Comparison of different resolutions of six free online DEMs with GPS elevation data on a new 6th of October City, Egypt

The Digital Elevation Models (DEMs) are used to represent the topographic surface of the earth as it is used in many applications. Nowadays, many free DEMs are available online. This study aims to assess the accuracy of six free DEMs. This research uses the GPS observations to determine the elevation stations in the study area by using the Post-Processing Kinematic (PPK) technique and compares these elevations against the elevations computed at the same stations by using six free online Digital Elevation Models (DEMs): SRTM 1 Arc-Second, ASTER 1 Arc-Second, NASADEM 1 Arc-Second, SRTM 3 Arc-Second, GTOPO30, and ALOS PALSAR 12.5 m to find the best results from the free DEMs. The elevations of all stations of different data sources were calculated using the geoid model EGM96. The results indicate that the DEMs that give the best results in this study are obtained from NASADEM 1 Arc-Second, where (RMSE is 2.99 m, and the SD is 2.97 m), and SRTM 3 Arc-Second, where the (RMSE is 3.70 m, and the SD is 3.53 m).


Introduction
A Digital Elevation Model (DEM) is a valuable data source for numerous studies and viewing the earth's 3D surface topography (Ibrahim 2018). The era of satellite remote sensing has made the Digital Elevation Model (DEM) available through its onboard active and passive sensors. Accuracy and extent of use based on spatial resolution and temporal resolution of topographical information quality. Detailed spatial resolution varied from 10 cm, 5-15 m, 30-90 m, and 1 km (Usman et al. 2021). DEMs are a standard way for making topographic maps, orthophotos, civilian engineering projects, and other engineering applications. Several software applications have recently been developed and used to construct DEMs from a variety of sources, including field surveying, scanned topographic maps, 3D laser scanning, and stereo images viewed from the air or space. The primary goal of creating a DEM is to depict surface elevation (Hapep and Al-Bakri 2020). Three names, DSM, DTM, and DEM, are used interchangeably in scientific literature to depict elevation models. The elevations of the earth's surface, including natural and man-made things like plants and buildings, are referred to as a Digital Surface Model (DSM). The elevation values for bare earth are referred to as a Digital Terrain Model (DTM), which excludes features such as vegetation and buildings. The term Digital Elevation Model (DEM) is widely used to refer to either a DSM or a DTM (Delgado et al. 2018). DEMs come in a variety of resolutions and are created using a variety of methods. DEMs that provide good terrain representation are extremely expensive. The data used in this study was acquired through two main sources: firstly, ground elevation data obtained by using the GPS Post-Processing Kinematic (PPK) technique and secondly the six free online Digital Elevation Model (DEMs) data obtained from SRTM 1 Arc-Second, ASTER 1 Arc-Second, SRTM 3 Arc-Second, NASADEM 1 Arc-Second, GTOPO30, and ALOS PALSAR DEM 12.5 m. This research aims to examine six free online DEMs on the new 6th of October City, Egypt, using the elevations of 572,200 stations obtained from the GPS by Post-Processing Kinematic (PPK) technique to determine the best DEM in the study area, where the vertical accuracy of the stations obtained from using GPS by Post-Processing Kinematic (PPK) technique was 5 cm.

Study area
The study area is a new 6th of October City which is located to the West of Cairo governorate. The study area is located in the Egypt Red Belt coordinate system on the Egyptian Transverse Mercator (ETM). The study area is approximately 326 km 2 (see Fig. 1). The elevations of the study area vary between 105.50 m and 270.30 m.

Data sources
This research used six free online DEMs with different resolutions and compared them with the GPS elevation data. The following procedure illustrates the acquisition of data used in this study:

ALOS-PALSAR DEM
The Japan Aerospace and Exploration Agency launched ALOS-PALSAR in 2006. (JAXA). Until May 12, 2011, the ALOS-PALSAR was operational. (Khal et al. 2020). The Advanced Land Observing Satellite Phased Array type L-band Synthetic Aperture Radar (ALOS PALSAR) DEM data collection has a spatial resolution of 12.5 m. The ALOS PALSAR is an active microwave sensor that operates at 28 MHz and has a fine range resolution of 7-44 m. The L-band frequency is used by the sensor (Khatriker et al. 2019). The ellipsoid WGS 1984 is used to obtain DEM data (Ferreira and Cabral 2021).

NASADEM
The NASADEM information products were derived from original telemetry information from the Shuttle Radar Topography Mission (SRTM), a collaboration between the National Aeronautics and Space Administration (NASA) and the National Geospatial-Intelligence Agency (NGA), in addition to the involvement of German and Italian space agencies. In addition to Terra Advanced Spaceborne Thermal and Reflection Radiometer (ASTER) Global Digital

GPS elevations
This research uses GPS observations to determine the elevations of the study area by the Post-Processing Kinematic (PPK) technique. In December 2018 the GPS observations were taken from the Leica Viva instrument and processing the observations by LEICA Geo Office LGO 7 software, where the GPS observations were taken using a four-wheel-drive vehicle in which the device was installed and movement in the study area was carried out with distances ranging from 10 to 100 m. The output data resulting from processing the GPS observations by using the LGO 7 software was obtained on the Egyptian Transverse Mercator and EGM96.

