Evaluation of vertical accuracy of five open-source DEM datasets in a mountainous city: A case study of Chongqing central urban area, China

Accurate digital elevation models (DEMs) are essential for terrain analysis, urban planning, hydrological modeling, and disaster risk assessment in mountainous cities. However, complex topography, heterogeneous land cover, and dense high-rise buildings can substantially affect DEM vertical accuracy. Taking the central urban area of Chongqing, China, as a representative mountainous city, this study evaluates the vertical accuracy of five freely available global DEMs, namely ALOS PALSAR, Copernicus DEM, NASADEM, AW3D30, and ASTER GDEM V3. High-quality ICESat-2 ATL08 laser altimetry points were used as reference elevations after data filtering and vertical datum harmonization. Accuracy was assessed using mean error (ME), mean absolute error (MAE), standard deviation (STD), and root mean square error (RMSE), and the effects of land cover, slope, aspect, and building height were further analyzed. The results show that ALOS PALSAR achieves the highest overall accuracy, with an RMSE of 2.97 m, followed by NASADEM, AW3D30, Copernicus DEM, and ASTER GDEM V3, with RMSE values of 3.76 m, 5.67 m, 6.33 m, and 9.85 m, respectively. Land cover significantly affects DEM accuracy, with larger errors generally observed in forested areas, whereas ALOS PALSAR maintains relatively stable performance across different land cover types. DEM errors tend to increase on steeper slopes, while aspects have a comparatively limited influence. Building height is also an important factor, with errors increasing in high-rise built-up areas and showing a tendency toward elevation underestimation. These findings provide practical guidance for selecting suitable open-source DEMs in mountainous urban environments and highlight the need to consider terrain, land cover, and urban morphology when applying DEMs in complex cities.