Physical Principles and Data Acquisition
Open in the Padhaiko app1.1 Electromagnetic Radiation
Remote sensing acquires information about targets without physical contact by measuring electromagnetic radiation (EMR) reflected or emitted from the Earth's surface and atmosphere. Two complementary physical descriptions govern how EMR propagates and interacts with matter: the wave model (Maxwell's electromagnetic theory) and the particle (quantum) model.
Maxwell's Wave Formulation
James Clerk Maxwell unified electricity and magnetism into four coupled differential equations showing that a time-varying electric field generates a time-varying magnetic field, and vice versa. In a homogeneous, isotropic, source-free medium (such as vacuum or dry air), Maxwell's equations reduce to the electromagnetic wave equation:
where is the electric field vector, is the magnetic flux density, is magnetic permeability, and is electric permittivity.
Electromagnetic Wave Properties
An electromagnetic wave is a transverse wave in which oscillating electric () and magnetic () fields are mutually perpendicular to each other and to the direction of propagation:
- Propagation speed in vacuum:
- Fundamental wave relation: Wavelength (m) and frequency (Hz) satisfy
- Wavenumber (): Often used in infrared spectroscopy, defined as (commonly in ).
- Wave amplitude and irradiance: The instantaneous energy flux density is given by the Poynting vector , whose time average is proportional to .
| Spectral Region | Approximate Wavelength Range | Primary Remote Sensing Mechanism |
|---|---|---|
| Ultraviolet (UV) | Mostly scattered/absorbed by ozone; fluorescence studies | |
| Visible (Blue, Green, Red) | Solar reflection; true-color imaging, chlorophyll & water color | |
| Near-Infrared (NIR) | Solar reflection; vegetation cell structure, biomass, land–water boundary | |
| Short-Wave Infrared (SWIR) | Solar reflection; leaf/soil moisture, mineral & clay mapping | |
| Mid-Wave Infrared (MWIR) | Mixed solar reflection and thermal emission; active fires, hotspots | |
| Thermal Infrared (TIR) | Self-emission from Earth; Land Surface Temperature (LST), heat islands | |
| Microwave / Radar | Active backscatter (SAR/GPR) & passive emission; all-weather, cloud-penetrating |
Electromagnetic Energy and Energy Quantization
While wave optics describes propagation, interference, and diffraction, the interaction of EMR with detectors (photodiodes, CCDs) and atoms (absorption and emission lines) requires quantum theory. Max Planck and Albert Einstein showed that electromagnetic energy is transferred in discrete packets called photons (quanta).
Photon Energy Quantization
The energy (or ) of a single photon is directly proportional to its frequency and inversely proportional to its wavelength :
where is Planck's constant and is the speed of light in vacuum.
- Longer wavelengths (thermal IR, microwave) carry much lower energy per photon than shorter wavelengths (visible, UV).
- Consequently, passive sensors operating at longer wavelengths require larger instantaneous fields of view (IFOV) or wider spectral bandwidths to collect enough photons for an acceptable signal-to-noise ratio.
Solar Irradiance and Blackbody Radiation
All objects at absolute temperature emit electromagnetic radiation due to thermal agitation of charged particles. An idealized blackbody is a hypothetical object that absorbs of all incident radiation at every wavelength () and emits the maximum possible radiant flux for a given temperature .
The Sun behaves approximately as a blackbody at (often approximated as ). Outside Earth's atmosphere at the mean Sun–Earth distance (), the total integrated solar irradiance across all wavelengths is the solar constant:
Because Earth's surface has an average ambient temperature around , its thermal emission peaks in the thermal infrared rather than the visible spectrum.
Planck's Law, Stefan-Boltzmann Law, and Wien's Displacement Law
Fundamental Laws of Blackbody Radiation
For an ideal blackbody at absolute temperature (in Kelvin):
- Planck's Law (Spectral Radiant Exitance): Gives the spectral exitance ( or ) emitted into a hemisphere at wavelength :
where is the Boltzmann constant, , and .
- Stefan-Boltzmann Law (Total Radiant Exitance): Integrating Planck's law over all wavelengths yields the total power emitted per unit area:
where is the Stefan-Boltzmann constant. For a real greybody with total emissivity , .
