniform gray oxidized fine sand; (iii) 70 cm thick silt unit; (iv) 50 cm of coarse sand.
Three charcoal samples were collected from the fine sand unit and one from the coarse sand unit. Radiocarbon ages of the charcoal pieces range from 6.7 ka to 7.1 ka BP by radiocarbon dating at the Beta Analytic laboratory, USA (Table 1). Four samples were also collected for OSL dating at the OSL laboratory of the Institute of Mountain Hazards and Environment, Chinese Academy of Sciences. The OSL dating results are shown in Table 2. Additionally, OSL dating of samples from each of the lower three units was performed on a Lexsyg Research automatic TL/OSL instrument at the Institute of Mountain Hazards and Environment, CAS. The sample from the silt unit gave an OSL age of 7.6 ± 0.7 ka, which is consistent with the radiocarbon age. The other two samples gave ages of 18.5 ± 1.7 ka (fine sand) and 15.3 ± 1.4 ka (coarse sand) (Fig. 2D); the difference may be due to their larger grain size, or lack of bleaching in a catastrophic flood event. Combining the OSL and radiocarbon dating results gives a probable age of ~ 7 ka BP for the Shuangyu SWD.
Table 1
Radiocarbon dating result for samples from shuangyu slackwater deposit
Sample | Material analyzed | Percent modern carbon | Modern carbon fraction | Age (BP) |
SC1 | Charred material | 41.73 +/- 0.16 | 0.4173 +/- 0.0016 | 7020 ± 30 |
SC2 | Organic sediment | 42.94 +/- 0.16 | 0.4294 +/- 0.0016 | 6790 ± 30 |
SC3 | Charred material | 43.10 +/- 0.16 | 0.4310 +/- 0.0016 | 6760 ± 30 |
SC4 | Charred material | 42.89 +/- 0.16 | 0.4289 +/- 0.0016 | 6800 ± 30 |
Table 2
Sample dose rate and optically stimulate dluminescence dating results for shuangyu slackwater deposit
Sample | Depth (m) | K (%) | Th (10− 6) | U (10− 6) | Dosage rate (Gy/ka) | Number of test pieces | De/Gy | OSL age (ka) |
SO1 | 2.3 | 3.07 ± 0.04 | 14.24 ± 0.70 | 2.44 ± 0.40 | 3.88 ± 0.28 | 8a + 12b | 71.6 ± 4.21 | 18.5 ± 1.7 |
SO2 | - | - | - | - | - | - | - | - |
SO3 | 3.4 | 3.22 ± 0.04 | 22.97 ± 0.80 | 1.74 ± 0.30 | 4.61 ± 0.34 | 8a + 12b | 35.1 ± 1.83 | 7.6 ± 0.7 |
SO4 | 3.9 | 2.36 ± 0.04 | 25.96 ± 0.80 | 1.46 ± 0.30 | 4.16 ± 0.30 | 8a + 12b | 63.8 ± 3.83 | 15.3 ± 1.4 |
a the number of test pieces using Single Aliquot Regenerative-dose method b the number of test pieces using Standardised Growth Curve method |
In addition, we found lacustrine deposits (N 30°09'05", E 94°59'42") at the mouth of Dayi creek, a Yigong’s tributary, 2 km upstream of the Shuangyu SWDs (Fig. 1C). The 2.2m parallel lacustrine lamination with yellow-brown silt is elevated c. 40.0 m above the adjacent floodplain. It is capped by 4.3 m debris-flow accumulation composed of angular and sub-angular gravels and boulders. A 1.5 m no-bedding mixture with sand and gravels is exposed under the lamination (Fig. 3A). Two pieces of charcoal were taken from the silt deposit and radiocarbon dated at Beta Analytic laboratory, USA (Fig. 3B). The 14C ages are determined as 6.3k Bp and 6.6k Bp, which means the Dayi lacustrine deposits are a little later than the Shuangyu SWD. The sediment sequence is located at the mouth of the branch Dayi creek on the right bank of the Yigong river. The bank is a concave bend behind a narrow reach, where the transport capacity of the 7k Bp superflood decreased and large volume of sediment that carried by the superflood stopped here. We speculate the paleoflood deposits jammed the Dayi’s outlet, forming a small temporary dammed lake in the creek. Inflow silt had deposited in the dammed lake and produced the lacustrine lamination. The upper debris-flow accumulation on the lacustrine deposits implies that it is very likely a large-scale debris flow happened in Dayi and broke out the dammed lake.
3.2 The modern flood SWDs
Another SWD was found on the opposite bank of Jiazhong village, a hydraulically sheltered area 3.9 km downstream of the 2000AD breached dam (Fig. 1C). The Jiazhong SWD is located within a cove on the river right bank, and its surface is a grassland about 6.0 m higher than the floodplains on the both banks (Fig. 4A). The excavated section is a 1.2 m thick sequence of fine to medium-grained sand capped by a ~ 10 cm thick sandy loam at ~ 21.0 m above the river level. The dark gray fine sand is interbedded with three units of ~ 10 cm thick grey medium sand and mingled with sub-angular gravels (Fig. 4B). The 0–70 cm upper part is nearly parallel laminated, but the lower part inclines to the river at an angle of ~ 5°, indicating an original local gradient and flow direction. A piece of charcoal collected in the middle of the SWD section was dated at the Beta Analytic laboratory. The measured percent modern carbon is 100.12 ± 0.37 and the radiocarbon age ranges from 1880 to 1956 AD with 87.3% probability. Monsoon seasonal floods of ~ 2100 m3/s peak unlikely reach the location of Jiazhong SWD. The upmost sandy loam is very thin, which means it formed in a relatively short period. Moreover, the SWD’s elevation is close to the 2000AD flood deposits next to it (Fig. 4A). Therefore, we interpret the Jiazhong SWD as the product of the 2000 outburst flood. The Jiazhong SWD’s level can be used as the flood stage indicator of the 2000 event.
