ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume X-4/W1-2022

GeoSpatial Conference 2022 – Joint 6th SMPR and 4th GIResearch Conferences, 19–22 February 2023, Tehran, Iran (virtual)

VERTICAL ACCURACY ASSESSMENT OF COPERNICUS DEM

(CASE STUDY: TEHRAN AND JAM CITIES)

M. A. Ghannadi 1*, S. Alebooye 2, M. Izadi 3, A. Ghanadi4

1 Arak University of Technology, Department of Geoscience Engineering, Arak, Iran - m.ghannadi@arakut.ac.ir
2 University of Tehran, School of Surveying and Geospatial Engineering, College of Engineering, Tehran, Iran – s.alebooye@ut.ac.ir
3 Western Michigan University, Department of Geological and Environmental Sciences, Kalamazoo, USA - moein.izadi@wmich.edu

4 Department of Civil Engineering, Faculty of Technology and Engineering, Central Tehran Branch, Islamic Azad University, Tehran,

Iran- alighanadii@gmail.com

Commission IV, WG IV/3

KEY WORDS: Copernicus, DEM, DSM, Accuracy Assessment, SRTM, WorldDEM.

ABSTRACT:

Digital Elevation Models (DEMs) are widely utilized in a variety of fields, including, hydrology, geomorphology, and geoscience.
The dataset is now openly and freely available to the public, with access permissions for data at 30- and 90-meter resolution having
been expanded. The Copernicus DEM is a Digital Surface Model (DSM) that depicts the Earth's surface, which includes structures,
infrastructure, and vegetation. This DEM is based on the WorldDEM DSM, which has been altered. The Copernicus DEM's vertical
accuracy is assessed in this paper. The study area is in Iran, north of Tehran and west of Jam. The Copernicus DEM is compared to
the Shuttle Radar Topography Mission (SRTM) in terms of vertical accuracy. The results of the experiments reveal that the
Copernicus DEM is more accurate than the SRTM. In the Tehran and Jam examples, the root mean square error (RMSE) for
Copernicus DEM (30 m) is 5.72 and 2.19 meters, respectively. Also, this value for SRTM is 6.10 and 3.95 respectively.

