Premium
A Physics‐Informed Bayesian Storyline Approach to Assess Sediment Transport in the Mekong
Author(s) -
Xu Bo,
He Xiaogang
Publication year - 2022
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/2022wr032681
Subject(s) - bayesian inference , climate change , mekong river , environmental science , range (aeronautics) , precipitation , sediment , structural basin , bayesian probability , sediment transport , hydrology (agriculture) , physical geography , geography , statistics , geology , meteorology , mathematics , geomorphology , oceanography , materials science , geotechnical engineering , composite material
The Mekong Delta, home to 20 million people, is experiencing significant land loss due to rising sea levels, accelerating land subsidence, and declining sediment supply. Robust estimates of the sediment flux delivered to Mekong Delta (SF MD ) and the relative contribution of sediment load (RCSL) from individual subbasins are key to designing future adaptation strategies, such as strategic dam planning, sand mining, and delta groundwater management. However, existing estimates of SF MD and RCSL are largely deterministic without uncertainty quantification or using a uniform sampling to represent uncertainty. They also remain questionable due to data inconsistency and methodological biases caused by overlooked physical processes. Here, we develop a hybrid physics‐based data‐driven modeling framework to constrain the probability distribution of SF MD and RCSL and explore how they change under plausible climate change and land‐use change scenarios, leveraging recent advances in Bayesian inference (i.e., reasoning by refutation). We find that pure yield‐based approaches, which typically ignore sediment retention, can lead to higher estimates of RCSL from upstream regions compared with the physics‐based approach. Our best estimate of historical (1962–2005) SF MD combining multiple lines of evidence shows a median of 106 Mt/yr with a 5%–95% range of 66–160 Mt/yr. Over the analyzed range of land‐use and climate change scenarios, future changes in SF MD seem to be more sensitive to the latter, especially changes in wet‐season precipitation. Our estimate of RCSL from the Upper Mekong River Basin is likely (>66% probability) in the range of 0.25–0.39 with a mean of 0.34 in all plausible scenarios.