Listen to audio version of this blog post:
For 5G RF planning, the minimum acceptable resolution depends on frequency band and environment type. At 3.5 GHz in urban areas, a 5 m Digital Terrain Model, 2–5 m building height data, and 5–10 m LULC clutter classification are the practical minimums. At 26 GHz mmWave, resolution requirements tighten to 1–2 m DTM, 1 m building data, and individual tree-level vegetation - coarser data produces unacceptable LoS prediction errors.
What Are the Resolution Minimums by Layer and Frequency?
| Geodata Layer | Sub-1 GHz | 3.5 GHz (n78) | 26 GHz mmWave |
|---|---|---|---|
| Digital Terrain Model (DTM) | 25 m | 5–10 m | 1–2 m |
| 3D Buildings / DSM | 10–25 m | 2–5 m | 0.5–1 m |
| LULC / Clutter | 25–50 m | 5–10 m | 2–5 m |
| Vegetation (height) | Not critical | 5–10 m | Individual tree (1–5 m) |
These values represent practical minimums for production network planning. Using coarser data is possible in rural or low-density areas where the propagation environment is relatively uniform, but in urban environments, coarser data directly increases prediction error.
Why Does Frequency Drive Resolution Requirements?
Higher frequencies have shorter wavelengths and interact more specifically with physical geometry. At 26 GHz, a 3 m height error in a building dataset can shift the calculated diffraction path loss by several decibels. A single tree - measurable only in datasets with individual-tree resolution - can attenuate a 26 GHz signal by 35.3 dB. At 700 MHz, the same tree has a marginal effect that is safely absorbed within the model's link budget margin.
Is There a "Good Enough" Threshold Below Which Accuracy Degrades Sharply?
Yes - particularly for mmWave LoS/NLoS classification. If building height data has a vertical RMSE greater than 3 m, the model begins misclassifying borderline LoS/NLoS links - those where the signal just grazes a building rooftop. Each misclassification introduces a discrete 10–30 dB error in the predicted link budget, which cannot be corrected by post-processing or model calibration.
For 5G Fixed Wireless Access (FWA) at 26 GHz, where operators must predict per-premise coverage before truck rolls, this threshold is critical. A 2 m height error on a building that is exactly at the edge of coverage can mean falsely qualifying or disqualifying customer premises.
How LuxCarta Addresses This
LuxCarta delivers building data with a 93%+ capture rate and height accuracy sufficient for 26 GHz planning - validated by Telefónica Deutschland for FWA trials. LULC data is produced at 50 cm resolution from sub-meter satellite imagery (18–19 classes) or 10 m resolution from Sentinel-2 for wider-area projects. DTM products are available at multiple resolutions including hydro-enforced versions for flood-prone or waterway-adjacent areas. The BrightEarth platform allows engineers to extract precisely the area and resolution needed for their project without ordering a full national dataset.
Frequently Asked Questions
Can I use 30 m SRTM elevation data for 5G planning?
SRTM at 30 m horizontal resolution and ~5–10 m vertical RMSE is acceptable only for sub-1 GHz macro planning in rural or semi-rural environments. For urban environments or any frequency above 1 GHz, SRTM is too coarse - building heights are averaged into the DSM surface, terrain features are smoothed, and propagation predictions carry errors of 5–15 dB.
Does horizontal resolution or vertical accuracy matter more for DTM?
Both matter, but in different ways. Horizontal resolution determines whether terrain features (hills, valleys, ridgelines) are captured at useful scale for the propagation path. Vertical accuracy determines whether elevation differences between transmitter and receiver are computed correctly. For 5G planning, vertical RMSE ≤1 m at 5–10 m horizontal resolution is the recommended combination for urban deployments.
What happens if I use a coarser LULC dataset for 5G planning?
Coarser LULC (e.g., 25 m or 50 m per pixel) misclassifies mixed-use areas, a pixel that contains both a road and adjacent building is assigned a single class. This produces incorrect attenuation coefficients along propagation paths and degrades model accuracy, particularly in suburban transition zones where clutter type changes frequently within a short distance.
LuxCarta provides high-quality 3D geospatial data solutions for telecom operators worldwide.
Contact our team to access data tailored to the exact resolution requirements of your 5G projects.