Microwave and LiDAR remote sensing

In short

Microwave remote sensing senses the Earth with wavelengths of about 1 mm to 1 m, and LiDAR with laser pulses in the visible to near-infrared, both usually as active systems that supply their own energy and measure the echo. Because they are active and (for microwaves) long-wavelength, they work at night, see through cloud, haze and smoke, and measure range directly — so they give all-weather imaging, millimetre-level ground deformation and bare-earth terrain under forest canopy, none of which optical imaging can do. The topic is examined heavily in GATE Geomatics (GE) and university remote-sensing papers: radar geometry, resolution expressions, backscatter and polarisation, interferometry, and lidar point-cloud processing. In India it underpins RISAT-1A/2B series SAR, EOS-04, the NASA-ISRO NISAR mission, flood and crop mapping at NRSC/MOSDAC, and the bulk of commercial aerial-lidar corridor and city mapping work (metros, highways, power lines, river basins).

Explained

Microwave fundamentals: bands, geometry and terms

• Active vs passive microwave: radar/SAR and scatterometers/altimeters transmit and receive (active); microwave radiometers only receive natural emission (passive) and give brightness temperature for soil moisture, sea-surface wind and atmospheric sounding.
• Bands (approximate wavelengths): Ka 0.8–1.1 cm, Ku 1.7–2.4 cm, X 2.4–3.8 cm, C 3.8–7.5 cm, S 7.5–15 cm, L 15–30 cm, P 30–100 cm. Longer wavelength = deeper penetration into canopy, soil and dry sand. RISAT-1/1A and Sentinel-1 are C-band; RISAT-2/2B series X-band; NISAR carries L-band (ISRO contributes S-band).
• Side-looking geometry is compulsory: a nadir-looking radar would receive equal-range echoes from both sides of the track and could not separate them. Hence the terms nadir, look angle, incidence angle, local incidence angle, near range, far range, swath.
• Slant range vs ground range: the radar measures slant range R = c·t/2; ground range = R·sin θ (θ = incidence angle). Imagery must be projected to ground range, which removes the near-range compression of the slant-range presentation.
• Depression angle = 90° − look angle; the incidence angle is measured from the local surface normal, and over sloping ground the local incidence angle controls the brightness.
• Pulse parameters: pulse width τ, pulse repetition frequency PRF, bandwidth B, antenna length L and width W, beamwidth β ≈ λ/L (radians).

Resolution, SAR principle and geometric distortions

• Range (across-track) resolution depends only on pulse length/bandwidth, not on height: Rr = cτ/(2 sin θ) = c/(2B sin θ). A chirp (linear FM pulse) with pulse compression gives large B with enough transmitted energy.
• Azimuth (along-track) resolution of a real-aperture radar (SLAR) = β·R = λR/L — degrades with range, so a fine resolution from orbit would need an impossibly long antenna.
• Synthetic Aperture Radar (SAR): the moving platform records amplitude and phase of many echoes from each target; coherent processing synthesises an aperture of length Ls equal to the footprint, giving azimuth resolution ≈ L/2, independent of range and wavelength. A 10 m antenna gives ~5 m azimuth resolution at any altitude.
• Imaging modes: Stripmap (fixed beam, moderate swath/resolution), ScanSAR/wide swath (beam stepped in elevation — wide swath, coarser resolution), Spotlight (beam steered on one patch — finest resolution, small scene), TOPSAR (Sentinel-1 IW), and RISAT's FRS/MRS/CRS modes.
• Geometric distortions: slant-range (near-range) compression; foreshortening (slopes facing the radar appear shortened); layover (slope angle > incidence angle — the top is imaged before the base); shadow (back slopes steeper than the depression angle get no illumination). Layover and shadow are irreparable; correction uses a DEM in range-Doppler terrain correction.
• Radiometric effects: speckle — the salt-and-pepper grain from coherent interference of scatterers inside a resolution cell; it is a *noise-like signal*, multiplicative, reduced by multi-looking (averaging N independent looks cuts the standard deviation by √N) or by adaptive filters (Lee, Frost, Kuan, Gamma-MAP), at a cost in resolution.