- Wien's Displacement Law (Peak Wavelength): Differentiating with respect to and setting yields the wavelength of maximum emission:
where (commonly rounded to ).
1.2 Atmospheric Propagation and Windows
Before reaching a satellite or airborne sensor—and on the downward solar path—radiation traverses Earth's atmosphere. The atmosphere modifies both the intensity and spectral composition of EMR through absorption and scattering.
Atmospheric Absorption and Transmission
Gas molecules selectively absorb photons whose energies match quantized rotational, vibrational, or electronic transitions. The fraction of incident radiation transmitted through a vertical or slant atmospheric path is the atmospheric transmittance , related to optical depth by Beer–Lambert's law:
where is the solar or viewing zenith angle.
| Atmospheric Constituent | Key Absorption Bands | Physical Effect & Operational Consequence |
|---|---|---|
| Ozone () | (Hartley–Huggins UV), (TIR) | Blocks harmful UV radiation; creates a narrow absorption notch inside the thermal window near |
| Water Vapor () | , , , , , | Highly variable in space and time; strongly absorbs NIR/SWIR bands (e.g., Landsat 8/9 & Sentinel-2 use for cirrus detection) |
| Carbon Dioxide () | , , | Well-mixed gas; blocks mid-IR around and sets the long-wave limit of the TIR window near |
| Oxygen () | (-B), (-A), (microwave) | Sharp narrow bands used for cloud-top pressure and surface pressure retrieval; used for atmospheric temperature sounding |
| Methane () & Nitrous Oxide () | , , | Minor absorbers targeted by greenhouse-gas monitoring spectrometers (e.g., TROPOMI, MethaneSAT) |
Scattering Mechanisms: Rayleigh, Mie, and Non-Selective Scattering
Scattering redirects photons out of their original propagation path without permanent absorption. The physical regime depends on the size parameter comparing particle diameter to wavelength :
Three Regimes of Atmospheric Scattering
- Rayleigh Scattering (): Caused by gas molecules (, ) whose diameters () are much smaller than optical wavelengths. The scattering intensity scales inversely with the fourth power of wavelength:
- Blue light () is scattered roughly times more strongly than red light ().
- Causes the blue sky, reddish sunsets, and hazy contrast reduction in short-wavelength satellite bands (known as path radiance).
- Mie Scattering (): Caused by suspended aerosols, dust, pollen, smoke, and fine haze droplets whose diameters () are comparable to visible and NIR wavelengths.
- Scattering varies roughly as to (much weaker wavelength dependence than Rayleigh) and is strongly forward-directed.
- Non-Selective Scattering (): Caused by large water droplets () and ice crystals in clouds and fog.
- All visible and NIR wavelengths are scattered nearly equally (), which is why clouds and fog appear bright white and opaque to optical sensors.
Atmospheric Transmission Windows
An atmospheric window is a wavelength region where atmospheric absorption is minimal and transmittance is high, allowing radiation to pass between the Earth's surface and spaceborne sensors.
| Atmospheric Window | Wavelength Range | Key Sensors / Bands Operating Here |
|---|---|---|
| Visible – NIR (VNIR) | Optical cameras, Landsat OLI, Sentinel-2 MSI (Blue, Green, Red, Red-Edge, NIR) | |
| SWIR Windows | and | Vegetation water stress & mineral absorption (Landsat/Sentinel-2 SWIR-1 and SWIR-2) |
| MWIR Window | Nighttime SST, active fire & volcanic hotspot detection (MODIS, VIIRS) | |
| TIR Window | (notable dip at ) | Thermal imaging, Land Surface Temperature (Landsat TIRS , ) |
| Microwave Window | (X, C, L, P bands) | Virtually transmission through clouds and light rain (Sentinel-1, ALOS PALSAR) |
1.3 Spectral Signatures
When incident solar radiation strikes a surface feature, conservation of energy requires that it be partitioned into reflected, absorbed, and transmitted components:
Dividing by defines spectral reflectance , absorptance , and transmittance :
The curve of as a function of wavelength is the target's spectral signature (spectral reflectance curve), which enables discrimination of land-cover materials.
Specular and Diffuse Reflection
The angular distribution of reflected energy depends on surface roughness relative to the wavelength (the Rayleigh roughness criterion):
- Specular Reflector: Surface height variations are much smaller than (e.g., calm water, smooth paved roads at microwave wavelengths). Incident radiation reflects at an angle equal to the angle of incidence () in the forward plane, producing bright sunglint when viewed directly and appearing dark from all other viewing angles.
