9.3 Active Remote Sensing: LiDAR, RADAR, Synthetic Aperture Radar (SAR) & UAS/Drone Mapping

9.3 Active Remote Sensing: LiDAR, RADAR, Synthetic Aperture Radar (SAR) & UAS/Drone Mapping

Core Principle: Unlike passive optical sensors that rely on reflected sunlight or emitted terrestrial heat, active remote sensing systems emit their own controlled beam of electromagnetic energy and measure the backscattered return signal. LiDAR uses high-frequency laser pulses to generate high-density 3D point clouds of terrain and vegetation canopy; RADAR/SAR uses microwave pulses to penetrate clouds and foliage, measuring surface roughness, dielectric properties, and millimeter-scale ground deformation; and UAS mapping leverages Structure from Motion (SfM) computer vision to construct high-resolution orthomosaics and 3D surface models.

---

1. Active vs. Passive Remote Sensing Systems

The fundamental architectural distinction in remote sensing engineering divides sensors into passive and active systems.

          PASSIVE REMOTE SENSING                        ACTIVE REMOTE SENSING
          
          [ Sun / Solar Radiance ]                 [ Sensor (LiDAR / RADAR) ]
                     \                                       |         ^
                      \  (Incident)           (Transmitted   |         | (Backscattered
                       \                         Pulse)      |         |  Echo)
                        v                                    v         |
                  [ Earth Target ]                     [ Earth Target ]
                        |
                        | (Reflected Energy)
                        v
              [ Passive Sensor ]

• Passive Remote Sensing: Measures naturally occurring radiation reflected from the Sun (optical visible, near-infrared, shortwave infrared) or emitted directly by Earth surface materials as thermal infrared energy. Passive sensors cannot operate in optical bands during darkness, cannot penetrate cloud cover or dense smoke, and are subject to shadowing caused by sun angle variations.
• Active Remote Sensing: Supplies its own source of electromagnetic illumination. The sensor transmits a pulse of known energy, wavelength, polarization, and phase, and subsequently measures the fraction of that pulse scattered back (backscattered) to the receiving antenna or photodetector. Active sensors operate equally well during the day or night, can penetrate adverse atmospheric conditions (such as haze, clouds, and light rain depending on wavelength), and provide direct geometric distance measurements.

---

2. Light Detection and Ranging (LiDAR) Principles and Hardware

LiDAR (Light Detection and Ranging) is an active optical remote sensing technology that measures distances by illuminating targets with rapid, pulsed laser light and recording the time elapsed between pulse emission and return detection.

                             AIRBORNE LIDAR HARDWARE TRIAD
   
                            [ Kinematic GNSS Receiver ]
                            (Determines Aircraft XL, YL, ZL)
                                          |
                                          v
                                [ Inertial Measurement Unit (IMU) ]
                                (Measures Pitch, Roll, Yaw at 200-500 Hz)
                                          |
                                          v
                                [ Scanning Laser Unit ]
                                (High-rate Pulse Emission: 100 kHz - 2 MHz)
                                         /|\
                                        / | \
                                       /  |  \
                                      /   |   \
                                     v    v    v
                              Scan Angle (θ) Limits: ±15° to ±20°

The Hardware Integration Triad

High-accuracy airborne LiDAR positioning is achieved through the integration of three hardware components:

1. The Laser Scanner / Rangefinder: Emits focused laser pulses at exceptionally high pulse repetition frequencies (typically $100\text{ kHz}$ to over $2\text{ MHz}$) and measures round-trip time-of-flight with picosecond-level precision. Most topographic mapping LiDARs operate in the near-infrared spectrum (typically $1{,}064\text{ nm}$). Bathymetric LiDAR systems use green lasers ($532\text{ nm}$), which can penetrate water columns to map seafloors and riverbed bathymetry.
2. Differential Kinematic GNSS Receiver: Mounted directly to the aircraft fuselage, logging multi-constellation satellite carrier-phase signals at high rates ($1-10\text{ Hz}$). Differential post-processing against base station networks establishes the trajectory coordinates $(X_L, Y_L, Z_L)$ of the laser sensor within 2 to 5 centimeters.
3. Inertial Measurement Unit (IMU): Rigidly fixed to the optical scanner head, containing high-precision solid-state or fiber-optic gyroscopes and accelerometers. The IMU measures aircraft orientation—Roll ($\omega$), Pitch ($\phi$), and Yaw/Heading ($\kappa$)—at high frequencies ($200 - 500\text{ Hz}$) to compensate for aircraft vibrations and atmospheric turbulence.

