Vertical accuracy assessment of fusion Global Digital Elevation Model (NASADEM and MERIT) over Peninsular Malaysia / Nur fatihah Mohd Rosli

This study focuses on to assess the vertical accuracy of the fusion Global Digital Elevation Model (GDEM) comprising NASADEM and MERIT datasets over Peninsular Malaysia. The primary objectives were to derive the vertical accuracy of fusion GDEM based on landuse classification and to improve the accuracy of fusion GDEM by fit to the local vertical datum. To achieve these objectives, the study utilized LIDAR datasets to assess the vertical accuracy by comparing the elevation surfaces between LIDAR datasets and the fusion GDEM datasets. This analysis involved subtracting the elevation values to determine the discrepancies. In addition, the MERIT dataset, which exhibited the best Root Mean Square Error (RMSE) for landuse classification, was identified as the optimum fusion GDEM. To align the fusion GDEM with the local vertical datum, a Corrective Vertical Datum (CVD) approach was employed. The CVD method involved calculating the difference between the EGM96 geoid surface and the MyGEOID, which was used to fit the MERIT dataset to the local vertical datum. The results demonstrated that the MERIT dataset performed well in urban, forest and non-forest areas, with vertical accuracy RMSE values of 4.334 m, 6.744 m, and 6.081 m respectively meanwhile NASADEM performed worst for each area which urban with 9.394 m, forest 9.304 m and lastly 8.888 m for non-forest area. The new MERIT dataset, which was fitted to the MyGEOID, exhibited improved accuracy, and successfully aligned with the local datum in Peninsular Malaysia, as indicated by an RMSE value of 2.627 m. In comparison, the MERIT dataset fitted to EGM96 provided an RMSE value of 3.012 m. Although the difference was slight, the new MERIT dataset aligned with the local datum, demonstrating its suitability for accurate geospatial applications in Peninsular Malaysia. In conclusion, this study assessed the vertical accuracy of the fusion GDEM over Peninsular Malaysia, focusing on landuse classification. By incorporating lidar datasets and employing the CVD method, the fusion GDEM was aligned with the local vertical datum, resulting in improved accuracy. The findings highlight the effectiveness of the fusion GDEM, particularly the new MERIT dataset, for geospatial analyses and applications in the region.