Stereoscopy, DEM and Orthophoto - Cheat Sheet
Your final 30-minute reference for stereoscopy principles, base-height ratio, DEM types, and orthophoto production. Master the formulas, definitions, and exam pitfalls.
Stereoscopy & Stereo Overlap
Formulas
B/H = air-base ÷ flying height
Meaning
B = successive exposure separation (m); H = flight altitude above datum (m); ratio dimensionless
Watch out
Do NOT confuse air-base (distance between camera positions) with baseline (distance between principal points on stereo pair). Air-base is measured in object space, not image space.
When to use
Whenever you need to evaluate height-measurement strength, vertical exaggeration, or stereo model geometry
Forward overlap % = (1 − ground_advance ÷ photo_ground_side) × 100
Meaning
Ground advance = ground distance covered per exposure; photo ground side = photo format × scale denominator
Watch out
Overlap refers to FORWARD overlap (along flight direction); SIDELAP (~25–30%) is overlap between adjacent flight lines. Do not mix these up.
When to use
Design of flight mission; calculation of air-base from required overlap
Air-base = (1 − overlap_fraction) × ground_side_length
Meaning
Overlap fraction (decimal, e.g., 0.60 for 60%); ground side = photo width at ground scale
Watch out
This assumes constant altitude and nadir photography. Terrain relief and aircraft drift may alter actual spacing.
When to use
Find spacing between successive exposures in a flight line given desired overlap
Common values
Value
60%
Symbol
O_f
Quantity
Typical forward overlap (3-D stereo)
Value
25–30%
Symbol
O_s
Quantity
Typical sidelap between flight lines
Value
~920 m
Symbol
B
Quantity
Common air-base for 1:10,000 photo (230 mm format)
Value
0.40–1.00
Symbol
B/H
Quantity
Base-height ratio range (photogrammetry)
Important facts
- Stereoscopy requires two images taken from different viewpoints separated by air-base B.
- The base-height ratio B/H is THE key parameter controlling height-determination sensitivity: larger B/H → stronger height model → less vertical exaggeration.
- Typical forward overlap for stereo photogrammetry is 60%; this leaves 40% new terrain per photo.
- Sidelap (~25–30%) between flight lines prevents gaps and allows bundle adjustment tie-points.
- Vertical exaggeration in a stereo model ≈ H/(B × scale), so small B/H ratios exaggerate terrain relief.
- Perfect stereoscopy requires the two images to have the same scale, similar tilt, and sufficient contrast.
- Relief displacement in raw photos disappears when viewed stereoscopically because both photos contain matching displacements.
- The distance between principal points in image space (baseline) relates to air-base B through scale: baseline = B × scale.
Key definitions
Stereoscopy
Example
Viewing overlapped aerial photos through a stereoscope creates an artificial 3-D model from which parallax readings yield terrain heights.
Fusion of two slightly offset images by the human visual system to perceive three-dimensional relief; basis for photogrammetric height measurement.
Stereo Pair
Example
Photos exposed at 900 m separation at 1500 m altitude form a stereo pair with B/H = 0.60.
Two overlapping photos (typically ~60% forward overlap) taken from successive exposures, one stereo left, one stereo right.
Base-Height Ratio (B/H)
Example
B/H = 0.60 gives moderate vertical exaggeration; B/H = 1.0 gives 1:1 scaling (exaggeration = 1).
Ratio of air-base to flying height; controls vertical exaggeration and geometric strength of stereo model.
Vertical Exaggeration
Example
Small B/H values (e.g., 0.40) exaggerate terrain; large B/H (e.g., 1.0) minimize exaggeration.
Factor by which vertical distances appear enlarged in a stereo model; equals (H/B) × (photo scale) for typical viewing geometry.
Forward Overlap
Example
60% forward overlap on 230 mm format at 1:10,000 scale gives air-base ~920 m.
Percentage of successive photos that overlap along the flight direction; typically 60% for 3-D photogrammetry.
Sidelap
Example
Sidelap ensures no gaps between strips and allows tie-point matching between neighboring lines.