- Diffuse (Lambertian) Reflector: Surface roughness is comparable to or larger than (e.g., matte soil, dense grass). An ideal Lambertian surface scatters incident radiation equally in all directions so its apparent radiance is independent of viewing angle.
Bidirectional Reflectance Distribution Function (BRDF)
Real Earth surfaces are neither purely specular nor perfectly Lambertian; their reflectance depends on both the illumination geometry and the viewing geometry.
Bidirectional Reflectance Distribution Function (BRDF)
The BRDF, denoted (units: ), is defined as the ratio of the infinitesimal radiance reflected in viewing direction to the infinitesimal irradiance incident from direction :
where is the zenith angle and is the azimuth angle.
- For an ideal Lambertian surface of reflectance , radiance is isotropic and , giving a constant BRDF:
- Real canopies exhibit strong hotspot effects (bright backscatter when sun and sensor align, , , hiding internal shadows) and anisotropic forward/backward scattering.
Characteristic Spectral Reflectance Curves
| Surface Type | Visible () | Near-Infrared () | Short-Wave Infrared () | Governing Physical Mechanism |
|---|---|---|---|---|
| Healthy Green Vegetation | Low (); slight green peak at ; strong blue () & red () troughs | Very high (); steep red-edge rise between | Moderate, with strong water absorption dips near and | Chlorophyll absorption in visible; multiple internal scattering in spongy mesophyll in NIR; leaf water content in SWIR |
| Dry Bare Soil | Moderate, rising monotonically from blue to red | Continues rising gradually () | High, with clay mineral absorption dip around | Iron oxide content reddens soil; moisture darkens soil across all wavelengths; organic matter lowers reflectance |
| Clear Liquid Water | Low (), peaking slightly in blue-green () | Near zero () | Zero () | Strong absorption by liquid water beyond ; turbidity/sediment raises red reflectance; chlorophyll raises green |
| Built Structures (Concrete, Asphalt, Roofs) | Flat to gently rising (; asphalt low, concrete high) | Nearly flat or slightly rising () without a sharp red-edge | Moderate to high () depending on mineral/paint composition | Broad mineral, aggregate, and bitumen absorption; lack of chlorophyll absorption or spongy mesophyll jump |
1.4 Sensor Electronics and Architectures
An optical remote sensing payload collects reflected or emitted photons using fore-optics (mirrors/lenses), separates wavelengths using filters or dispersive elements (prisms/gratings), and converts photons into electrical charge and digital numbers at the focal plane.
Optical Sensor Payloads and Focal-Plane Arrays (FPAs)
A focal-plane array (FPA) is a 1D linear or 2D rectangular matrix of photosensitive detector elements (pixels) placed at the focal plane of the optical telescope. Each detector converts incident spectral radiance into charge carriers via the internal photoelectric effect (silicon for VNIR; InGaAs or HgCdTe for SWIR and TIR, which often require cryogenic cooling to suppress dark current).