Backscatter, polarisation and radar interpretation

• Radar brightness is reported as the backscattering coefficient σ⁰ in dB; typical values: calm water −25 to −20 dB (specular, dark), smooth bare soil −15 dB, vegetation −10 to −6 dB, urban/corner reflectors 0 dB and above.
• Controls on σ⁰: surface roughness relative to λ (Rayleigh/Fraunhofer criterion), dielectric constant (hence moisture — dry soil ε ≈ 3–8, water ε ≈ 80, so wet soil is bright), local incidence angle, wavelength, polarisation, and target geometry (corner reflection from buildings, volume scattering from canopy, double-bounce from flooded forest and inundated paddy).
• Scattering types: specular (water, roads — dark), diffuse (bare fields), volume (forest at L-band), double-bounce (walls, stems), point/corner (bright spots).
• Polarisation: like-polarised HH, VV and cross-polarised HV, VH; cross-pol arises from depolarising volume/multiple scattering, so HV is sensitive to vegetation structure. Dual-pol, compact (hybrid, circular transmit) pol — RISAT-1's hybrid-polarimetry — and quad-pol/fully polarimetric systems allow decompositions (Pauli, Freeman-Durden, Cloude-Pottier H/A/α) to separate surface, double-bounce and volume scattering.
• Interferometric SAR (InSAR): phase difference between two images from slightly different positions gives topographic height; the key terms are baseline B, perpendicular baseline B⊥, ambiguity height, coherence, interferogram, flat-earth removal, phase unwrapping. SRTM (2000) mapped near-global topography by single-pass C-band InSAR.
• Differential InSAR (DInSAR) and PSInSAR: repeat-pass phase differences measure line-of-sight ground movement to a few millimetres — subsidence (Kolkata, Delhi NCR, Joshimath), mining, tunnelling, landslides, earthquakes, dam and bridge monitoring. Persistent-scatterer/SBAS stacks beat temporal decorrelation.
• Other active microwave sensors: radar altimeter (sea-surface height, significant wave height — SARAL/AltiKa, Jason series), scatterometer (ocean-surface wind vectors — Oceansat-2/-3 OSCAT, SCATSAT-1), ground-penetrating radar (subsurface utilities, archaeology), radargrammetry (stereo SAR heights without phase).

LiDAR fundamentals, platforms and returns

• LiDAR = Light Detection And Ranging: a pulsed laser (commonly 1,064 nm NIR for topography; 1,550 nm eye-safer SWIR; 532 nm green for bathymetry) measures two-way travel time, hence range, while GNSS and an IMU give position and attitude. The geolocation equation combines laser range and scan angle with the trajectory, lever-arm offsets and boresight angles.
• Components: laser transmitter, scanner (oscillating mirror → zigzag, rotating polygon → parallel lines, Palmer → elliptical, fibre), receiver and detector, timing electronics, GNSS/IMU (direct georeferencing), data storage; on aircraft, usually a co-mounted camera.
• Platforms: ALS (airborne laser scanning — corridor and area mapping), UAV-LiDAR (small areas, very dense), TLS (terrestrial/static laser scanning — façades, tunnels, quarries; registration via targets or cloud-to-cloud ICP), MLS (mobile/vehicle-mounted — road assets), spaceborne (ICESat-2 ATLAS photon-counting, GEDI on the ISS for forest structure).
• Ranging principle: time-of-flight (pulsed, most survey lidar) vs phase-shift/continuous wave (short-range TLS, very precise) vs FMCW. Modern ALS also uses MPiA (multiple pulses in air) to fly higher without ambiguity.
• Returns: a pulse spreads over a footprint and may be split by canopy, so discrete-return systems log first, intermediate and last returns (first ≈ canopy top, last ≈ ground); full-waveform systems digitise the whole echo for vertical structure; photon-counting/single-photon systems detect individual photons (ICESat-2).
• Intensity is the recorded return amplitude — useful for classification and for a pseudo-image, but it needs range and incidence-angle normalisation before it is quantitative.
• Bathymetric lidar uses green light to penetrate water, giving depth ≈ 1–3 × Secchi depth; it maps the shallow nearshore zone that ship echo-sounding cannot reach.

LiDAR data processing and products

1. Trajectory solution: differential GNSS (with a CORS/base station) plus IMU in a tight-coupled Kalman filter.
2. Point-cloud generation and strip adjustment / boresight calibration: remove systematic offsets between overlapping strips (checked on flat hard surfaces and roof planes).
3. Noise/outlier removal: low points (multipath) and high points (birds, cloud).
4. Classification (filtering): separate ground from non-ground by progressive TIN densification, morphological/slope-based filters or cloth simulation; then buildings, vegetation (low/medium/high), water, bridges. The ASPRS LAS format fixes standard class codes (2 = ground, 3–5 vegetation, 6 = building, 9 = water); LAZ is the compressed form.
5. Products: DTM/DEM (bare earth, from ground points), DSM (first returns — includes buildings and canopy), nDSM/CHM = DSM − DTM (object/canopy height), contours, break-lines, 3-D building models (LoD1/LoD2), cross-sections, volumes, power-line catenary and clearance, forest metrics (canopy cover, height percentiles, AGB models).
6. Accuracy drivers: vertical accuracy typically 5–15 cm (1σ) for ALS, horizontal 2–3× worse and strongly dependent on flying height and IMU attitude error; errors grow with range × angular error. Accuracy is validated against GNSS/DGPS check points on open hard surfaces (RMSEz).
7. Specification parameters: point density (points/m²), swath width = 2H·tan(FOV/2), overlap (usually 20–30%), pulse rate, scan frequency, flying height, speed — these are what a tender asks for.

