**4. Results and Discussions**

The experimental results and conditions are summarized in Table 1. In Table 1, *P* is the probability of a flood event; *Q* is the prototype flood event discharge; *t* is the prototype flood event duration; *z* is the prototype grade control datum elevation; *i* is the measured average bed slope (model) upstream of the GCS; *dp* is the maximum general scour depth at the site of 105th Provincial Highway Bridge in terms of *z* = 544 m (i.e., the bed elevation before the Wenchuan Earthquake). In this study, as the detailed hydrograph is not available, the peak flood rate was applied during each flood event. As the flood in this area is caused by intense rain, the flood rising and recession periods are very short and can be neglected. In order to avoid the impacts of local scour and constriction scour on the bed profile, the general scour depth at bridge site was measured upstream of the scour area of the bridge foundation.


**Table 1.** Summary of the experimental conditions and results.

## *4.1. Talweg Profile Upstream of GCS*

Figure 4 shows the talweg profile upstream of the GCS of each GCD *z* for different flood events. Figure 4 indicates that, for each test, the bed elevation immediately upstream of the GCS is approximately equal to the GCS crest (i.e., GCD). For each *z*, the upstream talweg elevation is lower with a lower *P* (higher discharge, *Q*). This is because a higher flowrate has a greater capacity for sediment transport, causing more erosion on the bed. For *z* = 539 m, the bed in −150 m < *x* <0 is flatter with a higher *Q*, as some of the sediment driven by the approach flow is blocked by the GCS, resulting in an aggraded bed near the GCS. For *z* = 533 m and *z* = 527 m, the aggradation in −150 m < *x* < 0 disappears. This is because, for a lower GCD, the sediment above the GCD is flushed downstream over the GCS without any blockage. Figure 4 also indicates that, for a lower *z*, the difference in the talweg profile caused by increasing *Q* is larger. The bed incision due to GCD drop can be affected by two factors: (i) the approach flow capacity for sediment transport; (ii) GCS blockage. The bed is stabilized when these two factors reach a balance. As the GCS only blocks the sediment when the GCD is above the bed, for a lower GCD, factors (i) and (ii) reach the balance at a lower bed level.

**Figure 4.** Upstream talweg profile after different flood events for *z* = 539 m (**a**), *z* = 533 m (**b**) and *z* = 527 m (**c**).

Figure 5 indicates that for each *Q*, the upstream talweg elevation decreases with decreasing *z*. For *P* = 50%, there is an abrupt steepening in the talweg profile for *z* = 533 m and 527 m at *x* ≈ 200 m and *x* ≈ 300 m, respectively. For *P* ≤ 10%, the abrupt steepening in the talweg profile for *z* = 533 m and 527 m disappears. This is because the bed degradation induced by the drop in GCD begins as a head cut erosion process from *x* = 0. For small discharges (*P* ≥ 20%), the sediment transport rate is slow and is unable to fill the head cut erosion area during the flood event. For high discharges (*P* ≤ 10%), the upstream sediment transport rate is high enough to fill the head cut erosion area within the duration of the flood. Figure 5 also indicates that for *P* ≤ 3.3%, the talweg profile of −200 m < *x* <0 is flatter than that of *x* < −200 m for *z* = 527 m. This is because the grade control datum protrudes high enough above the upstream river bed to block the sediment from the upstream, inducing an aggraded bed near the GCS.

**Figure 5.** Upstream talweg profile of each GCDs *z* for *P* = 50% (**a**), *P* = 20% (**b**), *P* = 10% (**c**), *P* = 5% (**d**), *P* = 3.3% (**e**) and *P* = 1% (**f**).

Figure 6 highlights the dependence of the average upstream bed gradient *i* on the flood discharge *Q*. Figure 6 shows that *i* is greater with a lower *z*. As the GCD drop enlarges the elevation difference between *x* = 1300 m and *x* = 0, the bed tends to develop to be steeper. Figure 6 also indicates that *i* is approximately independent of *Q* for *z* = 539 m, but decreases with *Q* for *z* = 533 m and *z* = 527 m. As shown in Figures 4 and 5, the final bed elevation at *x* = 0 is fixed at the GCD. Thus, the increasing *Q* erodes more sediment from the upstream bed, resulting in a lower upstream bed level and a smaller average gradient.

**Figure 6.** Dependence of average talweg gradient *i* on discharge *Q* for different GCDs *z*.

#### *4.2. Transverse Profile Upstream of GCS*

Figure 7 indicates that, for each *z*, the channel cross-section is deeper and wider with a higher *Q*. This is because the increase in *Q* increases the sediment transport capacity of the flow, resulting in greater erosion of the riverbed and riverbanks. Figure 7 also shows that, for a lower *z*, the difference in the cross-section depth and width caused by increasing *Q* is greater. More specifically, as shown in Figure 8, the channel cross-section of each discharge is wider and deeper for a lower *z*. The GCD drop can cause significant bed incision as shown in Figures 4 and 5, inducing mass failures in the riverbank as the incised bed can not support the bank material. Our experimental observations also confirm that, for a lower GCD, more mass failure occurs in the bank.

**Figure 7.** Transverse profile of different flood events (*P* = 1%–50%) at *x* = −800 m for *z* = 539 m (**a**), *z* = 533 m (**b**) and *z* = 527 m (**c**).

**Figure 8.** *Cont*.

**Figure 8.** Transverse profile of different GCDs *z* for *P* = 50% (**a**), *P* = 20% (**b**), *P* = 10% (**c**), *P* = 5% (**d**), *P* = 3.3% (**e**) and *P* = 1% (**f**).

#### *4.3. General Scour Depth at the 105th Provincial Highway Bridge*

Figure 9 highlights the dependence of the maximum general scour depth *dp* at the 105th Provincial Highway Bridge on the discharge *Q* for different GCDs. In Figure 9, for *z* = 539 m, with the help of the GCS, *dp* is approximately independent of *Q*. For a lower *z*, even very small *Q* (*P* = 50%) can cause serious general scour at the bridge site (*dp* ≈ 9.4 m and 14.8 m). For *z* = 527 m, i.e., the GCS is completely removed, the *dp* can reach up to 17.6 m. For the new GCS design plan *z* = 533 m, the minimum *dp* is 9.4 m. Thus, this study suggests building a new GCS with *z* ≥ 539 m or a new bridge with much deeper foundations.

**Figure 9.** Dependence of the maximum general scour depth *dp* at the 105th Provincial Highway Bridge on discharge *Q* for different GCDs *z*.