CCD vs. CMOS Sensors
| Aspect | CCD (Charge-Coupled Device) | CMOS (Active Pixel Sensor) |
|---|---|---|
| Charge-to-Voltage Conversion | Charge packets are shifted pixel-by-pixel across the chip to a single (or few) shared output amplifier(s) | Each pixel contains its own photodiode and amplifier/readout transistor, converting charge to voltage locally |
| Readout Speed & Power | Slower serial readout; higher drive voltage and power consumption | Very fast parallel/column readout; lower power consumption (favored in smallsats) |
| Uniformity & Dynamic Range | Historically superior pixel-to-pixel uniformity and low fixed-pattern noise | Modern scientific CMOS (sCMOS) matches CCD noise with on-chip ADC and radiation-tolerant design |
| Susceptibility to Artifacts | Can suffer vertical charge blooming/smearing under intense saturation | Susceptible to rolling-shutter distortion unless global shutter is used |
| Satellite Examples | Landsat 7 ETM+ (discrete detectors), older pushbroom SPOT instruments | Landsat 8/9 OLI (14-bit), Sentinel-2 MSI VNIR FPA |
Scanning Architectures: Whiskbroom vs. Pushbroom Systems
| Aspect | Whiskbroom (Across-Track) Scanner | Pushbroom (Along-Track) Scanner |
|---|---|---|
| Scanning Mechanism | An oscillating/rotating mirror sweeps the field of view perpendicular to the flight track across a single or small group of detectors | A linear array of thousands of detectors spans the entire swath width; forward motion of the satellite provides along-track scanning |
| Moving Parts | Mechanical scan mirror (wear, jitter, calibration drift) | No moving scan mirror (higher mechanical reliability and geometric stability) |
| Dwell Time () | Very short (one pixel at a time across the swath), limiting SNR and forcing wider bands or larger pixels | Much longer (-fold increase for cross-track detectors), yielding higher SNR and finer spectral/spatial resolution |
| Calibration Challenge | Only one/few detectors to calibrate radiometrically | Thousands of detectors must be cross-calibrated; mismatch causes vertical striping |
| Heritage Missions | Landsat 1–5 MSS, Landsat 4–5 TM, Landsat 7 ETM+, MODIS | Landsat 8/9 OLI, Sentinel-2 MSI, SPOT HRV |
1.5 System Resolutions and Trade-offs
The capability of any Earth-observation sensor is defined by four fundamental resolutions, together with radiometric fidelity (bit depth and SNR).
Spatial, Spectral, Radiometric, and Temporal Resolution
The Four Sensor Resolutions
- Spatial Resolution (Ground Sampling Distance, GSD): The ground size represented by a single pixel, governed by the Instantaneous Field of View (IFOV) (in radians) and orbital altitude :
where is the detector pixel pitch and is the telescope focal length. Smaller GSD means finer spatial detail (e.g., vs ).
- Spectral Resolution: The number and width () of wavelength bands measured by the sensor. Narrower, more numerous bands (e.g., or contiguous hyperspectral bands) resolve subtle absorption features that broad panchromatic bands () merge together.
- Radiometric Resolution (Bit Depth): The sensitivity of the detector to small differences in incoming radiance, expressed as the bit depth of the Analog-to-Digital Converter (ADC). An -bit sensor quantizes radiance into discrete Digital Numbers (DN) from to .
- Temporal Resolution (Revisit Time): The time elapsed between consecutive observations of the same ground target. Depends on orbital altitude/inclination, swath width, off-nadir pointing capability, and the number of satellites in the constellation.
Bit Depth and Signal-to-Noise Ratio (SNR)
Signal-to-Noise Ratio (SNR) quantifies how clearly true target radiance stands out above random detector noise (shot noise, thermal dark current, and readout noise):
Higher bit depth (-bit levels, -bit levels) reduces quantization noise, but extra bits are only useful if the sensor's physical SNR is high enough that the least significant bits are not dominated by random noise.
Trade-offs Among Different Types of Resolution
Because a sensor receives a finite photon flux within a limited satellite mass, power, and downlink bandwidth budget, improving one resolution forces compromises in others:
- Spatial vs. Spectral vs. SNR: Smaller pixels (finer spatial resolution) and narrower bands (finer spectral resolution) both reduce the number of photons collected per sample, lowering SNR unless dwell time or aperture diameter is increased. For this reason, panchromatic bands are wider spectrally and achieve finer spatial resolution (e.g., on Landsat 8/9 vs multispectral).
- Spatial vs. Temporal (Swath Width): High spatial resolution sensors typically have narrow fields of view and narrow swaths, resulting in longer revisit intervals unless multiple satellites fly in a constellation.
- Radiometric/Spatial vs. Data Volume: Higher bit depth and smaller GSD multiply onboard storage and downlink data rates ().
Modern Earth-Observation Constellations
| System / Constellation | Spatial Resolution | Spectral Bands | Radiometric Depth | Temporal Revisit | Architecture & Role |
|---|---|---|---|---|---|
| Landsat 8 / 9 (OLI & TIRS) | MS, Pan, TIR (resampled to ) | 11 bands (VNIR, SWIR, Cirrus, Pan, 2 TIR) | -bit (L8) / -bit (L9) | each ( combined) | Pushbroom; swath; long-term + year calibrated archive |
| Sentinel-2 (2A / 2B / 2C MSI) | (RGB+NIR), (Red-Edge+SWIR), (Atmos) | 13 bands (including 3 Red-Edge bands) | -bit | single ( pair at equator) | Pushbroom; swath; European Copernicus agriculture & land monitoring |
| Commercial Smallsats (e.g., PlanetScope, SkySat) | (PlanetScope), (SkySat) | 4–8 VNIR bands | -bit | Daily (PlanetScope constellation of CubeSats) | CMOS FPAs on agile micro/nanosatellites; rapid disaster & precision monitoring |
Worked Out Examples
1.1
Model the Sun as a blackbody at and an agricultural land surface as a blackbody at .