Elevation data validation
In this study, statistical indices were employed to analyze, validate, and compare the DEM to GPS data. The vertical accuracy of the DEM may be assessed using this statistical analysis. An elevation error was calculated for each station as the difference in elevation between model and ground reference data (see Eq. (1)).
In the above equation, Z diff (i) is the elevation error at the station, Z model (i) is the elevation of the station on DEM, and Z ref (i) is the elevation of the GPS station. The value of Mean Error (ME), Root Mean Square Error (RMSE), and Standard Deviation (SD) were computed for each DEM (see Eqs. (2)-(4)). (1) RMSEs are measures of surface quality and provide the understanding of differences between two types of data (predicted by the model and observed data) (Khalid et al. 2016).   Fig. 3, where the Office of Geomatics website (NGA -Office of Geomatics 2022) was used to convert the ellipsoidal heights to orthometric heights on EGM96 by calculating the geoid heights. • Calculating the difference in heights between the GPS station heights and the station heights obtained by different DEMs. The difference in heights is used for calculating the values of Mean Error (ME), Root Mean Square Error (RMSE), and Standard Deviation (SD).

Results
The comparison between the different DEMs with the GPS data in the study area is illustrated hereafter.
• Fig. 4 shows the difference in elevations between 313,149 GPS stations and the same stations at the ALOS PALSAR DEM. The statistical analysis revealed that the maximum difference, minimum difference, ME, RMSE, and SD are 12.40 m, − 42.15 m, − 2.91 m, 3.85 m, and 5.13 m, respectively (see Table 1). • Fig. 5 depicts the difference in elevations between 572,200 GPS stations and the same stations at the ASTER 1 Arc-Second DEM. The statistical analysis revealed that the maximum difference, minimum difference, ME, RMSE, and SD are 51.52 m, − 70.24 m, − 2.66 m, 6.74 m, and 6.65 m, respectively (see Table 1). • The maximum difference, minimum difference, ME, RMSE, and SD are 39.52 m, − 40.99 m, − 2.68 m, 4.07 m, and 3.92 m, respectively; this result is the statistical analysis between the GPS data and SRTM 1 Arc-Second DEM (see Table 1). The difference in elevations of 572,200 stations between GPS data and SRTM 1 Arc-Second DEM is shown in Fig. 6 Fig. 8. • Fig. 9 shows the difference in elevations between 572,200 GPS stations and the same stations at the GTOPO30 (1000 m) DEM. The statistical analysis revealed that the maximum difference, minimum difference, ME, RMSE,and SD are 53.19 m,13.17 m,and 13.14 m, respectively (see Table 1).
Also in this research, we excluded the outlier values in the four DEMs that gave the best results in order to note whether removing these stations affected the accuracy of the results of these models or did not affect, as the number of stations that were excluded in each DEM was as follows:  Fig. 11).F ig. 10 Difference in elevations between the GPS data and the NASADEM 1 Arc-Second after excluding the outlier values

Conclusions
In this research, we evaluate the elevation data extracted using six free different DEMs (SRTM 1 Arc-Second, ASTER 1 Arc-Second, NASADEM 1 Arc-Second, SRTM 3 Arc-Second, GTOPO30, and ALOS PALSAR 12.5 m) by comparing these DEM elevation data on EGM96 with the 572,200 ground stations obtained by using the DGPS on EGM96 on the new 6th of October City. From the results, we can conclude that.
• Through the results of the evaluations of DEMs (NASA-DEM, SRTM, and ASTER) with a spatial resolution of 30 m, the best results we got were from NASADEM, where the results of the statistical analysis are the maximum difference, minimum difference, ME, RMSE, and SD that are 40.52 m, − 37.99 m, − 0.76 m, 2.99 m, and 2.97 m, respectively. Then came second in order of accuracy SRTM DEM. Finally, the ASTER DEM gave the worst results; this is evident by giving the highest values, whether in the difference in elevations, ME, RMSE, and SD. • The second-best results obtained in this study were those obtained from SRTM 3 Arc-Second DEM whose spatial resolution is 90 m. The results were as follows: the maximum and minimum differences in elevations are 39.44 m and − 24.99 m, respectively; the mean error is − 2.70 m; the root mean square error is 3.70 m; and the standard deviation error is 3.53 m. • Although the spatial resolution of ALOS PALSAR DEM is 12.5 m, the results we obtained from this model (using 313,149 ground stations) are less accurate than the SRTM DEM 30 m that appears from the results. • The statistical analysis showed that the ASTER DEM whose spatial resolution was 30 m and the GTOPO30 DEM whose spatial resolution was 1000 m have given the worst results in the study area.
In this research, the results of the statistical analysis of the four DEMs: SRTM 1 Arc-Second, NASADEM 1 Arc-Second, SRTM 3 Arc-Second, and ALOS PALSAR 12.5 m, were the best, so we excluded the outlier stations in these models in the hope of improving the results of these models. The results showed that the improvement was made at the maximum and the minimum differences, but the improvement was on ME, RMSE, and SD no more than 4 cm. This means that the exclusion of outlier stations did not affect the accuracy of the DEMs.