1. INTRODUCTION latitudes and across Antarctica, such as GTOPO, GLOBE,

Greenland Ice Sheet Mapping Project (RAMP), and
In numerous scientific and commercial uses, digital elevation GLAS/ICESat (Rizzoli et al., 2017).
models (DEMs) are essential. Many geoscience fields need The WorldDEM (Bayburt et al., 2017) is another global DEM
precise and up-to-date information about the Earth's surface that is now available to numerous users for a variety of
such as geology (Elmahdy et al., 2021, Aljammaz et al., 2021), applications. The new TanDEM-X DEM has been regarded as
forestry (Gdulová et al., 2020, Schlund et al., 2016), glaciology one of the most accurate and consistent global DEM data sets of
(Błaszczyk et al., 2019), oceanography (Abrams et al., 2020), the Earth's surface since September 2016 (Wessel et al., 2018).
and hydrology (Moretti and Orlandini, 2018, Mohammadi et al., This ground-breaking product will be used in a variety of
2020) and risk management of natural disasters (Hawker et al., regional and worldwide applications for evaluating biological
2019, Florinsky and Bliakharskii, 2019). DEMs can be obtained evaluating biological and physical processes on the planet's
from contour lines, topographic maps, land surveys, surface (Zink et al., 2014). Single-pass SAR interferometry was
photogrammetry techniques, Synthetic Aperture Radar (SAR) used to obtain the height data. The satellites TerraSAR-X and
interferometry, and light detection and ranging (LiDAR) TanDEM-X, which fly in a near helix configuration with
instruments (Arun, 2013). distances between 300 and 500 meters, captured the
Until now, our society has created three DEMs that encompass corresponding sets of photos (Zink et al., 2014). The measured
virtually all the world's land masses. Even though globes and height corresponds to the reflective surface of the X-Band
maps of the world depicting land topography existed well signal, even though SAR interferometry is well adapted to
before, none of these topographic products were available at a globally map the Earth's surface in a short period due to its "day
spatial resolution and precision efficient to meet the multiple and night" and "all-weather" observation capacity (Rossi and
needs of civilization's long-term development. (Becek et al., Gernhardt, 2013). The SAR data utilized for the global DEM
2016). The Shuttle Radar Topography Mission (SRTM), which production were gathered in StripMap mode with horizontal
has a spatial resolution of 30 m between 56° South and 60° transmit and receive polarization between December 2010 and
North latitude, has been the principal source of elevation data January 2015.
on an almost global scale-up until now (Rizzoli, Martone et al. The Copernicus DEM (Cop-DEM) is the most recent DEM
2017). Furthermore, DEMs with a 30-meter resolution are from around the world to become publicly available. The goal
available from the ALOS global digital surface model of this paper is to assess Cop-vertical DEM's accuracy and to
(AW3D30) (Caglar et al., 2018) and the Advanced Spaceborne achieve this, more precise DEMs (derived from aerial images
Thermal Emission and Reflection Radiometer (ASTER) and GeoEye satellite images) were employed. The paper is
(Sujatha et al., 2015). The AW3D30 one covers latitudes structured as follows: the Cop-DEM characteristics are
between 83° South and 83° North, but the ASTER has introduced in Section 2; Section 3 introduces the research topic
significant holes in both Antarctica and the arctic areas (Rizzoli and the data employed; the experimental results are discussed in
et al., 2017). Only lower resolution DEMs (in the range of Section 4 and the study will be concluded in the summary
hundreds of meters to kilometers) are available for higher section.

* Corresponding author

This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-X-4-W1-2022-209-2023 | © Author(s) 2023. CC BY 4.0 License. 209
 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume X-4/W1-2022
GeoSpatial Conference 2022 – Joint 6th SMPR and 4th GIResearch Conferences, 19–22 February 2023, Tehran, Iran (virtual)

2. COPERNICUS DEM 3. STUDY AREA

The Cop-DEM is a digital elevation model (DEM) which The northern section of Tehran city and the neighboring areas of
depicts the Earth's surface, which includes buildings, Jam city in Iran are the research areas for Cop-DEM (30m and
infrastructure, and vegetation. Flattening of water bodies and 90m) validation. The UTM coordinates for the research region
continuous river flow have been added in this DEM, which is in Tehran City (zone 39N) extend from the northeast (X:
produced from an edited DSM named WorldDEM. Shorelines 542450, Y: 3964590) to the southwest (X: 542450, Y:
and coasts have been edited, as well as specific features such as 3964590). Most of this area is urban. (X: 539010, Y: 3962350).
airports and improbable terrain formations. The WorldDEM Jam City (zone 40N) coordinates are (X: 630100, Y: 3082200)
product is based on radar satellite data collected during the and (X: 625500, Y: 3079000), respectively. This area is
TanDEM-X Mission, that is supported by a public-private completely mountainous. In Tehran, the study area elevation
partnership between the German government, represented by extends from 1555 to 2080 m, while in Jam, the range is 579 to
the German Aerospace Center (DLR), and Airbus Defence and 830 m. Figure 1 shows a Cop-DEM (30 m) of the study area.
Space. The Cop-DEM is available in three different versions:
EEA-10 (for EEA39 nations and regions), GLO-30, and GLO- For comparison with Cop-DEM from the research areas,
90. The Cop-DEM GLO-30 and GLO-90 instances cover the SRTM-DEM (30m) and SRTM-DEM (90m) were prepared. In
whole worldwide landmass during the data collection period the Tehran and Jam cases, two networks with 8816 and 16632
(2011-2015). Each instance covers a total area of 148.5 million elevation points were employed to compare Cop-DEM (30m)
square kilometers (including inland water bodies. Cop-DEM and SRTM-DEM (30m) models with reference-DEMs. DEMs
data is provided in Integer 32-bit and 16-bit formats in the (90m) have values of 1014 and 1924. Two 1-meter spatial
GEOTIFF, DTED, DGED, and INSPIRE formats. The resolution referenced DEMs (produced from GeoEye-2 stereo
resolution of Cop-DEM data ranges from 0.3 to 3.0 arc seconds. imagery and aerial photogrammetry for Jam and Tehran,
This data is divided into units of 1°×1°. The Cop-DEM respectively) are used for evaluation and assessment. DEM
instances are accessible in Geographic Coordinates, with the accuracy is measured using the Root Mean Square Error
World Geodetic System 1984 as the horizontal reference datum (RMSE) and Mean Absolute Error (MAE) (Ghannadi et al.,
(WGS84-G1150; EPSG 4326). The Earth Gravitational Model 2020). The RMSE and MAE formulas are depicted in Equations
2008 (EGM2008; EPSG 3855) is the vertical reference datum 1 and 2.
(https://spacedata.copernicus.eu).

