Digital Elevation_Model

Provide Feedback on this Article

In short: A Digital Elevation Model (DEM) is a digital representation of the elevation of the Earth’s surface (Guth et al., 2021). DEMs provide a visualization of topographic features by assigning an elevation value to each cell of a grid or data point, similar to a chessboard, where each square represents a specific area on Earth. Scientists have struggled to find a common definition, so two main types of DEMs must be distinguished based on the surfaces they include: Digital Terrain Models (DTMs) display the bare ground without vegetation or human-made infrastructure, whereas Digital Surface Models (DSMs) additionally include features such as vegetation, water, ice, or buildings (Guth et al., 2021). DEMs are generated using different remote sensing techniques such as photogrammetry, LiDAR, and radar, each with its own strengths and challenges regarding accuracy, costs, resolution, and environmental constraints (Croneborg et al., 2020). Beyond visualizing terrain, DEMs form the basis for many further analyses, including the calculation of slope, aspect, or watershed boundaries, and find application in a wide range of disciplines, ranging from hydrology to video game design.

Background[edit]

DEM 1..png

DEMs have become a core tool in geographic information systems (GIS) and remote sensing. Their development is connected to advances made in orthophotography at the end of the past century, where aerial images were geometrically corrected (orthorectified) so that measurements of distances, angles, and areas were improved. Initially, orthoimages were produced using analogue methods, but from the 1970s to the 2000s, digital orthoimages and photogrammetrically derived DEMs have experienced a “sudden explosion” regarding their availability and use (Jensen, 1995). The number of publications on DEMs has been steadily increasing since the 1970s, with a peak in 2024, reaching nearly 9,000 documents on Scopus (Figure 1). Until around 1990, topographic maps in Germany were published exclusively in analogue printed form by the state surveying authorities (Landesvermessungsbehörden) of the individual states. They displayed terrain using contour lines, supplemented by high points like peaks and saddles, and the course of waterways, roads, railway lines, larger buildings, towns, and technical features like boundaries, water pipes, and power lines (Wikipedia-Authors, 2004). Originating from that, DEMs have come a long way, including great areas of application on and beyond Earth, with modern elevation models measuring Mars and Venus (Wikipedia-Authors, 2003).

Generally, DEMs represent the elevation of the terrain in a structured grid or matrix format, allowing analysis of topography and hydrology. Modern DEM creations involve digital image processing, ground control points, and satellite data for improved accuracy and resolution. These models are widely used in scientific, engineering, and planning applications, providing the foundational data for analyses regarding flood modeling, slope assessment, watershed delineation, and infrastructure planning (Jensen, 1995).

Over the past decades, practitioners struggled to find a common understanding of DEMs. The interdisciplinary application field and the diverse actors involved in the production of data for DEMs highlight the need for a common understanding. The former understanding of DEMs as a statistical representation of the continuous surface of the ground by a large number of selected points with known xyz coordinates from 1958 soon evolved into showing 3D terrain models. Later it was replaced by the participants of the 2019 Joint Research Centre meeting and the Digital Elevation Model Intercomparison eXperiment working group, who agreed to treat DEMs as a general term, encompassing both Digital Terrain and Digital Surface Models (Guth et al., 2021).

What the Method does[edit]

Terminology[edit]

According to Guth et al. (2021), a DEM describes “a digital representation of the elevation or height of a topographical surface.” Depending on the incorporated spheres, the DEM can be subdivided into a Digital Terrain Model (DTM) and Digital Surface Model (DSM). A DTM, or “bare-earth DEM,” as it is called in the US-American context, represents the boundary between the lithosphere and the atmosphere (Fig. 2B), while a DSM also includes the hydrosphere, cryosphere, biosphere, or anthroposphere (Guth et al., 2021).

Figure 2: Representation of the terrain (A); and the differentiation of DTM and DSM (B) (Retrieved from Guth et al., 2021).


Technical Background[edit]

Data Acquisition

There are different methods available for collecting elevational data. All approaches have in common that they collect coordinates with a horizontal datum, i.e., latitude and longitude in a certain coordinate reference system (CRS), and a vertical datum, i.e., height measurement. However, as Table 1 shows, they differ in terms of accuracy, i.e., the vertical error, spatial resolution, spatial extent, and ability to handle environmental disturbances, such as weather, smoke, and foliage (Croneborg et al., 2020).

Depending on the purpose of the DEM, different data types should thus be considered. On a local scale, higher accuracy and higher spatial resolution are required (Croneberg et al., 2020). These data can be obtained with photogrammetry methods or LiDAR (Croneberg et al., 2020; Gupta, 2017). Photogrammetry is a passive remote sensing technique, recording the optical reflection of the surface (Gupta, 2017). Hence, it cannot penetrate clouds, dense foliage of tree canopies, or smoke (Croneberg et al., 2020). In contrast to this, LiDAR is an active remote sensing technique that can therefore penetrate canopy and smoke (Croneberg et al., 2020). It can be applied for all spatial scales (Croneberg et al., 2020). From an aerial or satellite platform, laser beams are sent to the surface, and the two-way travel time and the intensity of the backscattered signal are recorded. Based on the travel time, the elevation of the ground object can be calculated. The intensity of the signal allows us to draw conclusions about the object attributes (Gupta, 2017). Last but not least, satellite “Synthetic Aperture Radar” (SAR) data is often used to obtain global DEMs. Similar to LiDAR, SAR interferometry (InSAR) is an active remote sensing technique, sending a signal to the surface and measuring the backscattering (Gupta, 2017). However, in comparison to LiDAR, not laser beams but electromagnetic pulses are used (Croneberg et al., 2020).

Table 1: Overview of Methods for elevational data collection. Spatial Resolution and Accuracy: high <1 m; medium 1-5 m; low >5 m (Based on Croneberg et al., 2020).
Method Spatial Resolution Accuracy Application Limitation
Ground Survey - High Local Scale Cost-, time-, labour-intensive
Digitisation of topographic contour maps - Low Old records, archive information Low accuracy
Conventional / digital / UAV aerial photogrammetry High High Local Scale Cannot deal with clouds, canopy, smoke
Satellite SAR Medium/Low Medium/Low Regional to global scale Lower precision than LiDAR
Aerial LiDAR High High Local to global scale High data volume

Generation and Representation of DEMs[edit]

DEM 3.jpg

To obtain a DEM, data is stored in a storage model, i.e., a grid, matrix, or array data structure (Guth et al., 2021). One can distinguish between point-based and area-based sampling techniques (Guth et al., 2021). Point-based methods, e.g., LiDAR, create so-called geospatial point clouds with X and Y coordinates and Z values for height. The coordinates are sampled irregularly so that no regular grid is created. Point clouds do not yet represent a DEM, but the required continuous surfaces can be created through interpolation techniques (Burrogh, 1998; Guth et al., 2021). In contrast, area-based methods, used by InSAR and photogrammetric techniques, use regular grid sampling methods. Each pixel in the grid represents an elevational value that is derived from the median or (weighted) mean value of the area covered (Guth et al., 2021).

Depending on the data structure, the DEM can be stored in different file types (Table 2). The data is now ready for analysis and visualization.

Derivatives of DEM[edit]

Besides the analysis and visualization of elevational data, DEMs can form the basis for further calculation of derivatives. Examples are shaded relief, slope, aspect, curvature, and viewshed, which are further explained in Table 3 (Gupta, 2018). Furthermore, based on the DEM and canopy height model, water depth and ice thickness can be assessed.

DEM 4.jpg

Visualisation of Typical Results[edit]

DEMs have a variety of different application fields among different disciplines. From the modelling of flood simulations in coastal areas (Xu et al., 2021) to the analysis of natural hazards like floods or landslides, as well as for tourism, sports, defense, or even video games, DEMs can be used in many different forms (Guth et al., 2021). This means that a vast number of case studies with different applications and resulting maps and research results can be looked at (Guth et al., 2021).

Visualisation of Tsunami Vertical Evacuation Routes with DEM[edit]

Here, a case study for planning tsunami vertical evacuation routes with the usage of a high-resolution unmanned aerial vehicles (UAV) DEM in the Drini Coastal Area in Java, Indonesia, is used to visualise how DEMs can be used and what the results can look like (Marfai et al., 2021). Tsunami vertical evacuation (TVE) routes are developed to improve the inadequate existing models provided by the institutional authorities. The goal is to identify better tsunami evacuation routes and assembly points for near-field tsunami hazards and determine safe areas for potential shelters. The results show that vertical evacuation could be more appropriate than the horizontal evacuation plans that have been established by the authorities. For this, high-resolution UAV photogrammetry is used to generate orthophotos, a DSM, and a DTM. The DEM terminology is used as follows: The DSM represents the surface, including vegetation and buildings, while the DTM—derived from the DSM by filtering out land cover—represents the bare terrain. In Figure 3 the UAV-based photogrammetry orthosomaic map and the derived DSM are seen (Marfai et al., 2021).

Figure 3: Results from (a) UAV orthosomaic and (b) DSM of Drini coastal area (retrieved from Marfai et al., 2021).

In a further step, the DSM is converted into a DTM in order to remove vegetation and other surface objects that could distort the inundation analysis and the extent of predicted flooding. Figure 4 demonstrates this conversion and highlights the significant elevation differences that often exceed 10 m, caused by tree and shrub cover. The resulting DTM represents the bare-earth surface, enabling more accurate slope calculations and hazard simulations. This distinction between DSM and DTM is essential, as only the DTM allowed a realistic modeling of tsunami inundation and the identification of truly safe areas for evacuation. (Marfai et al., 2021)

Figure 4: Digital surface model (a) and digital terrain model (b) (retrieved from Marfai et al., 2021).


The final stage of the analysis integrates the DTM into least-cost distance modeling and is subsequently validated through field simulations. As shown in Figure 5, three vertical evacuation points and seven possible routes are tested. Routes equipped with stairs to the hilltops prove to be the fastest and most effective, allowing evacuees to reach safe zones within 3–6 minutes, while others require 8–10 minutes and are less suitable due to steep slopes or dense vegetation. The results show that UAV-derived DSM/DTM data, when combined with evacuation simulations, can significantly improve the realism and effectiveness of tsunami evacuation planning (Marfai et al., 2021).

Figure 5: Tsunami evacuation routes based on field evacuation simulation (Retrieved from Marfai et al., 2021)

DEMs in Water Resource Management[edit]

A different field of application is Water Resource Management. This entry highlights water resource management as an example because it can be directly linked to other disciplines that rely on hydrological results, such as urban planning and ecology.

Water flow is fundamentally determined by the shape of the Earth’s surface. As such, water resource management depends on accurate elevation or topographic information, represented as elevation map layers. DEMs are used to combine hydraulic data (the physical mechanics of water flow) with hydrological data (the movement of ground and surface water within ecosystems). In contrast, bathymetric data represents underwater topography. (Global Facility for Disaster Reduction and Recovery, 2015)

Flow channels represent an open, natural, or man-made channel and guide water with a specific direction and speed, depending on their slope and size. In this case, DEMs are used to define specific metrics that influence the stream flow. Those metrics encompass characteristics such as bank conditions, water bodies, river centerlines, channel and streambed width, and artificial in-stream barriers. These metrics directly influence stream morphology and can be quantified using DEMs, as Figure 6 shows:

Figure 6: Representation of the interplay between the earths surface and waterflow in a channel (retrieved Global Facility for Disaster Reduction and Recovery, 2015).

Channel characteristics shape the hydraulics of water movement, which consequently drive hydrology. Ideally, flow channel characteristics are based on data for the entire streambed. For that, bathymetric LiDAR can provide detailed information on underwater topography, although water depth and turbidity may limit visibility of terrain features at the bottom of a water body. Through that, scientists can model flow characteristics such as velocity, turbulence, or pressure. These models are then used to derive risk assessments, for example of surface runoff, erosion, or flood parameters, which become increasingly relevant through the effects of climate change. For instance, the number of extreme weather events, including intense rainfall, will increase for the city of Lueneburg, pushing the use of DEMs into the foreground (GEO-NET Umweltconsulting GmbH, 2019). They successfully characterise the floodplain of low-lying land areas, determining areas of special risk for flooding above a certain threshold. In the field of stormwater management, DEMs can monitor and model watersheds, streams and other flow channels. Furthermore, they can precisely predict where water will flow to and how formed storm water interacts with existing structures such as culverts and bridges. For this calculation the digital surface model comes into play as it additionally includes human-made infrastructure. (Global Facility for Disaster Reduction and Recovery, 2015)

Since the shape of the earth governs water flow, DEMs assist in identifying flow channels, connecting those in the form of stream networks and delineating catchment areas. From this, multiple interdisciplinary applications arise to…

  • assess drought severity
  • monitor streamflow and water supply
  • drive policy decisions
  • plan infrastructure development to differentiate between risk and ideal areas for zoning- and land-use planning
  • direct land use and land cover to enhance water resource management
  • map downstream impacts of pollution

(Global Facility for Disaster Reduction and Recovery, 2015)

In general terms: A water drop can be placed at any point on Earth and its precise path can be computed by using the DEM (Global Facility for Disaster Reduction and Recovery, 2015).

