El Niño Southern Oscillation Influences River Bank Erosion on the Lower Mekong River

Here we use an excess shear stress model of the form E = k(τ-τc), where E is the erosion rate per unit time and unit bank area, τ is the boundary shear stress, and k and τc are erodibility parameters (erodibility coefficient, k, and critical shear stress, τc) to estimate river bank erosion at study sites (Ang Nyay and Pakse, both in Laos) on the Lower Mekong River (LMR). The model is parameterized using field measurements and analytical modelling. Bank erodibility parameters (k and τc) are estimated in situ using a Cohesive Strength Meter [Tolhurst et al., 1999, Estuarine, Coastal & Shelf Sci., 49, 281-294]. A shear stress partitioning model [Kean and Smith, 2006, J. Geophys. Res., 111(4), F04009, doi:10.1029/2006JF000467] is used to characterise τ. Input data for the Kean and Smith model is obtained via surveys of river bank topography (to characterise bank roughness) and acoustic Doppler current profiler measurements of near-bank flow velocities over a wide range of flows, thus enabling τ to be linked to flow discharge. In turn, this enables use of flow data from gauging stations to reconstruct multi-decadal time series of annual bank erosion rates. In this model variations in annual rates of bank erosion are induced by variations in the LMR’s flow discharge regime. Mann-Kendal tests reveal a slight, but statistically significant (at 95% confidence), downward trend of simulated bank erosion rate at Pakse (mean erosion rate 0.55±0.15 m/yr, declining at 0.002 m/yr during 1923-2007); but no significant trend about the long term mean of 0.30±0.08 m/yr further upstream at Ang Nyay (1913-2007). This indicates that the hydrological regime of the LMR above Ang Nyay has been relatively stable (at decadal scales) during the last ~90 years, but there is evidence of a minor reduction in runoff (more specifically, the annual volume of runoff over the threshold required to initiate bank erosion) from the tributaries that feed the LMR between Ang Nyay and Pakse. However, both simulated time series exhibit quasi-periodic oscillations about these means. The Mekong’s flow regime is controlled by variations in meltwater regime, the intensity of the SE Asian monsoon, and the frequency and intensity of tropical cyclones that penetrate inland from the South China Sea. Since melt contributes only to (non-erosive) base flows, variability in glacier and snow melt contributions from Himalayan source areas are unlikely to significantly impact inter-annual variations in river bank erosion. However, inter-annual variability in the SE Asian monsoon and tropical cyclone dynamics may both be related, albeit in a complex manner, to the El Niño Southern Oscillation (ENSO). Cross-wavelet transform and wavelet coherence analyses indicate statistically significant (at 95% limit) coherence between ENSO and simulated river bank erosion, but at both sites only since about 1980. During this recent period the ENSO and fluvial erosion time series are in anti-phase; that is ENSO cold phases are associated with enhanced rates of river bank erosion. This is as expected, since ENSO cold phases are associated with earlier onset and enhanced intensity of the monsoon, while the number of intense tropical storm systems making landfall over Vietnam and moving across the Lower Mekong Basin is also higher.