(a)

(b)
Figure 1. Cop-DEM of a) north of Tehran and b) west of Jam city

This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-X-4-W1-2022-209-2023 | © Author(s) 2023. CC BY 4.0 License. 210
 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume X-4/W1-2022
GeoSpatial Conference 2022 – Joint 6th SMPR and 4th GIResearch Conferences, 19–22 February 2023, Tehran, Iran (virtual)

In other words, the performance of COP-DEM is better in

(1) mountainous areas.

Case Resolution Points DEM RMSE MAE

Study (m) # (m) (m)
(2)
Cop-DEM 5.72 4.87
30
where M, N= rows and columns in DEM, respectively 8816 SRTM 6.10 5.11
, = ith point elevation over the reference Tehran
DEM and the Cop-DEM, respectively Cop-DEM 7.86 6.07
90
In the next section, experimental results are illustrated and
1014 SRTM 11.52 6.89
discussed.

4. RESULTS & DISCUSSION Cop-DEM 2.19 1.50

30
The Cop-DEM and SRTM DEM were compared with the 16632 SRTM 3.95 3.20
reference DEMs. The statistical comparison of the two DEMs is Jam
given in Table 1. RMSE and MAE of Cop-DEM were better
Cop-DEM 3.98 2.82
than SRTM by 6%-44% and 5%-53%, respectively. 90
Figure 2 shows error maps for Cop-DEM in two case studies.
These error maps are gained from the subtraction of the DEMs 1924 SRTM 5.66 4.51
from the referenced DEM.
The results reported in this study show that Cop-DEM is more Table 1. Accuracy assessment of Cop-DEM and SRTM
accurate than SRTM in urban areas as well as in mountainous High errors frequently occur in places with mountainous
areas. Table 1 shows that the accuracy difference of DEMs is topography, as shown in Figure 3. Furthermore, 100 random
more in mountainous areas. elevation points from the Cop-DEM and SRTM were chosen at
random and their accuracy was evaluated.

(a) (b)

(c) (d)

Figure 2. Cop-DEM (30 m) error maps of north of Tehran (a) and Jam city (c) Cop-DEM (90 m) error maps of north of Tehran (b)
and west of Jam city(d).