Overlap between adjacent flight lines; typically 25–30% to ensure continuous stereo coverage.
Diagrams to know
- Stereo pair geometry: two cameras separated by air-base B at height H above datum.
- Relief displacement diagram: point at height h above datum projects further from principal point than point at datum level.
- Vertical exaggeration in stereo model as a function of B/H.
Digital Elevation Models (DEM)
Common values
Value
1–2 m
Symbol
dx, dy
Quantity
Typical DEM grid cell size (urban)
Value
5–10 m
Symbol
dx, dy
Quantity
Typical DEM grid cell size (regional mapping)
Value
±0.05–0.30 m (RMSE)
Symbol
σ_z
Quantity
Vertical accuracy, LiDAR DEM
Value
±0.10–1.00 m (RMSE)
Symbol
σ_z
Quantity
Vertical accuracy, photogrammetric DEM
Value
±2–10 m (RMSE)
Symbol
σ_z
Quantity
InSAR DEM vertical accuracy
Important facts
- A DEM is essential for orthophoto production, contour generation, viewshed/visibility analysis, drainage pattern extraction, and volume computation.
- DEM grid cell size determines vertical and horizontal precision: smaller cells → finer detail but larger file size.
- DEM vs DSM: DEM is bare earth; DSM includes objects. Choose DEM for terrain analysis, DSM for urban planning with building heights.
- TIN models use fewer points than grids for same accuracy by adaptive refinement; easier to incorporate breaklines (ridges, cliffs, stream channels).
- DEM can be produced by three primary methods: (1) photogrammetric stereo image matching, (2) LiDAR scanning, (3) radar interferometry (InSAR).
- Photogrammetric DEM requires sharp image contrast and sufficient stereo overlap; fails in homogeneous areas (water, snow, dense vegetation).
- LiDAR DEM is fast, accurate, all-weather, and penetrates some vegetation; expensive and requires specialist equipment.
- InSAR (Interferometric Synthetic Aperture Radar) DEM works in cloud cover and at night; coarser resolution (10–30 m) than optical methods.
- Vertical accuracy of a DEM depends on flying height, image resolution, stereo base, and matching algorithm precision: typically 0.1–1 m for photogrammetry, ±10–30 cm for LiDAR.
- DEM voids (missing data) occur over water, shadow, or areas with no texture; must be filled by interpolation or secondary data.
Key definitions
Digital Elevation Model (DEM)
Example
1 m × 1 m grid DEM produced from LiDAR or photogrammetric stereo covering 100 km² area.
Gridded or triangulated dataset of terrain elevations at regular horizontal intervals; bare-earth surface without vegetation or buildings.
Digital Surface Model (DSM)
Example
Urban DSM includes building roofs; forest DSM includes tree-crown heights; DEM removes both for bare ground.
Gridded elevation dataset including terrain, vegetation canopy, buildings, and other objects; represents top-of-canopy/structure elevation.
Digital Terrain Model (DTM)
Example
DTM used for drainage analysis, earthwork computation, visibility analysis.
Synonym for DEM; emphasizes bare-earth surface representation after filtering of non-terrain features.
TIN (Triangulated Irregular Network)
Example
TIN preferred for irregular terrain; allows variable point density; easier to edit breaklines (ridges, stream channels).
Vector-based elevation model using connected triangles with vertices at measured height points; adaptive mesh refines in high-relief areas.
Grid DEM
Example
10 m grid DEM: each pixel represents 10 m × 10 m ground area with single elevation.
Raster elevation model with regular square or rectangular cells; each cell holds one elevation value.
Orthophotography (Orthophoto)
Example
1:10,000 orthophoto of Metro Manila with ±1 m planimetric accuracy; can be measured like a map.
Aerial photograph geometrically corrected to remove relief displacement and tilt distortion using a DEM, resulting in uniform scale suitable for measurement.
Orthomosaic
Example
10,000 km² orthomosaic of Mindanao at 2 m/pixel resolution.
Seamless mosaic of multiple orthophotos merged and blended into a single, continuous image product.
Diagrams to know
- Cross-section showing DEM (bare earth) vs DSM (with buildings and vegetation).
