Elevation uncertainties in the Mekong Delta quantified using a transferable approach

Abstract

The elevation of coastal lowlands relative to local sea level is a crucial determinant for their exposure and is key input for coastal hazard and relative sea-level rise impact assessments. For many data-sparse coastal lowlands worldwide, global digital elevation models are often the only source of information. While these provide an adequate spatial (i.e. horizontal) resolution for regional, delta-wide coastal assessments, their vertical errors in the range of several metres impede investigations of (relative) sea-level rise impact where changes occur on millimetre- to centimetre-scale. Assessing the quality of available elevation datasets is required to identify the best performing model(s) to use for generating reliable coastal impact and exposure assessments. While data-intrinsic inaccuracy has been extensively addressed both in dataset documentation and literature, the relevance and proper vertical datum conversion from global geoid and ellipsoid to local sea level is often still omitted in many applied studies from coastal research. Similarly, the impact of the actuality of elevation data (i.e. time since data acquisition) on assessments in coastal lowlands is so far understudied although elevation models may become quickly outdated, especially where coastal lowlands are facing high rates of elevation change resulting from the interplay of vertical land motion, (vertical) sediment accretion and sea-level change. Particularly for flat, low-lying subsiding coastal landscapes like the Mekong Delta, being in parts only a few decimetres elevated above sea level and experiencing land subsidence of up to several centimetres per year, the reliability of elevation data and adequate representation of elevation relative to local sea level as well as the consideration of factors impacting elevation over time is of utmost importance. We present a transferable and ultimately globally applicable approach to quantify and attribute uncertainties in elevation assessment for data-sparse coastal lowlands using global elevation models to sources such as inaccuracy, vertical datum offset and actuality. The approach combines openly available land elevation and sea-level datasets, integrated with non-linear time series and projections of sea-level change and vertical land motion while also pointing out the need of information on sediment accretion/erosion. We showcase this approach by revisiting land elevation in the Vietnamese Mekong Delta (i) by vertically referencing 11 commonly used global elevation models and an updated local elevation model to a common actual, local sea-level datum, and (ii) by conducting a thorough assessment of elevation model performance that not only allows for the quantification of errors and elevation assessment uncertainties but also their attribution to data-intrinsic inaccuracy, vertical datum offset and, tentatively, non-linear impact of elevation change due to vertical land motion (e.g. extraction-induced land subsidence) and sea-level change affecting the actuality of the elevation model. Our approach not only allows to improve the understanding of coastal elevation to further improve relative sea-level rise and flood impact assessments and to substantiate projections of future elevation in the Mekong Delta, but in its design, applying solely open data and commonly used GIS software, facilitates similar assessments of elevation model performance and elevation assessment uncertainties in other (data-sparse) coastal regions in the world.