- Compute the wavelength of peak emission for both bodies using Wien's displacement law ().
- Compute the total radiant exitance () emitted by each surface using the Stefan-Boltzmann law ().
Applying Wien's displacement law:
- For the Sun ():
- For the Earth surface ():
Applying the Stefan-Boltzmann law:
- For the Sun:
- For the Earth surface:
- Peak wavelengths: (visible), (thermal IR).
- Total radiant exitances: , .
1.2
- Compare the photon energy (in Joules and in , with ) of a green visible photon () and a thermal infrared photon ().
- Using Rayleigh's scattering law (), calculate how many times more strongly blue light () is scattered by atmospheric gas molecules than red light ().
Computing photon energies via : With and , we have :
- For :
- For :
A visible green photon carries times more energy than a TIR photon.
Comparing Rayleigh scattering intensities:
- Green photon (): (); TIR photon (): ().
- Blue light at undergoes times stronger Rayleigh scattering than red light at .
1.3
A pushbroom Earth-observation satellite orbits at altitude with focal length and detector pixel pitch . Its swath width is , and its ADC uses -bit radiometric quantization.
- Determine the Ground Sampling Distance (GSD) at nadir.
- Estimate the number of detector elements required across the linear focal-plane array.
- Find the number of discrete Digital Number (DN) levels and the dynamic range in decibels () of the -bit ADC.
Calculating Ground Sampling Distance (GSD): The sensor's IFOV is :
Multiplying by orbital altitude :
Estimating cross-track detector count:
Computing radiometric levels and dynamic range: With bits:
- .
- Minimum cross-track array size: detectors.
- discrete DN levels (), corresponding to .
Exercises
Exercise — Physical Principles and Data Acquisition
Derive the relationship between wavelength, frequency, and speed of an electromagnetic wave from Maxwell's wave formulation, and distinguish between the wave and photon descriptions of EMR. Using , calculate the peak emission wavelength and total blackbody radiant exitance of the Sun (, ).
Maxwell's wave equation and wave parameters: In a charge-free vacuum, Maxwell's curl equations combine to yield , which admits plane-wave solutions propagating at speed . A sinusoidal wave of spatial period and temporal frequency advances one wavelength per period , giving .
Wave vs. quantum (photon) models: The wave model explains propagation, polarization, interference, and diffraction, whereas photon quantization () governs discrete absorption and emission in atoms and semiconductor detectors.
Wien's and Stefan-Boltzmann calculations at :
follows directly from Maxwell's wave equation with ; photon energy is . At , and .
Explain the three atmospheric scattering mechanisms (Rayleigh, Mie, and non-selective) and list the major absorbers (, , , ) that define atmospheric transmission windows. Compute the ratio of Rayleigh scattering intensity at the violet-blue edge () to the red edge ().
Classifying the three scattering regimes:
- Rayleigh (): Gas molecules scatter strongly at short wavelengths (), causing path radiance and blue sky.
- Mie (): Aerosols, smoke, and dust scatter forward with weaker dependence ( to ).
- Non-selective (): Cloud droplets scatter all visible and NIR wavelengths equally, blocking optical ground viewing.
Absorbers and atmospheric windows: blocks UV () and absorbs at ; creates deep absorption bands around , and ; absorbs strongly at and beyond ; produces sharp bands at and . The gaps between these bands form the VNIR (), SWIR ( and ), MWIR (), TIR (), and microwave windows.
Rayleigh scattering ratio between and :
Rayleigh (, ), Mie (), and non-selective (, ) scattering interact with , , , and absorption bands to leave VNIR, SWIR, MWIR, TIR, and microwave windows. Rayleigh scattering at is times stronger than at .