3.3 Peak flow reconstruction
The well documented 2000 AD flood can be used as a benchmark to aid reconstruction of historic events. The peak 2000 AD discharge at Tongmai Bridge has been estimated as 126,400 m3/s 13, 120,000 m3/s 12, and 124,000 m3/s 20, with a peak water depth of 52 m. We compare peak discharge calculations for the 2000 AD flood using 15 empirical formulas and know data on breach depth, dam height, volume of released water, and impounded water volume (Table 3) 21. The MacDonald and Langridge-Monopolis (MLM) 22 formula produced the closest estimate to the measured discharge, at ~ 130,000 m3/s. So the MLM formula is used to estimate the peak discharge of the 1902 AD event, using data on breach height and released water volume from Delaney and Evans 15 (Table 4). The resulting discharge of ~ 168,000 m3/s is more than 50 times that of a large monsoon season flood.
Table 3
Estimates of peak discharge for the outburst flood generated by the 2000 ad yigong dam failure using 15 empirical models listed in Liu et al. 21
Author | Model* | Publication date | Peak discharge (m3/s) |
Kirkpatrick | Qp =1.268(Hw+0.3)2.5 | 1977 | 28835 |
SCS | Qp =16.6Hw1.85 | 1981 | 27528 |
USBR | Qp =19.1Hw1.85 | 1988 | 31674 |
USBR | Qp =48Hw1.85 | 1988 | 32963 |
Hagen | Qp =0.54(Vs-Hd)0.5 | 1982 | 24239 |
Singh and Snorrason | Qp =13.4Hd1.89 | 1984 | 26084 |
Singh and Snorrason | Qp =1.776Vs0.47 | 1984 | 41922 |
MacDonald and Langridge-Monopolis | Qp =3.85(HwVw)0.41 | 1984 | 129944 |
Costa | Qp =1.122Vs0.57 | | 225643 |
Costa | Qp =0.981(VsHd)0.42 | 1985 | 42698 |
Costa | Qp =2.634(VsHd)0.44 | 1988 | 2998208 |
Evens | Qp =0.72Vw0.53 | 1986 | 61459 |
Froechlich | Qp =0.607Hw1.24Vw0.295 | 1995 | 48528 |
Webby | Qp =0.0443g0.5Vw0.365Hd1.4 | 1996 | 94314 |
* Where Qp is outburst flood peak discharge, Vw is the water released (m3), Vs is barrier lake volume, Hw is breach depth, and Hd is dam height. In this study, we only considered complete dam breach, thus, Vw is the same as Vs and Hw is the same as Hd. |
Table 4
Basic parameters of the yigong barrier lakes in 1900 ad and 2000 AD
Year | Water depth (m) | Lake area (km2) | Lake volume* (Gm3) | Peak discharge† (m3/s) |
2000 | 55 | 48.93 | 2.015 | 129,944 |
1900 | 73 | 54.08 | 2.838 | 167,943 |
* Volume was calculated using \(V={A}^{3.424}*3304.5\) (Delaney and Evans, 2015), where A is the area of impounded water |
†Peak discharge was calculated using \({Q}_{p}=3.85*(H*V{)}^{0.41}\) (MacDonald and Langridge-Monopolis, 1984), where H is barrier lake water depth and V is lake volume. |
SWD surface elevation is widely used in paleoflood hydraulic reconstruction as a PSI18,23. The most commonly used paleoflood peak discharge estimation technique is the 1-D step-backwater method18,24. We applied the method to the 2000AD and 7 ka BP events using HEC-RAS 5.0.7 software25, assuming a subcritical flow regime. Contour lines interpolated from the ALOS DEM (in 2009 with12.5×12.5 m spatial resolution) correspond well with channel shape and location in the ETM + image for 1999, so provide an acceptable record of pre-2000 AD topography. The starting cross section for the step-backwater calculation was set at the Tongmai Bridge bedrock section. A total of 97 cross sections were extracted along the 19.2 km reach with an average spacing of 200 m using the RAS Mapper of HEC-RAS 5.0.7 (Fig. 5A).
Expansion and contraction coefficients of 0.1 and 0.3 were specified, based on channel width. Manning’s n roughness coefficient was calibrated using the peak discharge (126,000 m3/s) and water depth (52 m) of the 2000 AD flood. Computed water surface elevation at the Tongmai and Jiazhong cross sections corresponded well with the 2000 AD flood level when Manning’s n was set as 0.03 and 0.035 for river channel and floodplain, respectively. Sensitivity tests in the model show that a 25% variation in Manning’s n results in a maximum difference of 0.13% in the discharge result. Using these values of the coefficients, a discharge of 225,000 m3/s provides the best approximation for the Shuangyu deposits (Fig. 5B). The calculated values of the velocity and the flow area of the 2000AD flood at Tongmai bridge are 15.7 m/s and 8350 m2, closed to those estimated by Delaney and Evans15. For the 7 ka BP flood, the velocity and the flow area at Tongmai bridge are 19.5 m/s and 12360 m2. The paleo-flood with such a magnitude can yield a bed shear stress of 5 kPa, and move large boulders up to 5 ~ 6 m in diameter8.