Comparisons, limitations and Indian practice

• Microwave strengths: cloud/day-night independence (vital in the Indian monsoon), moisture and structure sensitivity, mm-level deformation, canopy and dry-soil penetration. Limits: speckle, layover/shadow in hills, non-intuitive interpretation, no spectral colour information, coarse-to-moderate resolution for wide swaths.
• LiDAR strengths: direct, dense 3-D geometry, bare-earth under forest, very high vertical accuracy, works at night. Limits: cannot see through cloud, fog or heavy rain; no return from clear calm water at NIR; costly flying; huge data volumes; no inherent spectral classification.
• Indian regulation and practice (as of November 2026): aerial surveys and the carriage of survey-grade sensors need clearance under India's aerial-survey rules administered through the Survey of India/MoD channel; drone operations follow the Drone Rules, 2021 (as amended) with Digital Sky registration, UIN, remote-pilot certificate, and airspace zones — Beyond-visual-line-of-sight work needs specific permission. Geospatial data acquisition and sharing follow the Guidelines for Geospatial Data, 2021 (DST), which deregulated most ground and aerial geospatial data for Indian entities, while finer-than-threshold attribute data and some sensitive-site data remain controlled; the National Geospatial Policy, 2022 sets the 2030 target of high-resolution topographic and terrain data. Survey of India's vertical datum and the national CORS network provide control; the Geospatial Information Regulation Bill was never enacted. Exact thresholds and clearance routes change — confirm the current notification before quoting a number in practice.
• Indian programmes: RISAT series (C- and X-band SAR), EOS-04 (C-band, 2022), Oceansat/SCATSAT scatterometry, SARAL/AltiKa altimetry, and NISAR (launched 2025, L- and S-band sweep-SAR, free data) for deformation, cryosphere and agriculture; NRSC/Bhuvan distribute flood maps derived largely from SAR, and NHP/river-basin and smart-city projects have been mapped with aerial lidar.

Key points

• Active sensing — the sensor supplies the energy, so SAR and lidar are independent of sunlight; passive microwave only measures emitted radiation.
• Side-looking — radar must look sideways; a nadir beam cannot resolve left-right ambiguity.
• Range resolution — Rr = c/(2B sin θ): bandwidth and incidence angle only, never altitude.
• Azimuth resolution (RAR) = λR/L, degrades with range; SAR ≈ L/2, constant with range.
• Chirp + pulse compression — gives wide bandwidth without a destructively short pulse.
• Speckle — multiplicative, coherent-interference grain; multi-looking N looks reduces σ by √N.
• σ⁰ — backscattering coefficient in dB; set by roughness/λ, dielectric constant (moisture) and local incidence angle.
• Layover vs shadow — layover when slope > incidence angle; shadow when back-slope > depression angle; both unrecoverable.
• Polarisation — HH/VV like-pol, HV/VH cross-pol; cross-pol responds to volume and multiple scattering.
• Double-bounce — bright corner/stem returns; the signature of flooded vegetation and urban walls.
• InSAR — phase difference gives height; DInSAR/PSInSAR gives line-of-sight displacement to millimetres; coherence is the quality measure.
• Band choice — X for detail and texture, C for operational all-purpose and soil moisture, L/P for canopy and subsurface penetration.
• LiDAR ranging — R = c·t/2; a 1 ns timing error gives ~15 cm of range error.
• First vs last return — first return builds the DSM, last return the DTM; CHM = DSM − DTM.
• Direct georeferencing — lidar needs no ground control for geometry, but needs check points for accuracy verification and strip adjustment.
• LAS/LAZ — the ASPRS exchange format with standard classification codes (2 ground, 6 building, 9 water).
• Swath — lidar swath = 2H·tan(FOV/2); point density falls as speed, height or scan angle rises.
• Complementary, not rival — SAR for all-weather change and deformation, lidar for precise 3-D geometry; optical for spectral content.