- Grid DEM cell structure with elevation values in each cell.
- TIN triangle mesh with vertices at varying heights and adaptive density.
- Flowchart: DEM generation methods (photogrammetry, LiDAR, InSAR).
DEM Production Methods
Common values
Value
1–4 points/m²
Symbol
ρ
Quantity
LiDAR point density (airborne)
Value
0.5–2.0 m
Symbol
Δ
Quantity
Photogrammetric point spacing (automated matching)
Value
30 m
Symbol
cell size
Quantity
SRTM InSAR DEM resolution (global)
Value
750 m
Symbol
H
Quantity
Typical flight altitude (photogrammetry, 1:5000 scale)
Important facts
- Photogrammetric DEM: requires overlapping stereo images, good image contrast, and tied ground control points; can fail in featureless areas.
- LiDAR DEM: direct range measurement, penetrates thin vegetation, all-weather, no GCP required for relative heights; very expensive.
- InSAR DEM: global coverage possible (SRTM 30 m), coherent for vegetated and bare areas, no optical imagery needed; phase unwrapping challenging over mountains.
- Photogrammetry best for detailed urban/industrial mapping; LiDAR best for vegetation mapping and terrain under forest canopy; InSAR best for rapid global DEMs.
- Hybrid DEMs combine photogrammetric and LiDAR data for accuracy and completeness.
- DEM quality controlled by ground control point density, image resolution, baseline/altitude ratio (photogrammetry), and laser pulse density (LiDAR).
Key definitions
Photogrammetric Image Matching
Example
Correlation-based image matching finds pixel-level correspondence between left and right images; outputs x,y,z for each matched point.
Automated or manual identification of corresponding points in stereo images to compute 3-D coordinates and generate DEM by parallax.
LiDAR (Light Detection and Ranging)
Example
Airborne LiDAR at 1000 m altitude yields 1–4 points/m² density; ground-based LiDAR (TLS) yields millions of points/m².
Active remote sensing using laser pulses to directly measure range to terrain; produces point cloud with x,y,z coordinates.
InSAR (Interferometric Synthetic Aperture Radar)
Example
Spaceborne InSAR (ALOS-2, Sentinel-1) produces 30 m resolution global DEM; applied to SRTM (Shuttle Radar Topography Mission).
Radar technique using phase difference between two synthetic aperture radar (SAR) images to compute elevation; works day/night and through clouds.
Point Cloud
Example
LiDAR point cloud: 50 million points covering 100 km² → gridded to 1 m DEM.
Unstructured set of x,y,z coordinates; output of LiDAR scanning or photogrammetric bundle adjustment; basis for gridding into DEM.
Diagrams to know
- Photogrammetric stereo: two images → image matching → parallax → 3-D point cloud → DEM grid.
- LiDAR system: laser pulse transmission → reflection from terrain/canopy → time delay → range → x,y,z.
- InSAR geometry: two radar passes → interferogram → phase unwrapping → elevation.
Orthophoto Production & Orthorectification
Formulas
Scale variation in raw photo ∝ relief displacement + tilt distortion
Meaning
Raw photo scale changes with elevation and camera tilt; no uniform scale possible without correction
Watch out
Do NOT measure distances on a raw aerial photo. The scale varies across the image. You must use an orthophoto (corrected with DEM) or rectified orthophoto.
When to use
Explain why raw aerial photos cannot be used for measurement; why orthophotos are necessary
Orthorectification: resample using DEM to place each pixel at correct planimetric (x, y) position
Meaning
DEM provides elevation z for each ground point; backward ray tracing from ortho grid through DEM to original image recovers source pixel.
Watch out
Orthorectification REQUIRES an accurate DEM. Without DEM, you cannot remove relief displacement. The quality of orthophoto depends on DEM accuracy.