This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-X-4-W1-2022-209-2023 | © Author(s) 2023. CC BY 4.0 License. 211
 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume X-4/W1-2022
GeoSpatial Conference 2022 – Joint 6th SMPR and 4th GIResearch Conferences, 19–22 February 2023, Tehran, Iran (virtual)

(a)

(b)

(c)

(d)

Figure 3. Accuracy assessment of the random elevation points of Cop-DEM (blue) and SRTM (red) in Tehran area with resolution 30
(a) and 90 m (b) and Jam area with resolution 30 (c) and 90 m (d)

This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-X-4-W1-2022-209-2023 | © Author(s) 2023. CC BY 4.0 License. 212
 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume X-4/W1-2022
GeoSpatial Conference 2022 – Joint 6th SMPR and 4th GIResearch Conferences, 19–22 February 2023, Tehran, Iran (virtual)

grain algorithm and the SRTM DEM. Geocarto Int.,

5. SUMMERY 36, 76-95. doi: 10.1080/10106049.2019.1594393.

It is now a necessity to have DEMs with high (or medium) Florinsky, I. & Bliakharskii, D. 2019. The 2017 catastrophic
resolution and vertical accuracy from all over the world. SRTM, subsidence in the Dålk Glacier, East Antarctica:
ASTER, and AWA30 have all been used as Global-DEMs. unmanned aerial survey and terrain modelling.
Another global DEM, the WorldDEM, is now available to a Remote Sens. Lett., 10, 333-342. doi:
wide range of customers in a variety of fields. The Cop-DEM is 10.1080/2150704X.2018.1552810.
derived from a WorldDEM that has been altered. The Cop-
vertical DEM's accuracy is tested in this investigation. The Gdulova, K., Maresova, J. & Moudry, V. 2020. Accuracy
north of Tehran and the west of Jam located in Iran are the assessment of the global TanDEM-X digital elevation
research locations for Cop-DEM (30m and 90m) assessment. model in a mountain environment. Remote Sens.
Cop-DEMs and SRTM-DEMs from the study areas are Environ., 241, 111724. doi:
compared. The comparison of Cop-DEM and SRTM-DEM 10.1016/j.rse.2020.111724.
models with precise reference-DEMs is done using network
elevation points. Vertical accuracy is measured using RMSE Ghannadi, M. A., Alebooye, S., Izadi, M. & Moradi, A. 2020. A
and MAE. Cop-RMSE DEM and MAE values were found to be method for Sentinel-1 DEM outlier removal using 2-
6%-44% and 5%-53% percent better than SRTM in study areas, D Kalman filter. Geocarto Int., 1-15. doi:
respectively. 10.1080/10106049.2020.1815866.

Hawker, L., Neal, J. & Bates, P. 2019. Accuracy assessment of

REFERENCES the TanDEM-X 90 Digital Elevation Model for
selected floodplain sites. Remote Sens. Environ., 232,
Abrams, M., Crippen, R. & Fujisada, H. 2020. ASTER global 111319. doi: 10.1016/j.rse.2019.111319.
digital elevation model (GDEM) and ASTER global
water body dataset (ASTWBD). Remote Sens., 12, Mohammadi, A., Karimzadeh, S., Jalal, S. J., Kamran, K. V.,
1156. doi: 10.3390/rs12071156. Shahabi, H., Homayouni, S. & Al-ansari, N. 2020. A
Multi-Sensor Comparative Analysis on the Suitability
Aljammaz, A., Sultan, M., Izadi, M., Abotalib, A. Z., Elhebiry, of Generated DEM from Sentinel-1 SAR
M. S., Emil, M. K., Abdelmohsen, K., Saleh, M. & Interferometry Using Statistical and Hydrological
Becker, R. 2021. Land Subsidence Induced by Rapid Models. Sens., 20, 7214. doi.org/10.3390/s20247214.
Urbanization in Arid Environments: A Remote
Sensing-Based Investigation. Remote Sens., 13, 1109. Moretti, G. & Orlandini, S. 2018. Hydrography‐driven
doi: 10.3390/rs13061109. coarsening of grid digital elevation models. Water
Resour. Res., 54, 3654-3672. doi:
Arun, P. V. 2013. A comparative analysis of different DEM 10.1029/2017WR021206.
interpolation methods. Egypt. J. Remote. Sens., 16,
133-139. doi.org/10.1016/j.ejrs.2013.09.001. Rizzoli, P., Martone, M., Gonzalez, C., Wecklich, C., Tridon,
D. B., Brautigam, B., Bachmann, M., Schulze, D.,
Bayburt, S., Kurtak, A., Buyuksalih, G. & Jacobsen, K. 2017. Fritz, T. & Huber, M. 2017. Generation and
Geometric accuracy analysis of WorldDEM in performance assessment of the global TanDEM-X
relation to AW3D30, SRTM and ASTER GDEM2. digital elevation model. ISPRS J. Photogramm.
Int. Arch. Photogramm. Remote Sens.-ISPRS Remote Sens., 132, 119-139. doi:
Archives 42 (2017), Nr. 1W1, 42, 211-217. doi: 10.1016/j.isprsjprs.2017.08.008.
10.5194/isprs-archives-XLII-1-W1-211-2017.
Rossi, C. & Gernhardt, S. 2013. Urban DEM generation,
Becek, K., Koppe, W. & Kutoglu, Ş. H. 2016. Evaluation of analysis and enhancements using TanDEM-X. ISPRS
vertical accuracy of the WorldDEM™ using the J. Photogramm. Remote Sens., 85, 120-131.
runway method. Remote Sens., 8, 934. doi.org/10.1016/j.isprsjprs.2013.08.006.

