A comparison of two global datasets of extreme sea levels and resulting flood exposure

Abstract Estimating the current risk of coastal flooding requires adequate information on extreme sea levels. For over a decade, the only global data available was the DINAS‐COAST Extreme Sea Levels ( DCESL ) dataset, which applies a static approximation to estimate extreme sea levels. Recently, a dynamically derived dataset was developed: the Global Tide and Surge Reanalysis ( GTSR ) dataset. Here, we compare the two datasets. The differences between DCESL and GTSR are generally larger than the confidence intervals of GTSR . Compared to observed extremes, DCESL generally overestimates extremes with a mean bias of 0.6 m. With a mean bias of −0.2 m GTSR generally underestimates extremes, particularly in the tropics. The Dynamic Interactive Vulnerability Assessment model is applied to calculate the present‐day flood exposure in terms of the land area and the population below the 1 in 100‐year sea levels. Global exposed population is 28% lower when based on GTSR instead of DCESL . Considering the limited data available at the time, DCESL provides a good estimate of the spatial variation in extremes around the world. However, GTSR allows for an improved assessment of the impacts of coastal floods, including confidence bounds. We further improve the assessment of coastal impacts by correcting for the conflicting vertical datum of sea‐level extremes and land elevation, which has not been accounted for in previous global assessments. Converting the extreme sea levels to the same vertical reference used for the elevation data is shown to be a critical step resulting in 39–59% higher estimate of population exposure.