The Time-of-Flight Ranging Equation

The fundamental range distance ($R$) from the sensor to the reflecting surface target is calculated directly from the round-trip travel time ($\Delta t$):

$R = \frac{c \cdot \Delta t}{2}$

Where:

• $R$ = range distance between scanner and ground target (in meters).
• $c$ = speed of light through the atmosphere ($\approx 2.997 \times 10^8\text{ m/s}$).
• $\Delta t$ = total elapsed time from pulse emission to return detection (in seconds).

Combining range ($R$), instantaneous scanner mirror deflection angle ($\theta$), aircraft trajectory $(X_L, Y_L, Z_L)$, and IMU attitude angles transforms every reflected pulse into an absolute three-dimensional coordinate $(X, Y, Z)$ on the terrestrial datum.

Multi-Return Discrete LiDAR

A laser beam diverges slightly as it travels toward the ground, illuminating a circular "footprint" typically $10$ to $30\text{ centimeters}$ in diameter at operational flying altitudes. When this footprint strikes a semi-porous feature—such as a forest canopy—the beam splits, with portions of the pulse reflecting back from different vertical layers:

• First Return: Backscattered from the highest reflective element, such as the top of the tree crown or a building rooftop.
• Intermediate Returns: Reflected from intermediate canopy layers, including sub-canopy tree branches, leaves, understory brush, and transmission lines.
• Last Return: Reflected from the lowest solid boundary encountered—ideally the bare-earth ground surface beneath the trees.

                     MULTI-RETURN DISCRETE LIDAR INTERACTION
   
   Laser Pulse Emission (1064 nm) ---------------------------------------------\
                                                                               \
           +-----------------------+                                            v
           |    Upper Tree Crown   | ====> [ FIRST RETURN: Top of Canopy ]
           +-----------------------+                                            |
                      |                                                         |
                      v                                                         v
           +-----------------------+                                            |
           | Intermediate Branches | ====> [ INTERMEDIATE RETURN: Mid-Canopy ]
           +-----------------------+                                            |
                      |                                                         |
                      v                                                         v
           ~~~~~~~~~~~~~~~~~~~~~~~~~                                            |
           |   Understory Shrub    | ====> [ INTERMEDIATE RETURN: Understory ] |
           ~~~~~~~~~~~~~~~~~~~~~~~~~                                            |
                      |                                                         |
                      v                                                         v
   =======================================                                      |
   [   Bare-Earth Ground Surface         ] ====> [ LAST RETURN: Bare Earth Ground ]

LiDAR Return Intensity

Alongside 3D coordinates, modern LiDAR systems record the intensity of each return—a measure of the backscattered signal amplitude. Intensity is governed by the surface reflectivity of the target at the laser's wavelength ($1{,}064\text{ nm}$), the target roughness, and the angle of incidence. Paved roads, concrete structures, green vegetation, and exposed soils exhibit distinct intensity signatures, allowing intensity rasters to be used similarly to black-and-white near-infrared photography.

ASPRS LAS Classification Standards

The American Society for Photogrammetry and Remote Sensing (ASPRS) defines the standard open binary file format (.las) used globally for storing and exchanging 3D point cloud data. ASPRS standardizes numeric point classification codes:

Elevation Surfaces Derived from Classified LiDAR

• Digital Terrain Model (DTM / Bare-Earth DEM): Generated by interpolating Class 2 (Ground) points exclusively, representing the bare-earth topography with all buildings, towers, and vegetation digitally removed.
• Digital Surface Model (DSM): Generated by interpolating first returns and non-ground features, representing the top surface envelope of the Earth, including rooftops, power lines, and canopy tops.
• Normalized Digital Surface Model (nDSM) / Canopy Height Model (CHM): Calculated via simple map algebra by subtracting the bare-earth DTM from the surface DSM:

$\text{nDSM} = \text{DSM} - \text{DTM}$

The nDSM isolates the true physical height of all above-ground features above the local terrain, making it the primary analytical tool for urban tree canopy studies, building height extraction, and wildfire fuel modeling.