When to use
Convert raw aerial photograph to orthophoto; remove relief displacement and tilt distortion in one step
Common values
Value
±0.5–2.0 m (at 90% confidence)
Symbol
σ_xy
Quantity
Typical orthophoto planimetric accuracy
Value
0.25–2.0 m/pixel
Symbol
GSD
Quantity
Common orthophoto ground resolution
Value
WGS84 / UTM Zone 51N–53N (PH) or PRS92 / PPCS
Symbol
CS
Quantity
Typical orthophoto vertical datum
Value
~0.5 mm on photo (~5 m on ground at 1:10,000 scale)
Symbol
d_r
Quantity
Relief displacement per 100 m elevation (at 1000 m altitude, 230 mm photo)
Important facts
- Raw aerial photo has varying scale due to relief and tilt; cannot be used for direct measurement like a map.
- Orthophoto is the photo resampled using DEM to remove relief displacement and tilt → uniform scale → measurable like a map.
- Orthophoto planimetric accuracy ≈ image ground resolution + DEM vertical error / slope angle; typically ±0.5–2.0 m.
- Orthomosaic tiles multiple orthophotos into seamless product; requires color balancing and edge-matching between tiles.
- Orthorectification workflow: raw image + DEM + camera model → backward ray tracing → resampled orthophoto.
- Orthophoto retains photographic detail (buildings, roads, vegetation) while providing map-accurate geometry.
- Orthophoto resolution (e.g., 1 m/pixel, 0.5 m/pixel) set by original image resolution; DEM resolution should match or exceed image resolution.
- Common orthophoto applications: urban planning, land use mapping, change detection, infrastructure inventory, emergency response.
- Quality control: check for edge mismatch in mosaics, color discontinuities, and positional accuracy against known ground survey points.
- Philippine PRS92 / PPCS or WGS84 / UTM commonly used coordinate systems for orthophoto georeferencing per RA 4374.
Key definitions
Relief Displacement
Example
Mountain peak 1000 m above sea level displaces ~5 cm on a 1:10,000 photo from 1200 m altitude; tree top displaces ~2 cm.
Radial shift of image point from principal point due to elevation; points higher than datum appear further from principal point.
Tilt Distortion
Example
Camera tilted 10° forward → near end of photo larger scale, far end smaller scale.
Scale and shape distortion in image caused by non-vertical camera tilt; image scale increases in direction of tilt.
Orthophoto (Orthographic Projection)
Example
1:5000 orthophoto of Quezon City with ±0.5 m planimetric accuracy; scale is constant across entire image.
Image geometrically corrected using DEM so every point is at correct planimetric position with uniform scale; can be measured like a map.
Orthomosaic
Example
10,000 ha orthomosaic of Batangas province at 1 m/pixel; 100+ individual orthophoto tiles blended.
Multiple orthophotos seamlessly merged, color-balanced, and clipped to a common boundary; single seamless product.
Backward Ray Tracing (Resampling)
Example
Orthophoto pixel at (100 m, 200 m, elevation 50 m) → ray through camera model → locate source pixel in raw left image → resample intensity.
Algorithm for orthorectification: for each orthophoto grid cell, trace ray back through DEM and camera model to find source pixel in raw image.
Pansharpening
Example
Sentinel-2 10 m multispectral + 1 m panchromatic orthophoto → 1 m colored orthophoto.
Fusion technique merging low-resolution multispectral imagery with high-resolution panchromatic orthophoto to produce sharp, colored orthophoto.
Diagrams to know
- Relief displacement diagram: point at elevation h projects further from principal point than point at datum; radial shift ∝ h.
- Tilt distortion diagram: tilted camera produces varying scale across image; scale larger at near end, smaller at far end.
- Orthorectification process: raw image + DEM + camera model → backward ray tracing → orthophoto.
- Orthomosaic assembly: multiple orthophotos → color balance → edge blend → clipped mosaic.
- Orthophoto vs raw photo comparison: scale uniformity, relief-distortion removal, measurability.
Applications & Uses of DEM and Orthophoto
Common values
Value
5–10 m
Symbol
Δz
Quantity
Typical contour interval (mountainous)
Value
1–2 m
Symbol
Δz
Quantity
Typical contour interval (gentle terrain)
Value
>0.5%
Symbol
S_min
Quantity
Minimum slope for contour definition
Value
5–10 m
Symbol
dx
Quantity
Typical DEM cell size for slope mapping
Important facts
- DEM enables automated generation of contours; manual methods are slow and prone to error.