Formulas and examples

Example 1 — SAR resolutions. A spaceborne SAR: λ = 5.6 cm (C-band), antenna L = 10 m, slant range R = 850 km, chirp bandwidth B = 50 MHz, incidence angle θ = 30°.

• Range resolution Rr = 3 × 10⁸ / (2 × 50 × 10⁶ × sin 30°) = 3 × 10⁸ / (10⁸ × 0.5) = 6 m.
• Beamwidth β = 0.056/10 = 5.6 × 10⁻³ rad → real-aperture azimuth resolution = 5.6 × 10⁻³ × 850,000 = 4,760 m (≈ 4.8 km).
• SAR azimuth resolution = L/2 = 5 m. The synthetic aperture used is Ls = λR/L = 4,760 m.
• Interpretation: SAR improves azimuth resolution here by a factor of about 950.

Example 2 — slant range, ground range and target height. Travel time t = 5.8 ms from a platform at H = 800 km with θ = 35°.

• R = 3 × 10⁸ × 5.8 × 10⁻³ / 2 = 870 km.
• Ground range from nadir = R·sin θ = 870 × 0.5736 = 499 km.
• Flat-earth target elevation h = H − R·cos θ = 800 − 870 × 0.8192 = 800 − 712.7 = 87.3 km — physically impossible, which shows the flat-earth form fails at orbital ranges: Earth curvature must be included. For an airborne case (H = 6 km, R = 7.32 km, θ = 35°): h = 6 − 7.32 × 0.8192 = 6 − 6.00 = 0 m, i.e. the target lies at the datum.

Example 3 — speckle and multi-looking. A single-look intensity image has a coefficient of variation of 1.0 (100%).

• 4 looks → 1/√4 = 0.50 (50%); 16 looks → 1/√16 = 0.25 (25%).
• If one look is 5 m × 5 m, a 4-look azimuth average gives a 5 m × 20 m cell, usually resampled to about 20 m × 20 m square pixels — a 4× loss of azimuth detail for a halving of noise.

Example 4 — DInSAR displacement. A C-band interferogram (λ = 5.6 cm) over a subsiding city shows 3.5 fringes in one year between a stable reference and a settlement bowl.

• One fringe = λ/2 = 2.8 cm of line-of-sight change.
• LOS displacement = 3.5 × 2.8 = 9.8 cm/year away from the sensor.
• Vertical component (assuming purely vertical motion, θ = 34°): d_vertical = d_LOS / cos θ = 9.8 / 0.829 = 11.8 cm/year of subsidence.

Example 5 — lidar mission planning. ALS flown at H = 1,500 m AGL, total FOV = 40°, PRF = 300 kHz, aircraft speed = 60 m/s, side overlap 30%.

• Swath SW = 2 × 1,500 × tan 20° = 2 × 1,500 × 0.36397 = 1,091.9 m.
• Line spacing at 30% overlap = 0.7 × 1,091.9 = 764.3 m.
• Point density (single strip) = 300,000 / (60 × 1,091.9) = 300,000 / 65,514 = 4.58 points/m².
• Effective density allowing for the 30% double-covered strip edges: average over the line spacing = 300,000/(60 × 764.3) = 6.54 points/m².
• Range error from timing: with δt = 0.5 ns, δR = 3 × 10⁸ × 0.5 × 10⁻⁹ / 2 = 7.5 cm — which is why timing and attitude, not the laser, usually limit accuracy.

Where it is used

• Flood and disaster response: NRSC/Bhuvan flood inundation maps for Assam, Bihar, Kerala and Odisha are made mainly from C-band SAR because the monsoon keeps optical sensors blind; also cyclone wind fields from scatterometer data for IMD.
• Deformation and infrastructure monitoring: DInSAR/PSInSAR for city subsidence, Himalayan landslides (Joshimath), metro tunnelling, mines, dams and bridges — used by NRSC, CGWB-linked studies, IITs and private geospatial consultancies.
• Agriculture and forestry: SAR-based rice and crop area/condition for the FASAL and crop-insurance (PMFBY) work streams; lidar/GEDI-type canopy height and biomass for FSI-style forest inventory.
• Corridor and city mapping with aerial/UAV LiDAR: highways (NHAI DPRs), railway and metro alignments, transmission-line clearance (POWERGRID), river-basin and flood-plain DTMs (NHP), smart-city 3-D models and property surveys — delivered by firms such as Genesys International, Ceinsys, RMSI, ADCC and L&T Geo-informatics.
• Coastal and ocean work: radar altimetry for sea level and wave height (SARAL/AltiKa, INCOIS services), bathymetric lidar and hydrographic charting under the Indian hydrographic programme.
• Job roles: SAR analyst/InSAR processing engineer, LiDAR data processor and point-cloud classifier, geospatial project engineer, UAV survey pilot-operator, DEM/DTM QC analyst — with SNAP, GMTSAR/ISCE, ENVI SARscape, TerraSolid, LAStools, Global Mapper, Cyclone/RealWorks and PDAL/Python as the working tools.