---

3. RADAR and Synthetic Aperture Radar (SAR)

RADAR (Radio Detection and Ranging) systems operate within the microwave region of the electromagnetic spectrum (wavelengths ranging from $1\text{ millimeter}$ to $1\text{ meter}$). Because microwave radiation is several orders of magnitude longer than visible light, it travels through clouds, dense fog, aerosol haze, and moderate rainfall with minimal attenuation, providing an all-weather, day-or-night imaging capability.

Microwave Frequency Bands and Canopy Interaction

Radar backscatter is influenced by the relationship between the radar wavelength and the physical dimensions of surface scattering structures (leaves, twigs, tree branches, boulders), as well as the electrical moisture content (dielectric constant) of the target.

                      MICROWAVE RADAR FREQUENCY BANDS
   
   Wavelength:  ~3 cm (X-band)      ~5.6 cm (C-band)          ~23 cm (L-band)
   Frequency:   (8 - 12 GHz)          (4 - 8 GHz)               (1 - 2 GHz)
   
   Canopy
   Top ------ [ X-Band Scatter ] --> Bounces off small leaves and top twigs
                 \                  
   Mid-Canopy --- v ------------- [ C-Band Scatter ] -------> Scatters within canopy branches
                                     \                         
   Trunk / Ground -------------------- v ------------------ [ L-Band Scatter ]
                                                               Penetrates through foliage
                                                               to trunks & soil surface

Side-Looking Radar Geometry and Geometric Distortions

Airborne and spaceborne imaging radars utilize a Side-Looking Airborne Radar (SLAR) configuration, illuminating terrain at an oblique side angle perpendicular to the flight path (the cross-track range direction). Imaging directly nadir is avoided because symmetrical reflections from the left and right sides of the aircraft would return at identical time delays, causing left-right spatial ambiguity.

Oblique side-looking geometry introduces three distinct terrain-induced geometric distortions:

1. Slant Range vs. Ground Range Distortion: Radar systems measure the direct line-of-sight distance from the antenna to the ground—the slant range. Because slant range varies non-linearly with horizontal ground distance, raw radar imagery appears compressed in the near-range (closest to flight track) relative to the far-range. Preprocessing algorithms must apply trigonometric corrections to convert slant range coordinates into true horizontal ground range.
2. Foreshortening: When a radar beam encounters a terrain slope tilted toward the sensor, the distance between the base and top of the slope is compressed in the slant-range dimension. Foreshortened mountain faces appear compressed and display disproportionately bright backscatter.
3. Layover: An extreme manifestation of foreshortening that occurs when the terrain slope facing the radar is steeper than the incoming radar wavefront (slope angle exceeds the radar depression angle). The radar pulse reaches the mountain summit before it reaches the mountain base. Consequently, the top of the mountain is imaged prior to the bottom, causing the peak to appear laid over toward the sensor in front of the base.
4. Radar Shadow: Slopes tilted steeply away from the radar beam receive no illumination. Because the radar beam cannot illuminate behind the ridgeline, these areas produce no backscattered signal, appearing as solid black regions in the image devoid of information.

                      RADAR GEOMETRIC DISTORTIONS
   
   Radar Antenna [Flight Track]
         \
          \  (Radar Wavefront)
           \ 
            v
            /\ [Layover: Peak imaged before base]
           /  \ 
          /    \ [Radar Shadow: Steep back slope receives no beam]
         /      \ 
   _____/        \____________
     Foreshortening: Front slope compressed

Synthetic Aperture Radar (SAR) Principles

In standard Real Aperture Radar (RAR), spatial resolution in the along-track (azimuth) direction is governed by physical antenna length ($L$) and distance to the target ($R$):

$R_{\text{azimuth}} = \frac{\lambda \cdot R}{L}$

To achieve a fine azimuth resolution of $5\text{ meters}$ from an orbital altitude of $800\text{ kilometers}$ using a $5.6\text{ cm}$ C-band system, a real antenna would have to span over $9\text{ kilometers}$ in physical length—an engineering impossibility.

Synthetic Aperture Radar (SAR) circumvents this physical limitation. As the satellite or aircraft moves along its flight path, the radar transmits hundreds of pulses per second, illuminating a ground target from numerous successive positions. By recording both the amplitude and the phase of the returning echoes, the system uses Doppler frequency shifts to reconstruct the signals as if they were recorded by a single, continuous "synthetic" antenna kilometers long.