- Contour interval selection affects map readability: steep terrain needs smaller intervals (2–5 m); gentle terrain can use larger (10–20 m).
- Volume computation using DEM + design surface is faster and more uniform than manual cross-section methods.
- Viewshed analysis essential for site selection: communication towers, surveillance cameras, hazard warnings.
- Drainage extraction from DEM allows large-area watershed delineation; manual digitization infeasible for 100,000+ ha.
- Orthophoto + DEM enables rapid land use mapping, change detection (before/after), and feature extraction.
- Orthophoto is ideal base map for engineering design (roads, pipelines, buildings) because scale is uniform and accurate.
- Slope maps from DEM used for site suitability analysis: agriculture, landslide risk, road design.
- Aspect maps guide site selection for agriculture (sun exposure), solar panels, wind farms.
- Flood modeling with DEM + rainfall data essential for disaster risk reduction and evacuation planning per RA 4374 (Unified Building Code).
Key definitions
Contour Generation
Example
10 m contour interval extracted from 1 m DEM; visualizes terrain slope and morphology.
Extraction of elevation contour lines (constant elevation) from DEM at regular intervals; used for topographic maps.
Volume Computation (Cut/Fill, Earthwork)
Example
Airport runway expansion: compute volume of soil to excavate from existing terrain to achieve design grade.
Calculation of volume of earth to be excavated (cut) or placed (fill) by differencing original DEM and design surface DEM.
Viewshed (Visibility) Analysis
Example
Communication tower on hilltop; compute which areas have line-of-sight to tower using DEM.
Determination of which terrain areas are visible from a given observer point using DEM; accounts for terrain occlusion.
Drainage Analysis
Example
Automatic river centerline extraction and watershed delineation from DEM for hydrology.
Extraction of stream networks, flow direction, and catchment boundaries from DEM using flow algorithms.
Slope/Aspect Maps
Example
Slope map for landslide susceptibility zoning; aspect map for solar radiation estimation.
Derived raster map showing slope angle (steepness) and aspect (compass direction) at each DEM cell.
Flood Inundation Modeling
Example
Flood risk map for Laguna Province: combine 100-year design rainfall with DEM to map inundation depth and extent.
Prediction of flood extent using DEM, rainfall, and hydrological flow models; critical for hazard mapping.
Diagrams to know
- Contour line extraction algorithm: DEM grid → interpolate elevation along contour line → connect points.
- Viewshed diagram: observer point → terrain between observer and target → line-of-sight blocked/unblocked by terrain elevation.
- Flow direction algorithm on DEM: steepest descent routing from each cell determines stream direction.
- Cut/fill computation: original DEM – design DEM = cut (positive) / fill (negative) volume.
- Slope and aspect calculation from DEM: use neighboring cell elevations to compute gradient vector.
- Flood inundation: DEM + water surface elevation → inundation depth raster.
Practical Calculations & Board Exam Examples
Important facts
- Always state assumptions (e.g., nadir photography, constant altitude, datum reference) before calculating.
- Check units: air-base in meters, flying height in meters → B/H is dimensionless.
- For ground distance, use: ground side = photo format (mm) × scale denominator (e.g., 230 mm × 10,000 = 2,300 m).
- Overlap percentage: (1 − new_ground / photo_ground_side) × 100%; verify by summing image coverage over flight line.
- DEM vertical accuracy depends on matching precision in stereo images; texture-poor areas (water, snow) prone to large errors.
- Orthophoto planimetric accuracy is limited by image resolution (GSD) and DEM accuracy; expect ±1–2 × GSD under ideal conditions.
- When comparing orthophoto and raw photo: orthophoto scale is uniform; raw photo scale varies by ±5–20% over typical relief.
Must remember
- Base-height ratio B/H is the KEY parameter: larger B/H gives weaker height model (less exaggeration); smaller B/H gives stronger height model (more exaggeration). Typical range: 0.40–1.0.