Blaszczyk, M., Ignatiuk, D., Grabiec, M., Kolondra, L., Laska, Schlund, M., Von Poncet, F., Kuntz, S., Boehm, H.-D. V.,
M., Decaux, L., Jania, J., Berthier, E., Luks, B. & Hoekman, D. H. & Schmullius, C. 2016. TanDEM-X
Barzycka, B. 2019. Quality assessment and elevation model data for canopy height and
glaciological applications of digital elevation models aboveground biomass retrieval in a tropical peat
derived from space-borne and aerial images over two swamp forest. Int J Remote Sens., 37, 5021-5044. doi:
tidewater glaciers of southern spitsbergen. Remote 10.1080/01431161.2016.1226001.
Sens., 11, 1121. doi: 10.3390/rs11091121.
Sujatha, E. R., Selvakumar, R., Rajasimman, U. & Victor, R. G.
Caglar, B., Becek, K., Mekik, C. & Ozendi, M. 2018. On the 2015. Morphometric analysis of sub-watershed in
vertical accuracy of the ALOS world 3D-30m digital parts of Western Ghats, South India using ASTER
elevation model. Remote Sens. Lett., 9, 607-615. doi: DEM. Geomatics, Nat. Hazards Risk, 6, 326-341. doi:
10.1080/2150704X.2018.1453174. 10.1080/19475705.2013.845114.

Elmahdy, S. I., Mohamed, M. M. & Ali, T. A. 2021. Automated Wessel, B., Huber, M., Wohlfart, C., Marschalk, U., Kosmann,
detection of lineaments express geological linear D. & Roth, A. 2018. Accuracy assessment of the
features of a tropical region using topographic fabric global TanDEM-X Digital Elevation Model with GPS

This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-X-4-W1-2022-209-2023 | © Author(s) 2023. CC BY 4.0 License. 213
 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume X-4/W1-2022
GeoSpatial Conference 2022 – Joint 6th SMPR and 4th GIResearch Conferences, 19–22 February 2023, Tehran, Iran (virtual)

data. ISPRS J. Photogramm. Remote Sens., 139, 171-

182. doi: 10.1016/j.isprsjprs.2018.02.017.
Zink, M., Bachmann, M., Brautigam, B., Fritz, T., Hajnsek, I.,
Moreira, A., Wessel, B. & Krieger, G. 2014.
TanDEM-X: The new global DEM takes shape. IEEE
Geosci. Remote Sens. Mag., 2, 8-23.
doi:10.1109/MGRS.2014.2318895.

This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper.
https://doi.org/10.5194/isprs-annals-X-4-W1-2022-209-2023 | © Author(s) 2023. CC BY 4.0 License. 214