CoastalDEM v2.1 A New, High-Accuracy and High-Resolution Global Coastal Elevation Model Trained on ICESat-2 Satellite Lidar

CoastalDEM v2.1 is a near-global digital elevation model (DEM) developed by Climate Central to provide more accurate land elevation data for coastal flood risk assessments. By utilizing a sophisticated convolutional neural network trained on NASA's ICESat-2 satellite lidar data, the model corrects the positive vertical bias common in satellite radar-derived DEMs, which often mistake building rooftops for ground level. Testing shows that CoastalDEM v2.1 virtually eliminates global median bias in the critical 0 to 5 m elevation band, significantly outperforming other publicly available global DEMs such as NASADEM, TanDEM-X, MERIT, and AW3D30, particularly in densely populated urban areas.

• CoastalDEM v2.1 significantly reduces vertical bias compared to other global digital elevation models (DEMs), virtually eliminating global median bias to less than 0.01 m in the 0 to 5 m elevation band.
• In areas with moderate urban development (over 1,000 people per square kilometer) and elevations below 5 m, CoastalDEM v2.1's median vertical bias is 0.09 m, which is three times lower than CoastalDEM v1.1 and ten times lower than other comparable DEMs.
• In very high population density areas (over 10,000 people per square kilometer), CoastalDEM v2.1 shows a median bias of 0.42 m, whereas other DEMs range from 2.04 to 2.84 m; this difference is approximately three times the range of sea level rise projected by the IPCC AR6 for 2100 (0.5 to 1.0 m).
• The model's accuracy was validated against airborne lidar data in the U.S. and Australia, where it outperformed other global DEMs and even surpassed the performance of CoastalDEM v1.1 in the U.S., despite v1.1 being specifically trained on U.S. data.
• CoastalDEM v2.1 utilizes a more advanced convolutional neural network architecture with thousands of hidden units and over a thousand input variables per pixel, trained on global ICESat-2 satellite lidar data to better represent the built environment globally.
• The model expands the elevation range for predicted corrections to between -10 m and 120 m, compared to the 1 to 20 m range used in CoastalDEM v1.1, to better handle low flat regions and tall skyscrapers.

APA

Climate Central (2021). CoastalDEM v2.1 A New, High-Accuracy and High-Resolution Global Coastal Elevation Model Trained on ICESat-2 Satellite Lidar. https://assets.ctfassets.net/cxgxgstp8r5d/Ygz6c7lz0MaEneXJfE1xT/c9e07d21aa917dc2d7b77d4ba13e7395/CoastalDEM2.1-White-Paper_.pdf

Chicago

Climate Central. CoastalDEM v2.1 A New, High-Accuracy and High-Resolution Global Coastal Elevation Model Trained on ICESat-2 Satellite Lidar. 2021. https://assets.ctfassets.net/cxgxgstp8r5d/Ygz6c7lz0MaEneXJfE1xT/c9e07d21aa917dc2d7b77d4ba13e7395/CoastalDEM2.1-White-Paper_.pdf.

Wikipedia

{{cite report |author=Climate Central |title=CoastalDEM v2.1 A New, High-Accuracy and High-Resolution Global Coastal Elevation Model Trained on ICESat-2 Satellite Lidar |date=21 September 2021 |url=https://assets.ctfassets.net/cxgxgstp8r5d/Ygz6c7lz0MaEneXJfE1xT/c9e07d21aa917dc2d7b77d4ba13e7395/CoastalDEM2.1-White-Paper_.pdf |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{climatecentral2021coastaldem, author = {{Climate Central}}, title = {{CoastalDEM v2.1 A New, High-Accuracy and High-Resolution Global Coastal Elevation Model Trained on ICESat-2 Satellite Lidar}}, institution = {Climate Central}, year = {2021}, month = sep, url = {https://assets.ctfassets.net/cxgxgstp8r5d/Ygz6c7lz0MaEneXJfE1xT/c9e07d21aa917dc2d7b77d4ba13e7395/CoastalDEM2.1-White-Paper_.pdf}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }