Detailed Description 

The demand for precise and accessible terrain data has never been greater. From telecommunications and infrastructure planning to simulation and training, accurate elevation models are key to driving informed decisions. LuxCarta’s global 10-meter digital terrain model (DTM), built using the Copernicus GLO-30 dataset, offers a leap forward in resolution, precision, and global coverage.

Through the advanced capabilities of our BrightEarth™ platform, we have enhanced the original Copernicus GLO-30 data to deliver a highly detailed 10-meter resolution DTM. Covering vast expanses from 84°N to 56°S, this model provides the accuracy needed to tackle even the most demanding geospatial challenges, from urban planning to network optimization.

This post provides an in-depth look at the technical innovations behind our approach and the ways our global 10m DTM stands out compared to other models in the market.

Figure 1 Global Coverage

Precision and Accuracy

The Copernicus GLO-30 DEM displays a high vertical accuracy, with a global RMSE of around 4 meters. However, this accuracy can vary depending on the terrain, geographical region, and the surface being modeled (vegetation, bare ground, etc.). Generally, in flat terrains, the accuracy is higher, while in areas with more rugged topography, the precision can be slightly lower due to the challenges of representing steep slopes and other complex landforms. The GLO-30 DEM achieves this level of precision by integrating radar interferometry data and correction algorithms, further refining the raw satellite data.

When compared to other global DEMs, the GLO-30 stands out for its improved resolution over earlier models such as the Shuttle Radar Topography Mission (SRTM), ASTER GDEM or ALOS AW3D30 which offer lower spatial resolutions and lower accuracy. For instance, SRTM offers an RMSE of approximately 10-16 meters, making GLO-30 more precise. Similarly, ASTER GDEM, while offering a comparable 30-meter resolution, exhibits more significant elevation errors in many regions, particularly those with vegetation or cloud cover, due to the limitations of optical sensors. Same for ALOS AW3D30, 30m resolution and offering a global RMSE ranging from 5 to 6m.

Main data Sources: RADAR

The GLO-30 model is primarily generated using data from the TanDEM-X mission, a collaboration between the German Aerospace Center (DLR) and Airbus Defence and Space. TanDEM-X utilizes synthetic aperture radar (SAR) technology, specifically interferometric SAR (InSAR), to measure surface elevation. This technology allows for precise elevation measurements by analyzing the phase differences in radar signals received from two different satellite positions. TerraSAR-X, which operates in tandem with TanDEM-X, also contributes data, and the two satellites function as a single radar interferometer. The GLO-30 DEM is processed using advanced algorithms that refine and correct the radar data to minimize errors and ensure consistent coverage across the globe.

In addition to SAR data, GLO-30 also incorporates ancillary datasets, including ground control points and other elevation data, to improve accuracy in challenging regions. Corrections are applied to mitigate the effects of vegetation, buildings, and other surface features that might interfere with bare-earth elevation measurements. The radar-based nature of the GLO-30 model makes it particularly robust in areas where optical methods fail due to cloud cover or other atmospheric conditions.

Merging method with other higher resolution sources

The QTZ (Quantized Mesh format) used by Cesium is highly efficient for visualizing large-scale terrain datasets in 3D applications. It reduces file sizes by encoding elevation data using quantization, delta encoding, and octrees, optimizing it for streaming and rendering. A terrain tile set in quantized-mesh-1.0 format is a simple multi-resolution quadtree pyramid of meshes. To upscale a dataset such as Copernicus GLO-30 (30m resolution) to a 10m resolution using available data sources like LiDAR, NED10, or other DEM sources, the process involves several key steps.

Figure 2: Quantized Mesh for upscaling DTM from 30m to 10m

First, we start with GLO-30, a global dataset with approximately 30m horizontal resolution and varying vertical precision (1m in low-relief areas). To refine this to a 10m DEM, you would integrate higher-resolution datasets. LiDAR, with vertical accuracy in the centimeter range and horizontal resolution typically ranging from 1-3m, would provide the most precise data, particularly in regions where it's available. For areas without LiDAR coverage, the NED10 dataset, offering 10m resolution and vertical accuracy within 2.44m (95% confidence), is commonly used.

During the upscaling, terrain interpolation methods, such as bilinear or bicubic interpolation, can fill gaps between different sources. In regions with LiDAR data, it can directly replace lower-resolution DEMs, improving vertical precision (Z accuracy) to 10-50 cm. In areas without such detailed data, NED10 or similarly available DEMs provide sufficient vertical accuracy of around 2m (Z) and a horizontal resolution of 10m (XY).

It's also important to remind that our clean-up method consist in removing artifacts (natural or man-made, i-e trees or buildings) above ground leaving no bump due to the trace of those artifacts.

Figure 3: Not Applying clean-up method

This is essential if what we want is placing objects on it. Summarizing, our terrain is ready in any raster or QTZ format (mesh) to be used directly as a globe in Cesium.

Figure 4: Applying clean-up method

Once the source data is refined, the quantized mesh algorithm compresses the dataset by encoding the elevation values into 16-bit integers. For a final terrain model with 10m horizontal resolution, you can expect XYZ precision of approximately 10m in horizontal accuracy and sub-meter vertical precision (depending on data sources and processing), ideal for high-resolution 3D terrain visualization.

Global Coverage

One of the significant advantages of Luxcarta DTM 10m is its truly global coverage. It encompasses all land areas between 84 degrees North and 56 degrees South, including regions that are often excluded from other datasets due to political or logistical constraints. This means our DTM can be utilized for terrain analysis and modeling anywhere in the world, including previously underrepresented areas such as high-latitude regions or countries with limited access to high-quality geospatial data. This broad coverage makes Luxcarta DTM 10m an invaluable resource for international organizations, research institutions, and governments involved in global-scale projects.

While earlier DEMs such as SRTM exclude polar regions and ASTER GDEM suffers from gaps and inconsistencies, GLO-30 provides seamless coverage, ensuring that users can access consistent data regardless of their area of interest. Furthermore, the Copernicus program's commitment to regularly updating the DEM ensures that users benefit from the latest available data, reflecting changes in terrain due to natural processes or human activities.

Comparison With Other DEMs

SRTM (Shuttle Radar Topography Mission)

SRTM, one of the most well-known global DEMs, was produced by NASA and offers data at 30-meter resolution globally. It is based on radar data collected during the 2000 mission aboard the Space Shuttle Endeavour. The vertical accuracy of SRTM typically ranges between 5 and 16 meters RMSE, depending on the region. Its global coverage includes most of the world between 60°N and 56°S, with notable exclusions at high latitudes and steep mountainous regions. While widely used, SRTM’s data is now over two decades old, making it less reliable for regions that have undergone significant changes since the year 2000. Additionally, its coarser resolution outside the U.S. limits its utility for detailed terrain analysis. In comparison, GLO-30’s finer 30-meter resolution globally and vertical RMSE of 4 meters provide a more accurate and up-to-date alternative.

• In areas with complex topography, SRTM suffers from radar shadowing and layover effects, which can distort terrain representation, particularly in steep regions. In contrast, GLO-30 incorporates advanced algorithms to correct such errors, ensuring better representation of complex terrains. Moreover, SRTM data often suffers from voids (missing data), especially in mountainous or forested areas, which require interpolation or external data sources for filling. GLO-30 addresses many of these gaps, offering seamless coverage, including previously hard-to-map areas, though limitations in steep terrains or highly vegetated regions remain.
  
  
• Another significant limitation of SRTM is its outdated coverage. Over the past two decades, human activities such as urban expansion, deforestation, and natural events like landslides have dramatically altered the terrain in many regions. GLO-30, with data based on the more recent TanDEM-X mission, provides a more current depiction of global topography.

Figure 5: Challenging Luxcarta DTM vs SRTM

ASTER GDEM

The ASTER GDEM offers global coverage at a 30-meter resolution similar to GLO-30. However, ASTER GDEM is derived from optical stereo imagery, which can suffer from cloud cover and atmospheric distortion, leading to higher vertical errors. ASTER’s RMSE can range from 10 to 25 meters, which is considerably higher than GLO-30’s typical RMSE of 4 meters. Additionally, ASTER data has been known to contain anomalies and gaps, particularly in densely vegetated or snow-covered regions, where optical methods struggle to capture accurate elevation data.

Figure 6 Challenging Luxcarta DTM vs ASTER GDEM

ALOS AW3D30

Produced by the Japan Aerospace Exploration Agency (JAXA), the ALOS World 3D DEM provides global coverage at a 30-meter resolution. The vertical accuracy of AW3D30 is typically better than that of GLO-30, with an RMSE of around 5 meters in many regions. However, AW3D30 is not freely available globally, unlike GLO-30, which makes the latter more accessible to users requiring high-resolution elevation data for free.

• The ALOS World 3D DEM, produced using data from Japan’s ALOS (Advanced Land Observing Satellite) mission, provides global coverage at a 30-meter resolution, like GLO-30. ALOS DEM is highly regarded for its vertical accuracy, offering a global RMSE of 5 to 6 meters, which is comparable to GLO-30's 4 meters. ALOS uses optical data, in contrast to the radar-based GLO-30, which can lead to different strengths and weaknesses between the two. Optical data is sensitive to cloud cover and atmospheric conditions, which can introduce gaps and errors in the ALOS dataset, particularly in tropical and polar regions where cloud cover is persistent. In these cases, GLO-30’s radar-derived data provides superior coverage since radar can penetrate clouds, offering a more reliable elevation model for regions with challenging weather conditions.
  