Strengths and Challenges[edit]

Major strengths of DEMs include data availability and the ease of application. Several DEMs with different spatial resolutions are freely available to download. Examples include the ASTER Global DEM and the NASA Shuttle Radar Topographic Mission (SRTM), both with up to 30 m resolution (Kumar et al., 2018). Both models encompass the majority of the terrestrial surface, with the exception of the polar regions. This allows analysis of inaccessible or hazardous regions. The method’s effortless accessibility for everyone is another strength. Open-access data providers such as earthexplorer.org supply pre-processed, void-filled data, eliminating the necessity for advanced GIS expertise to utilise DEMs. The ready-to-use data is complemented by intuitive commands in the open-source software QGIS, enabling users to extract the necessary derivatives from DEMs. As a result, it is possible to create topographical visualisations using simple means, for example, for urban planning or flood protection.

Nevertheless, there are challenges associated with DEMs. The accessibility of the data and ease of use make DEMs useful for a wide range of users. However, this can lead to incorrect data being used for financial reasons or due to a lack of knowledge. Incorrect data selection can significantly influence the results of the DEM and, in turn, lead to incorrect real-world decisions. This can be a problem, since different DEMs differ in their elevational assessment for various reasons. First of all, there are different methods for the vertical datum defining the zero elevation, which can cause variation in the elevation up to 90 m among different DEMs (Guth et al., 2021). Furthermore, differences in accuracy and spatial resolution can impact the elevational assessment in different DEMs, as well as different CRS. Consequently, it is pivotal to carefully evaluate the parameters of a certain project and determine, based on this analysis, which DEM fulfills the criteria. Cronebach et al. (2021) offers an overview of requirements for different applications.

Figure 7: Elevational Color Ramp from QGIS-hub [1]

Last but not least, maps and models are merely simplified representations of reality. Consequently, the representation of elevation cannot depict reality. In 2D maps, elevation is often shown with color gradients or contour lines, but these can be misleading if not chosen carefully (Monmonier, 1996). For example, color schemes can influence perception. Therefore, it is recommended to apply standard elevation ramps, as depicted in Figure 7, or light colors for low and dark tones for high elevation (Monmonier, 1996). Likewise, the choice of elevation intervals assigned to each color or contour line can exaggerate or downplay terrain differences. Both design choices can deliberately or unintentionally distort the map’s message.

Normativity[edit]

“Normativity” in its most basic definition deals with any kind of evaluation by humans. Normativity deals with how the method is evaluated by entities such as disciplines, scientific communities, or civil societies, for example. (Sustainabilitymethods, 2021)

When looking at DEMs, normativity is an important factor to consider because they are not creating “neutral” representations of terrain. They come with assumptions, conventions, and choices that shape how we see and use landscapes. (Guth et al., 2021)

When working with DEMs or any form of visual representation of landscape, it is important to understand that the DEM process and all the associated errors that can occur make it effectively impossible that all digital records in a DEM are correct. While this is a problem, it does not bring the conclusion that DEMs cannot be used safely for knowledge generation, but it is rather something that needs to be considered in every step when looking at or creating DEMs. (Fisher & Tate, 2006)

There are various aspects when looking at DEMs where its normativity should be taken into account, such as the definition of DEMs that can vary greatly depending on which discipline, community and user is working with DEMs (Guth et al., 2021) or the uncertainty in data collection methods or the interpolation that can be reduced through statistical models of errors, but even then the data can have underlying bias or assumptions (Fisher & Tate, 2006). There is no such thing as “error free reference data”, so error propagation should be considered at every step (Fisher & Tate, 2006).

