Spaceborne digital elevation models (DEMs) are a fundamental input for many geoscience studies, but they still include nonnegligible height errors. Here we introduce a high‐accuracy global DEM at 3″ resolution (~90 m at the equator) by eliminating major error components from existing DEMs. We separated absolute bias, stripe noise, speckle noise, and tree height bias using multiple satellite data sets and filtering techniques. After the error removal, land areas mapped with ±2 m or better vertical accuracy were increased from 39% to 58%. Significant improvements were found in flat regions where height errors larger than topography variability, and landscapes such as river networks and hill‐valley structures, became clearly represented. We found the topography slope of previous DEMs was largely distorted in most of world major floodplains (e.g., Ganges, Nile, Niger, and Mekong) and swamp forests (e.g., Amazon, Congo, and Vasyugan). The newly developed DEM will enhance many geoscience applications which are terrain dependent.

Floods are a natural hazard that affect communities worldwide, but to date the vast majority of flood hazard research and mapping has been undertaken by wealthy developed nations. As populations and economies have grown across the developing world, so too has demand from governments, businesses, and NGOs for modeled flood hazard data in these data‐scarce regions. We identify six key challenges faced when developing a flood hazard model that can be applied globally and present a framework methodology that leverages recent cross‐disciplinary advances to tackle each challenge. The model produces return period flood hazard maps at ∼90 m resolution for the whole terrestrial land surface between 56°S and 60°N, and results are validated against high‐resolution government flood hazard data sets from the UK and Canada. The global model is shown to capture between two thirds and three quarters of the area determined to be at risk in the benchmark data without generating excessive false positive predictions. When aggregated to ∼1 km, mean absolute error in flooded fraction falls to ∼5%. The full complexity global model contains an automatically parameterized subgrid channel network, and comparison to both a simplified 2‐D only variant and an independently developed pan‐European model shows the explicit inclusion of channels to be a critical contributor to improved model performance. While careful processing of existing global terrain data sets enables reasonable model performance in urban areas, adoption of forthcoming next‐generation global terrain data sets will offer the best prospect for a step‐change improvement in model performance.

Current flood risk mapping, relying on historical observations, fails to account for increasing threat under climate change. Incorporating recent developments in inundation modelling, here we show a 26.4% (24.1–29.1%) increase in US flood risk by 2050 due to climate change alone under RCP4.5. Our national depiction of comprehensive and high-resolution flood risk estimates in the United States indicates current average annual losses of US$32.1 billion (US$30.5–33.8 billion) in 2020’s climate, which are borne disproportionately by poorer communities with a proportionally larger White population. The future increase in risk will disproportionately impact Black communities, while remaining concentrated on the Atlantic and Gulf coasts. Furthermore, projected population change (SSP2) could cause flood risk increases that outweigh the impact of climate change fourfold. These results make clear the need for adaptation to flood and emergent climate risks in the United States, with mitigation required to prevent the acceleration of these risks.

Past attempts to estimate rainfall-driven flood risk across the US either have incomplete coverage, coarse resolution or use overly simplified models of the flooding process. In this paper, we use a new 30 m resolution model of the entire conterminous US with a 2D representation of flood physics to produce estimates of flood hazard, which match to within 90% accuracy the skill of local models built with detailed data. These flood depths are combined with exposure datasets of commensurate resolution to calculate current and future flood risk. Our data show that the total US population exposed to serious flooding is 2.6–3.1 times higher than previous estimates, and that nearly 41 million Americans live within the 1% annual exceedance probability floodplain (compared to only 13 million when calculated using FEMA flood maps). We find that population and GDP growth alone are expected to lead to significant future increases in exposure, and this change may be exacerbated in the future by climate change.

Human exposure to floods continues to increase, driven by changes in hydrology and land use. Adverse impacts amplify for socially vulnerable populations, who disproportionately inhabit flood-prone areas. This study explores the geography of flood exposure and social vulnerability in the conterminous United States based on spatial analysis of fluvial and pluvial flood extent, land cover, and social vulnerability. Using bivariate Local Indicators of Spatial Association, we map hotspots where high flood exposure and high social vulnerability converge and identify dominant indicators of social vulnerability within these places. The hotspots, home to approximately 19 million people, occur predominantly in rural areas and across the US South. Mobile homes and racial minorities are most overrepresented in hotspots compared to elsewhere. The results identify priority locations where interventions can mitigate both physical and social aspects of flood vulnerability. The variables that most distinguish the clusters are used to develop an indicator set of social vulnerability to flood exposure. Understanding who is most exposed to floods and where, can be used to tailor mitigation strategies to target those most in need.