- Stereoscopy requires overlapping images (~60% forward overlap typical). Two photos taken from different viewpoints (separated by air-base B) fuse into 3-D model when viewed stereoscopically.
- DEM is bare-earth elevations only; DSM includes vegetation, buildings, and objects. For terrain analysis use DEM; for urban planning use DSM.
- Raw aerial photo has VARYING SCALE due to relief displacement and tilt distortion — it CANNOT be measured like a map. Orthophoto is the same photo resampled using a DEM to remove distortions and achieve UNIFORM SCALE.
- Orthophoto production requires THREE inputs: (1) raw aerial image, (2) DEM (digital elevation model), (3) camera interior/exterior orientation. Without DEM, you cannot create orthophoto.
- Three ways to produce DEM: (1) Photogrammetric stereo image matching, (2) LiDAR laser scanning, (3) Radar interferometry (InSAR). Each has trade-offs in accuracy, cost, weather sensitivity.
- Relief displacement formula concept: Points at higher elevation project further from principal point (radially outward) than points at datum level. Magnitude depends on height above datum and flying height.
- Orthophoto planimetric accuracy ≈ ground resolution + DEM error; typically ±0.5–2.0 m. DEM accuracy is the limiting factor for accurate orthophotos.
- Sidelap (~25–30%) is overlap between ADJACENT FLIGHT LINES; forward overlap (~60%) is overlap along the flight direction. Both essential: forward overlap for 3-D stereo; sidelap for seamless coverage.
- When designing a photogrammetric mission: air-base = (1 − overlap_fraction) × ground_side_length. Example: 60% overlap on 2300 m ground side → air-base = 0.40 × 2300 = 920 m.
Last minute tips
- EXAM TIP: If asked "Why orthophoto over raw photo?" — answer: Raw photo has varying scale due to relief and tilt (cannot measure); orthophoto has uniform scale (can measure like map).
- EXAM TIP: DEM vs DSM confusion is COMMON. Remember: DEM = bare earth (terrain only); DSM = surface (with buildings/trees). Ask: Is vegetation/building height included? If YES → DSM. If NO → DEM.
- EXAM TIP: For B/H calculations, always check units. Air-base and flying height must be in SAME units (both meters). Result B/H is dimensionless.
- EXAM TIP: Orthophoto requires a DEM. If DEM is poor quality or has voids (missing data), orthophoto will have errors. This is a common pitfall in practice.
- EXAM TIP: Relief displacement increases with elevation (h) and decreases with flying height (H). High mountain + low altitude → large displacement. This is critical for understanding why photogrammetry fails over tall mountains without proper DEM correction.
Comparison tables
DEM vs DSM vs DTM
| Feature | DEM (Digital Elevation Model) | DSM (Digital Surface Model) | DTM (Digital Terrain Model) |
|---|---|---|---|
| Definition | Bare-earth ground surface elevations | Top-of-canopy, building, object elevations | Synonym for DEM; bare earth only |
| Includes vegetation/buildings? | No (removed by filtering) | Yes (includes all objects) | No (bare earth) |
| Production method | Photogrammetry, LiDAR, InSAR (with filtering) | Photogrammetry, LiDAR (direct) | LiDAR (with ground classification) or manual |
| Use case | Terrain analysis, drainage, earthwork, contours | Urban planning, tree heights, building mapping | Hydrological modeling, slope stability, geomorphology |
| Example application | Airport earthwork design, flood plain mapping | City 3-D building model, forestry inventory | Watershed delineation, landslide hazard zonation |
Raw Photo vs Orthophoto
| Property | Raw Aerial Photo | Orthophoto |
|---|---|---|
| Scale uniformity | Varying (relief + tilt distortion) | Uniform across entire image |
| Relief displacement | Present (radial from principal point) | Removed using DEM |
| Tilt distortion | Present if camera not vertical | Removed in orthorectification |
| Measurability | Cannot measure distances directly (scale varies) | Can measure like a map (uniform scale) |
| Georeferencing | Requires tie-points; scale changes with relief | Georeferenced to map projection; accurate planimetry |