  
• Another key distinction is accessibility. ALOS World 3D is not fully free for all users, with some regions requiring purchase or specific licensing agreements, while GLO-30 is freely available as part of the Copernicus Program, making it a more accessible tool for a wide range of users globally. This level of accessibility, coupled with GLO-30’s high accuracy, makes it a compelling option for users who need cost-effective and accurate DEM data for large-scale projects.
  
  
• Global Coverage and Accessibility
  While both GLO-30 and ALOS offer 30-meter global coverage, there are important differences in the extent and reliability of this coverage. GLO-30 provides data for all landmasses between 84°N and 56°S, excluding only extreme polar regions, which are challenging to map due to satellite data limitations. ALOS, though also offering near-global coverage, has limitations in regions with frequent cloud cover, making GLO-30 more dependable for uninterrupted data availability. Furthermore, the Copernicus Program ensures GLO-30 is consistently updated and maintained, keeping pace with technological advancements and changing terrains, whereas updates to ALOS datasets are less frequent.
  
  
• One of the most critical features of GLO-30 is its seamless integration into the Copernicus Earth Observation infrastructure, allowing users to easily access, process, and analyze the data alongside other Copernicus datasets (e.g., land cover, climate data). This level of integration fosters interdisciplinary research and decision-making by providing coherent datasets across environmental, agricultural, and urban planning sectors.

In summary, while SRTM remains widely used due to its historical significance and availability, GLO-30 surpasses it in terms of spatial resolution, accuracy, and up-to-date data. Its superior precision and seamless coverage, combined with free access, make it an indispensable tool for users in various sectors. Compared to ALOS, GLO-30 also holds an edge in accessibility and reliability in regions with cloud cover. Though both offer comparable vertical accuracy, GLO-30’s consistent availability and frequent updates position it as a leading choice for global elevation modeling.

By combining advanced radar technology, seamless global coverage, and robust accuracy, the Copernicus GLO-30 DEM serves as a critical resource for applications ranging from disaster management to environmental monitoring, surpassing both SRTM and ALOS in various key aspects.

Figure 7: Challenging Luxcarta DTM vs ALOS AW3D30

Limitations and Challenges

Although it offers many advantages, it is not without its limitations. Like many radar-based DEMs, it represents the surface elevation, including vegetation and man-made structures, rather than the bare ground. In heavily forested areas, this can result in an overestimation of true ground elevation, for addressing these issues. Our method of DTM extraction from this DSM-ish works equally well on 30 m DSMs or high-resolution DSMs (e.g. coming from LIDAR or NED10). 

Luxcarta proposes a pipeline that has to be operational anywhere in the world. We validated our method with this data on all of New Zealand by using as reference 22,000 GCPs (Ground Control Point). We obtained an RMSE of 3.49 m. According to the 30 m resolution of AW3D30 and the number of GCP, the RMSE validated our approach. An advantage of our approach is the possibility of easily adding constraints such as considering the GCPs to force the physics simulation to reach these GCPs without creating artefacts. In this use case the RMSE was 0.8 m.

To make sure we were perfectly aligned with our customer’s needs in terms of Z accuracy we performed the following analysis using 120,000 GCP (Ground Control Point) over the province of Madrid, Spain (sourced from official Open Data from CNIG) and the result was RMSE Z was 4.5 m. See some of the auto-explanative slides below : 

Figure 8: Challenging LuxCarta DTM vs CNIG GCP

Furthermore, while the model offers global coverage, certain localized errors may exist in regions with highly complex terrain, where radar signal interference can occur. Additionally, data collection limitations in extreme polar regions reduce its coverage to 84°N to 56°S.

Figure 9: Analysis of highest deltas over CNIG GCP

LuxCarta DTM 10m provides an exceptional balance of resolution, global coverage, and accessibility. With its combination of radar-derived data and advanced processing, it surpasses many older global elevation models in both accuracy and usability. Despite some limitations in vegetated areas, this DTM’s wide-ranging applications make it an invaluable tool for professionals in geospatial-related fields. As LuxCarta continues to update and refine this model, the global 10m DTM will remain a reliable and essential resource for terrain analysis, empowering industries and institutions to make data-driven decisions in a rapidly evolving world.

At LuxCarta, we deliver precise global terrain data to empower your geospatial projects with accuracy and reliability. Contact us today.