Define the Bidirectional Reflectance Distribution Function (BRDF) and contrast specular reflection with diffuse (Lambertian) reflection. If an ideal Lambertian soil plot has spectral reflectance at , what is its BRDF value in ? Also summarize the key spectral features that separate healthy vegetation, bare soil, clear water, and built structures across .
BRDF, specular vs. Lambertian reflection: The BRDF is (). Specular surfaces reflect predominantly at in the forward direction, whereas an ideal Lambertian surface scatters uniformly into the hemisphere so is constant in all directions.
Numerical BRDF for : For a Lambertian target, hemisphere integration gives , so:
Spectral curves across :
- Vegetation: Chlorophyll absorption in blue/red with green peak (), steep red-edge to high NIR plateau (), and water-absorption dips at and .
- Soil: Smooth, monotonically rising reflectance from visible through SWIR with minor clay dips near .
- Clear Water: Low visible reflectance and nearly zero reflectance in NIR and SWIR due to strong liquid-water absorption.
- Built Structures: Flat to gradually varying mineral/concrete/asphalt curves lacking both the sharp NIR red-edge and strong leaf water dips.
BRDF ; for a Lambertian surface with , . Vegetation shows chlorophyll troughs, a high NIR plateau, and SWIR water dips; soils rise monotonically; clear water absorbs nearly all NIR/SWIR; built surfaces remain relatively flat without a red-edge.
Compare CCD and CMOS focal-plane arrays and contrast pushbroom and whiskbroom scanning architectures in terms of dwell time, SNR, moving parts, and radiometric calibration.
CCD vs. CMOS comparison: CCDs transfer photo-generated charge packets serially to a shared output amplifier, giving high pixel uniformity and low fixed-pattern noise at the cost of slower readout, higher power, and blooming under saturation. CMOS Active Pixel Sensors integrate an amplifier per pixel, enabling rapid parallel readout, low power consumption, and on-chip ADCs suited to modern constellations and smallsats.
Whiskbroom vs. Pushbroom comparison:
- Whiskbroom: Uses a rotating/oscillating mirror sweeping across track over a few detectors; suffers mechanical wear and very short pixel dwell time, limiting SNR, though cross-detector calibration is simpler.
- Pushbroom: Uses a linear FPA covering the entire swath with no moving scan mirror; achieves an -fold longer dwell time (higher SNR and finer spatial/spectral resolution), but requires precise relative gain/offset calibration across thousands of detectors to prevent vertical striping.
CCDs excel in charge-transfer uniformity but consume more power and read out serially, whereas CMOS offers fast parallel readout and low power. Pushbroom scanners eliminate moving mirrors and gain -times longer dwell time (higher SNR) over whiskbroom scanners, at the expense of requiring multi-detector flat-field calibration to avoid striping.
Define spatial, spectral, radiometric, and temporal resolution and explain the fundamental engineering trade-offs among them. A Sentinel-2-like MSI operates at altitude with a VNIR GSD and swath width. Calculate (a) its IFOV in milliradians () and (b) the number of pixels across the swath, and briefly compare Landsat 8/9, Sentinel-2, and commercial smallsat constellations.
Definitions and resolution trade-offs:
- Spatial (GSD), spectral (band number and ), radiometric ( DN levels and SNR), and temporal (revisit period) resolutions compete for photons and downlink bandwidth. Narrowing bands or shrinking GSD cuts photon counts per pixel, degrading SNR unless aperture or dwell time grows; wider swaths improve temporal revisit at the expense of larger off-nadir viewing angles.
Computing IFOV and swath pixels: (a) . (b) Across a swath at GSD:
Constellation comparison:
- Landsat 8/9: MS ( Pan, TIR), 11 bands, /-bit, -day single (-day pair) revisit; benchmark radiometric continuity.
- Sentinel-2: , 13 bands including 3 Red-Edge bands, -bit, -day combined revisit, swath.
- Commercial smallsats (e.g., PlanetScope): VNIR with daily revisit via large CubeSat constellations, trading narrow SWIR/TIR bands and radiometric uniformity for high temporal and spatial detail.
(a) (); (b) pixels across the swath. Trade-offs balance photon-limited SNR, swath-driven revisit, and downlink capacity across Landsat 8/9 (, TIR), Sentinel-2 (, Red-Edge, -day), and daily smallsat constellations.