| Use in GIS | Background reference only; distortion too large for analysis | Base layer for mapping, measurement, analysis |
| Production requirements | Single photo, basic georeferencing | Photo + DEM + camera model + orthorectification software |
| Accuracy (planimetric) | ±5–50 m depending on relief (unreliable) | ±0.5–2.0 m depending on resolution and DEM quality |
Grid DEM vs TIN DEM
| Aspect | Grid (Raster) DEM | TIN (Vector) DEM |
|---|---|---|
| Structure | Regular square/rectangular cells; one elevation per cell | Triangulated mesh; vertices at measured points |
| Data density | Uniform across area; wasteful in flat zones | Adaptive; denser where terrain complex |
| Storage efficiency | Large file size for detailed DEMs | Smaller file size; fewer points for same accuracy |
| Breaklines | Hard to represent (ridges, cliffs averaged) | Easy to incorporate and enforce (forced edges) |
| Interpolation | Grid-based (bilinear, bicubic) | Within triangle: linear or smooth |
| Computation speed | Fast for grid operations (slope, aspect, contours) | Slower for raster operations; requires conversion to grid |
| Editing | Cell-by-cell editing; tedious | Vertex editing; easy to add/remove points or adjust breaklines |
| Best for | Rapid processing, slope/aspect maps, raster GIS workflows | Accurate representation, breakline preservation, manual editing |
DEM Production Methods Comparison
| Method | Photogrammetry | LiDAR | InSAR |
|---|---|---|---|
| Principle | Stereo image matching via parallax | Laser range measurement | Radar interferometry (phase difference) |
| Weather dependency | Requires clear sky, good lighting | All-weather capability | Day/night, cloud-penetrating |
| Point density | Variable (0.5–4 pts/m² for good overlap) | Controllable (1–20 pts/m² typical) | ~1 point per 30×30 m cell (SRTM) |
| Vertical accuracy | ±0.10–1.0 m (RMSE) | ±0.05–0.30 m (RMSE) | ±2–10 m (RMSE) |
| Vegetation penetration | Top-of-canopy only; no ground beneath forest | Partial (multiple returns can reach ground) | Limited; covers canopy top |
| Cost | Moderate (depends on image source, GCP) | High (equipment, processing) | Low (free global SRTM, Sentinel-1 data) |
| Turnaround time | Medium (image acquisition → processing weeks) | Fast for small areas; slower for large areas | Very fast (existing global data; new data weekly) |
| Global coverage | Partial (commercial, varies by region) | Partial (not global) | Complete (SRTM, Copernicus DEM) |
| Best application | Detailed urban/regional DEMs with good accuracy | High-accuracy DEMs for engineering, forest mapping | Rapid global DEMs, tectonic studies, large-area terrain |
Diagrams
Stereoscopy & Stereo Pair Concepts
Hierarchical overview of stereoscopy principles: stereo pair geometry, base-height ratio, overlap percentages, vertical exaggeration, and sidelap. Shows how these concepts interconnect in stereo photogrammetry.
DEM Production Workflow Comparison
Three parallel paths for DEM production: photogrammetric stereo matching, LiDAR scanning, and InSAR interferometry. All converge to point cloud → gridding → DEM → quality check → orthophoto production.
Orthophoto Production from Raw Photo
Complete workflow for converting raw aerial photo to orthophoto and potentially to orthomosaic. Shows four required inputs (photo, DEM, camera model, GCP) and key processing steps.
Image Quality and Feature State in Stereoscopy
State transition diagram showing how raw photo distortions are introduced (relief, tilt) and removed through DEM-based orthorectification to produce a measurable orthophoto.
Evolution of Elevation Data Technologies
Timeline showing technological evolution from analog stereoplotters to modern spaceborne LiDAR, illustrating the progression of DEM production methods over six decades.
DEM Production Cost Distribution (Typical Project)
Typical cost allocation for a photogrammetric DEM project: imagery acquisition (35%), ground control survey (20%), processing software (15%), QC (15%), contingency (15%). Shows that GCP fieldwork is significant cost driver.