Orthomosaic, DSM, DTM & DEM Explained

An orthomosaic is a single, large, geometrically corrected image of a site, stitched from many overlapping aerial photos. The key word is orthorectified: ordinary aerial photos contain distortion from camera tilt, lens geometry and terrain relief, so distances measured on them are unreliable. Orthorectification removes that distortion using the survey's geometry and an elevation model, producing an image with a uniform scale across its whole extent. The result looks like a map but carries the visual detail of a photograph.

Because every pixel sits at a true, consistent geographic position, you can measure distances, areas and positions directly on an orthomosaic, which an ordinary photo or a casual drone snapshot does not allow. This makes it the workhorse base layer of drone mapping: site documentation, planning, change tracking, asset location and area calculations all build on it. An orthomosaic answers what a site looks like and where things are; it is a true-to-scale picture, not an elevation model, so it does not by itself tell you how high anything is.

• Orthomosaic: one geometrically corrected, true-to-scale image of a site
• Orthorectification removes tilt, lens and terrain distortion for uniform scale
• Supports direct measurement of distance, area and position
• Answers what the site looks like and where things are, not their height

A Digital Surface Model, or DSM, is an elevation model that records the height of everything the survey saw from above, including the tops of buildings, trees, vehicles, equipment and other objects, as well as the bare ground in open areas. In one sentence: a DSM is the elevation of the visible top surface of the world. Photogrammetry produces a DSM naturally, because a camera images whatever is uppermost, so the model follows rooftops and canopies rather than the earth beneath them.

A DSM is the right product when the objects on the surface are exactly what you care about. It supports line-of-sight and viewshed analysis, flight-path and obstacle assessment, solar and telecom planning, canopy-height work and any task where the built and vegetated environment matters as much as the terrain. If you need to know how tall a structure or a tree is, or what blocks a sight line, the DSM is the model that holds that information, because it deliberately keeps every object in place.

• DSM: elevation of the visible top surface, including buildings and vegetation
• Produced naturally by photogrammetry, which images whatever is uppermost
• Best for line-of-sight, obstacle, solar, telecom and canopy analysis
• Use it when the objects on the surface are what you need to measure

A Digital Terrain Model, or DTM, is an elevation model of the bare earth with buildings, vegetation and other objects removed. In one sentence: a DTM is the elevation of the ground as if everything on top of it were stripped away. Creating one requires either a sensor that can see the ground, such as LiDAR penetrating vegetation, or processing that classifies and removes surface objects from a DSM. The DTM is the foundation for terrain analysis: earthworks and cut-and-fill volumes, drainage and hydrology, slope and contour mapping, and engineering design that must reference true ground level.

DEM, or Digital Elevation Model, is the term that causes the most confusion because usage varies. It is best understood as the umbrella category for any raster grid of elevation values; in much common usage, and particularly in many engineering and GIS contexts, DEM refers specifically to the bare-earth ground model and is used interchangeably with DTM. The safe practice is simple: when a deliverable is specified as a DEM, confirm whether the provider means the bare-earth ground surface or the general family of elevation models, so there is no ambiguity about whether objects are included.

• DTM: elevation of the bare earth with objects and vegetation removed
• Built via ground-seeing sensors like LiDAR or by filtering a DSM
• Basis for earthworks, drainage, slope, contours and engineering design
• DEM is an umbrella term often used to mean the bare-earth model; always confirm intent

A point cloud is the raw three-dimensional measurement from which most other products are derived. It is a dense collection of points, each with an X, Y and Z coordinate and often colour, captured by LiDAR or generated photogrammetrically from overlapping images. The point cloud is the richest geometric representation of a site, but in raw form it is heavy to handle and not directly a map; it is the source material that gets processed into the finished deliverables.

From a single point cloud you can derive several products. Take the highest points and you build a DSM; classify and keep only ground points and you build a DTM; drape the corrected imagery over an elevation surface and you produce an orthomosaic. So the point cloud relates to the orthomosaic and the elevation models as raw geometry relates to finished outputs: the cloud is three-dimensional source data you can model and measure within, while the orthomosaic is a flat, scaled image and the DSM, DTM and DEM are elevation surfaces extracted from that same source. Knowing this chain helps you specify exactly which stage of product you actually need.

• Point cloud: dense 3D points (X, Y, Z, often colour), the raw source data
• DSM derives from the top points; DTM from classified ground points
• Orthomosaic is a flat, scaled image, not a 3D or elevation product
• Specify the exact stage you need: raw cloud, image or elevation surface

Start from the question you must answer and the deliverable follows. Need a measurable map of what the site looks like? Order an orthomosaic. Need the height of buildings, trees or obstacles? You need a DSM. Need true ground level for earthworks, drainage or design? You need a DTM, and if a DEM is specified, confirm it means bare earth. Need full three-dimensional geometry to model, section or measure freely? You need the point cloud. Stating the decision first prevents the common mistake of commissioning a surface model when only a flat image was required, or vice versa.

The deliverable also influences how the data is captured. Bare-earth models under vegetation steer toward LiDAR; rich visual orthomosaics and surface models suit photogrammetry; many programmes need a combination. These outputs feed directly into digital terrain planning, and BotBit's digital terrain mapping supports the workflow from capture to product. Carry your survey sensor on a stabilised payload such as the BotBit payload and gimbal mount, and plan lawful operation, including DGCA and Digital Sky approvals in India, as part of scoping the deliverable.

• Map of the site: orthomosaic; object heights: DSM; bare ground: DTM/DEM
• Full 3D modelling and free measurement: point cloud
• Bare-earth-under-vegetation favours LiDAR; visual products favour photogrammetry
• Define the deliverable first, then the capture method, sensor and compliance