Interview questions

• Q: Why must a radar imaging system look sideways? A: A nadir-looking beam would receive echoes of equal slant range from targets on both sides of the flight line, which cannot be separated; side-looking makes range monotonic across the swath.
• Q: Does flying higher worsen SAR range resolution? A: No — range resolution depends only on pulse bandwidth and incidence angle (c/2B sin θ); altitude affects swath and signal strength, not range resolution.
• Q: How does SAR achieve fine azimuth resolution with a short antenna? A: It records the phase history of each target while the platform moves and coherently combines the echoes, synthesising an aperture equal to the beam footprint, giving resolution ≈ half the physical antenna length.
• Q: Why is calm water dark and a flooded forest bright in a SAR image? A: Smooth water reflects specularly away from the sensor; in a flooded forest the water-trunk corner produces strong double-bounce back to the sensor.
• Q: Is speckle noise, and can averaging remove it? A: It is a deterministic interference effect of coherent imaging, treated as multiplicative noise; multi-looking or adaptive filtering reduces its variance (by √N for N looks) at the cost of spatial resolution, but it cannot be eliminated.
• Q: What causes layover, and can it be corrected? A: Terrain slope steeper than the incidence angle, so the slope top is nearer the radar than its base; the information is irretrievably mixed, so it cannot be corrected — only masked or avoided by a different look direction.
• Q: What does coherence tell you in InSAR? A: The phase similarity between two acquisitions — high over bare rock and urban areas, low over vegetation and after surface change; low coherence makes the interferometric phase, and hence the height or displacement, unreliable.
• Q: Which band would you choose to map forest biomass, and why? A: L-band (or P-band), because the longer wavelength penetrates the canopy and scatters from trunks and large branches, which carry most of the biomass; C- and X-band saturate at the canopy top.
• Q: How does lidar give a bare-earth DTM under forest? A: Some pulses pass through canopy gaps and reflect from the ground as last returns; ground filtering separates those points, and interpolation over them produces the DTM.
• Q: Why does lidar not need ground control points the way photogrammetry does? A: It is directly georeferenced by GNSS/IMU, so each point's coordinates come from the trajectory and the measured range; check points are still needed to verify accuracy and to resolve strip offsets and systematic biases.
• Q: Why does a NIR lidar leave holes over lakes? A: Water absorbs 1,064 nm strongly and reflects specularly away from the sensor, so no return is recorded; green (532 nm) bathymetric lidar is used instead for shallow water.
• Q: SAR or LiDAR for monitoring 5 mm/year of city subsidence over 200 km²? A: SAR — PSInSAR on a long repeat-pass stack measures millimetre line-of-sight displacement over large areas from free archives, whereas repeated lidar flights cost far more and have 5–15 cm vertical accuracy, too coarse for that rate.

Common mistakes

• Inserting platform height into the range-resolution formula, or forgetting the sin θ term (and using the depression angle instead of the incidence angle).
• Writing SAR azimuth resolution as λR/L — that is the real-aperture value; SAR gives ≈ L/2.
• Saying microwaves "penetrate everything": they penetrate cloud and dry soil, but wet soil, dense wet canopy and water strongly attenuate the signal.
• Treating speckle as random sensor noise to be removed by a simple mean filter, which blurs edges and biases σ⁰.
• Confusing layover with foreshortening and shadow, or claiming layover can be geometrically corrected with a DEM.
• Interpreting raw lidar intensity as reflectance without normalising for range and incidence angle, and mixing up first return (DSM) with last return (DTM).
• Quoting a lidar point density without stating overlap, flying height, scan angle and speed — densities differ by a factor of two between single-strip and overlapped averages.
• Forgetting the λ/2 (not λ) per fringe in differential interferometry, and reporting line-of-sight motion as if it were vertical.

Related topics

Remote sensing sensors, platforms and Indian satellites · Digital image processing for remote sensing · Photogrammetry and drone surveys · GIS and remote sensing · DEM generation and terrain analysis · Geodesy, GNSS and coordinate systems