The quality of DEMs depends strongly on spatial resolution, which affects both terrain detail and measurement noise. Elevation errors propagate into derived products such as slope, aspect, or watershed boundaries, often amplifying uncertainty. Visualisation techniques like difference maps or uncertainty surfaces, along with correction methods such as filtering or hydrologic conditioning, help to manage these errors. A proper match between DEM resolution and terrain characteristics is essential, since coarse models may underestimate slopes in rugged areas, while overly fine models in flat areas can introduce noise. Distinguishing between gross, systematic, and random errors further clarifies error sources, and the principle of fitness for use highlights that DEM adequacy depends on application context. Error propagation can be studied through Monte Carlo simulation, analytical models, or sensitivity analysis, all of which emphasize the need for critical evaluation of DEM reliability (Fisher & Tate, 2006).

Outlook[edit]

In the future, the use of DEMs should focus on improving methodological consistency, approaches for quantifying and communicating uncertainty, and embracing errors as inherent characteristics of spatial data rather than as flaws (Wechsler, 2007). Instead of relying on a single grid as the “true” terrain representation, multiple plausible realisations and the use of several DEM datasets can help reduce bias and strengthen reliability (Xu, 2021). Artificial intelligence and machine learning are expected to open new opportunities for error detection, correction, and data fusion, though transparency and reproducibility must remain essential (Shawky, 2019). By sharpening analytical tools and critically addressing uncertainty, DEM research can push the boundaries of “truth” in terrain modeling and contribute to more robust and responsible applications in science and practice (Wechsler, 2007).  

References[edit]

  • Burrough, P. A., & McDonnell, R. (1998). Principles of geographical information systems: Spatial information systems and geostatistics (Repr. with corr). Spatial information systems and geostatistics. Oxford Univ. Press.
  • Croneborg, L., Saito, K., Matera, M., McKeown, D., & van Aardt, J. (2020). Digital Elevation Models: A Guidance Note on How Digital Elevation Models are Created and Used - Includes Key Definitions, Sample Terms of Reference, and How Best to Plan a DEM-Mission. World Bank. http://hdl.handle.net/10986/34445
  • Fisher, P. F., & Tate, N. J. (2006). Causes and consequences of error in digital elevation models. Progress in Physical Geography: Earth and Environment, 30(4), 467-489. https://doi.org/10.1191/0309133306pp492ra
  • Guth, P. L., van Niekerk, A., Grohmann, C. H., Muller, J.‑P., Hawker, L., Florinsky, I. V., Gesch, D., Reuter, H. I., Herrera-Cruz, V., Riazanoff, S., López-Vázquez, C., Carabajal, C. C., Albinet, C., & Strobl, P. (2021). Digital Elevation Models: Terminology and Definitions. Remote Sensing, 13(18), 3581. https://doi.org/10.3390/rs13183581
  • Jensen, J. R. (1995). Issues involving the creation of digital elevation models and terrain corrected orthoimagery using soft‐copy photogrammetry. Geocarto International, 10(1), 5–21. https://doi.org/10.1080/10106049509354475
  • Kumar, N., Singh, S. K., Mishra, V. N., Reddy, G. P. O., & Bajpai, R. K. (2018). Open-Source Satellite Data and GIS for Land Resource Mapping. In G. P. O. Reddy & S. K. Singh (Eds.), Geospatial Technologies in Land Resources Mapping, Monitoring and Management (Vol. 21, pp. 185–200). Springer International Publishing. https://doi.org/10.1007/978-3-319-78711-4_10
  • Marfai, M. A., Khakim, N., Fatchurohman, H. & Salma, A. D. (2021). Planning tsunami vertical evacuation routes using high-resolution UAV digital elevation model: case study in Drini Coastal Area, Java, Indonesia. Arabian Journal Of Geosciences, 14(19). https://doi.org/10.1007/s12517-021-08357-9
  • Shawky, M., Moussa, A., Hassan, Q. K., & El-Sheimy, N. (2019). Pixel-Based Geometric Assessment of Channel Networks/Orders Derived from Global Spaceborne Digital Elevation Models. Remote Sensing, 11(3), 235. https://doi.org/10.3390/rs11030235.
  • Xu, K., Fang, J., Fang, Y., Sun, Q., Wu, C. & Liu, M. (2021). The Importance of Digital Elevation Model Selection in Flood Simulation and a Proposed Method to Reduce DEM Errors: A Case Study in Shanghai. International Journal Of Disaster Risk Science, 12(6), 890–902. https://doi.org/10.1007/s13753-021-00377-z


The authors of this entry are: Georgina Emma Huber, Giulia Urgan, Janne Wohlberg.