The Fundamental Law of SAR: The theoretical azimuth resolution of a focused SAR system is independent of range and flying altitude, and equals approximately half the physical antenna length:

$R_{\text{azimuth}} \approx \frac{L}{2}$

Remarkably, a smaller physical antenna yields a finer spatial resolution in SAR, because a smaller physical antenna has a wider beamwidth, tracking targets over longer flight path distances and synthesizing a longer virtual aperture.

Radar Interferometry (InSAR and DInSAR)

Interferometric Synthetic Aperture Radar (InSAR) measures the phase difference between two coherent SAR acquisitions captured either from slightly different positions (spatial baseline) or at different times (temporal baseline).

• Topographic InSAR: Two radar antennas mounted on a single platform (e.g., the Shuttle Radar Topography Mission - SRTM) or repeat-pass orbits capture the terrain. The phase difference between the two signals directly corresponds to topographic elevation, generating high-accuracy global DEMs.
• Differential InSAR (DInSAR): By subtracting the topographic phase contribution from two SAR scenes acquired over the same geography weeks, months, or years apart, DInSAR isolates surface displacement. DInSAR measures ground deformation with millimeter-level precision, making it an effective tool for monitoring tectonic earthquake slip, volcanic chamber inflation, groundwater extraction subsidence, permafrost thaw cycles, and structural displacement of dams and bridges.

---

4. Unmanned Aerial Systems (UAS / Drone) Mapping Workflows

Small Unmanned Aerial Systems (sUAS), commonly known as drones, have transformed local-scale spatial data capture. Drone photogrammetry pairs high-resolution consumer or industrial digital cameras with computer vision algorithms to produce survey-grade 2D orthomosaics and 3D surface models.

Structure from Motion (SfM) Photogrammetry

Traditional aerial photogrammetry requires metric cameras with precisely calibrated focal lengths and rigorous flight lines with pre-calculated camera station geometries. In contrast, drone mapping relies on Structure from Motion (SfM)—a computer vision workflow that simultaneously resolves camera calibration parameters, exterior orientation, and three-dimensional scene geometry from overlapping, multi-angle photographs.

                         STRUCTURE FROM MOTION (SfM) PIPELINE
   
   [ Overlapping Drone Images ] (75-80% Front-lap, 65-75% Sidelap)
                 |
                 v
   [ Feature Identification: SIFT Algorithm detects invariant tie points ]
                 |
                 v
   [ Feature Matching: Matches identical keypoints across multiple photo pairs ]
                 |
                 v
   [ Bundle Block Adjustment (BBA): Solves camera poses & creates Sparse Point Cloud ]
                 |
                 v
   [ Multi-View Stereo (MVS): Dense image matching creates Dense Point Cloud ]
                 |
                 v
   [ 3D Triangular Mesh / DEM Generation ] ====> [ Orthorectified Orthomosaic ]

1. Keypoint Detection: Feature identification algorithms (such as the Scale-Invariant Feature Transform - SIFT) detect thousands of distinctive features (corners, texture edges, high-contrast patterns) across every image that remain invariant to scale, rotation, and illumination changes.
2. Tie-Point Matching: Keypoints are matched across overlapping photos, tracking points across 5 to 20 overlapping frames.
3. Sparse Bundle Block Adjustment: Solves for unknown camera positions, orientations, focal length, and lens distortions simultaneously, producing an initial sparse 3D point cloud.
4. Multi-View Stereo (MVS) Dense Matching: Uses pixel-by-pixel matching across the calibrated photo block to densify the point cloud into hundreds of millions of 3D points.
5. Product Generation: Generates high-density DSMs, digital terrain models (after running ground filtering routines), and orthomosaics with sub-centimeter Ground Sample Distances (GSD).

UAS Flight Planning and Overlap Requirements

Because small drones fly at low altitudes ($50 - 120\text{ meters}$ above ground level) over complex 3D structures, perspective changes between adjacent exposures are far more pronounced than in traditional high-altitude mapping. Consequently, SfM photogrammetry requires significantly higher overlap thresholds:

• Forward Overlap (Front-lap): $75%$ to $80%$ minimum along the flight path.
• Sidelap (Cross-lap): $65%$ to $75%$ minimum between parallel flight passes (increased to $80%$ in dense forest canopies or complex architectural environments).

