Global review reveals higher coastal sea levels than maps show

A new Nature article shows that most global and regional studies of sea level rise underestimate how high the sea actually is along the coast. The authors reviewed 385 scientific papers published between 2009 and 2025 and found that more than 99% did not correctly align sea level data with coastal elevation data. This means many assessments misjudge how much land is already below, or will fall below, sea level.

How were existing studies checked?

The researchers carried out a systematic review of peer-reviewed studies on sea level rise and coastal flooding. They focused on how each study used digital elevation models (DEMs), which provide land height, and how they linked these to sea level data. In 73% of the publications, key information on vertical reference levels was incomplete or missing. Only one study fully documented and correctly handled the vertical reference, including a proper link to observed sea level.

Most of the problems come from how authors treat vertical datums: the reference surfaces used to express elevation. Many studies simply took the geoid value of zero metres as if it were the same as mean sea level at the coast. A geoid is a gravity-based model surface, not a direct measurement of sea level. In reality, currents, winds, temperature and salinity all change the height of the sea surface compared with the geoid.

Sea level often sits well above the geoid

To quantify the effect of this simplification, the authors compared widely used global geoid models to measured mean sea level derived from satellite altimetry, known as mean dynamic topography. On average, actual coastal sea level stands 0.24–0.27 metres above the geoid-based assumption. In some regions, especially in the Indo-Pacific and parts of Southeast Asia, the difference can exceed 1 metre and locally reach several metres.

Maps in the article show that these large positive offsets occur mainly in data-sparse regions of the Global South, such as Southeast Asia, small Pacific islands, parts of Latin America and East Africa. In contrast, in North America and Western and Northern Europe, where gravity data are more complete, the geoid more closely matches measured sea level and the offsets are much smaller.

Underestimated land and population below future sea level

The team then used four modern global elevation datasets to run global calculations of exposure to relative sea level rise, both with and without a proper sea level reference. They applied a hypothetical 1 metre rise in relative sea level and compared the land and population that would fall below sea level in each case. When sea level is assumed to follow the geoid, the exposed area comes out at around 295,000–431,000 square kilometres. After correctly aligning elevation data to measured sea level, the estimated area below sea level increases to about 460,000–670,000 square kilometres, an increase of 31–37%.

The effect on population estimates is even larger. Using geoid-based methods, earlier studies suggested that 34–49 million people would live below sea level after 1 metre of relative sea level rise. With proper sea level referencing, this rises to 77–132 million people, an increase of 48–68%. The biggest changes appear in Southeast Asia, where the estimated area and population below sea level nearly double once the higher true sea level is taken into account.

The authors also looked at the so-called low-elevation coastal zone, the land within 10 metres of sea level. When using mean sea level instead of a geoid reference, the global area and population in this band increase by up to 4% and 8%, respectively. This suggests that many previous estimates of how many people live in low-lying coastal zones are too low.

Implications for global assessments and future practice

The study shows that vertical reference problems have not only affected individual papers, but have also passed into major international assessments. The authors identify 46 of the evaluated studies that are cited in the latest IPCC Sixth Assessment Report cycle. Many of these studies either skipped a sea level reference altogether or applied it incorrectly. When the authors recalculate exposure using corrected methods, they find that global estimates of population in the low-elevation coastal zone become higher than the numbers currently quoted in the IPCC reports.

To support better practice, the authors provide versions of several global elevation datasets that are already converted to mean sea level using the latest satellite-based sea level information. They also propose clearer documentation standards, review checklists and, where possible, data products that already combine elevation and sea level references.