Sensor Mechanics: Rolling Shutter vs. Global Shutter

A critical technical issue in UAS mapping is the distinction between rolling shutter and global shutter image capture mechanisms:

                ROLLING SHUTTER vs. GLOBAL SHUTTER CAPTURE
   
   ROLLING SHUTTER (Electronic Scan):           GLOBAL SHUTTER (Instantaneous Capture):
   Line 1 exposed at t = 0.00 ms               All rows (1 to N) exposed simultaneously
   Line 2 exposed at t = 0.05 ms               at the exact same instant (t = 0.00 ms).
   ...
   Line N exposed at t = 15.00 ms
   
   * Result: Moving drone distorts image.       * Result: Zero geometric distortion.
   * Structures warp, shear, and "jell-o".      * Meets survey-grade mapping standards.

• Rolling Shutter (Electronic CMOS): Most consumer drones utilize electronic rolling shutters that expose the sensor array row-by-row over a duration of 10 to 30 milliseconds. If the drone is moving at high speed ($8 - 15\text{ m/s}$) during image exposure, the top of the image is captured at a different physical position in space than the bottom. This differential motion introduces geometric shear, bending, and "jell-o" distortions that violate collinearity assumptions and degrade the spatial accuracy of the resulting SfM point cloud.
• Global Shutter (Mechanical or Advanced Electronic): Exposes every single pixel across the sensor array simultaneously in a single instant. Global shutters eliminate motion-induced geometric distortions, making them necessary for high-accuracy engineering surveys and survey-grade cadastral mapping.

Direct Georeferencing (RTK / PPK) vs. Ground Control Points (GCPs)

• Ground Control Points (GCPs): High-contrast targets laid across the site and surveyed using geodetic GNSS. Traditional SfM workflows require 5 to 10 GCPs evenly distributed across the project boundary, plus additional GCPs in the center and at elevation extremes, to anchor the bundle block adjustment and prevent "doming" or "bowling" deformations.
• Real-Time Kinematic (RTK) / Post-Processed Kinematic (PPK) Drones: Onboard dual-frequency GNSS receivers receive real-time corrections from a base station (RTK) or post-process carrier-phase logs against base files (PPK). Direct georeferencing calculates the 3D position of each exposure center to within $1$ to $3\text{ centimeters}$.
• Check Points (Mandatory Quality Control): Even when operating RTK/PPK drones with high positioning precision, photogrammetrists must place independent Check Points across the survey area. Check points are held out of the bundle adjustment to verify absolute horizontal and vertical accuracy (RMSE) in compliance with ASPRS accuracy standards.

---

5. Common GISP Exam Traps & Pitfalls

[!CAUTION] Exam Trap 9.3.1: Confusing ASPRS Classification Codes (Class 2 vs. Class 6). Questions testing LAS point cloud classification frequently target the numeric codes. Remember that Class 2 is Ground (bare-earth) and Class 6 is Building. A common error is mixing up Class 2 with Class 1 (Unassigned) or Class 8 (Model Key-point). If a question asks which points must be extracted to generate a bare-earth Digital Terrain Model, the answer is strictly Class 2.

[!CAUTION] Exam Trap 9.3.2: Assuming SAR Resolution Degrades with Distance (Orbital Altitude). In optical systems and real aperture radar, spatial resolution degrades as distance to the target increases ($R_{\text{az}} \propto R$). Candidates often assume this rule applies to Synthetic Aperture Radar. It does not. In SAR, the virtual aperture expands with range, causing the along-track (azimuth) resolution to be completely independent of range and altitude, depending solely on the physical antenna dimension ($R_{\text{az}} \approx L/2$).

[!CAUTION] Exam Trap 9.3.3: Assuming RTK Drones Eliminate the Need for Check Points. A widespread industry mistake is believing that flying an RTK or PPK drone eliminates the need for any ground survey measurements. While direct georeferencing can reduce the number of Ground Control Points (GCPs) required for the bundle adjustment, independent checkpoints are needed when the project claims independently tested horizontal or vertical accuracy; their number and distribution follow the applicable specification.

[!CAUTION] Exam Trap 9.3.4: Overlooking Rolling Shutter Distortions in Drone Photogrammetry. Drone imagery collected with low-cost consumer CMOS sensors that use rolling shutters experiences non-linear geometric warping when flying at operational survey speeds. For high-precision, survey-grade photogrammetric deliverables, specifications should require either a global shutter or advanced mathematical rolling-shutter compensation algorithms embedded in the SfM bundle adjustment.