Fact-checked by Grok 2 weeks ago

Spectral signature

A spectral signature is the unique pattern of electromagnetic radiation reflected, absorbed, or emitted by a material across different wavelengths, serving as a distinctive "fingerprint" for identifying substances in remote sensing applications.^[1]^[2] This signature arises from the interaction of light with molecular structures, where materials like vegetation reflect strongly in the near-infrared while absorbing in the visible red, contrasting with water's low reflectance across most wavelengths.^[2]^[3] In remote sensing, spectral signatures enable the classification and mapping of Earth's surface features by analyzing data from multispectral or hyperspectral sensors aboard satellites and aircraft, which capture reflectance in specific wavelength bands to differentiate land cover types such as forests, grasslands, soils, and water bodies.^[1] For instance, sensors like those on Landsat missions record these patterns to produce thematic maps that highlight variations in vegetation health or mineral composition.^[2] The precision of these signatures depends on spectral resolution—the number and narrowness of wavelength bands—allowing for finer distinctions in hyperspectral imaging compared to broader multispectral approaches.^[4] Spectral signatures find broad applications in environmental monitoring, agriculture, geology, and resource management, where they support tasks like assessing crop vitality through chlorophyll absorption peaks, detecting mineral deposits via unique absorption features, or tracking water quality by algal bloom signatures.^[1] In agriculture, farmers use these patterns to evaluate plant stress from drought or disease; geologists apply them to explore for ores; and ecologists monitor biodiversity via changes in forest cover.^[1] Advances in sensor technology continue to refine signature analysis, enhancing accuracy in global-scale observations despite challenges like atmospheric interference or mixed-pixel effects.^[4]

Fundamentals

Definition

A spectral signature is the unique pattern of interaction between a material and electromagnetic radiation across various wavelengths, characterized by the specific wavelengths where the material absorbs, reflects, transmits, or emits energy, functioning as a distinctive "fingerprint" for identifying and distinguishing materials.^[1] This pattern arises from the material's response to incident radiation, enabling differentiation between substances like vegetation, water, and soil based on their reflectance or emittance profiles.^[2] The term "spectral signature" originated in the 1960s amid the early development of multispectral remote sensing techniques, pioneered by NASA and the U.S. Geological Survey (USGS) to analyze Earth surface features through airborne and space-based imaging.^[5] It was formalized in literature as the spectral response of features observed over a range of wavelengths, supporting the transition from traditional aerial photography to quantitative spectral analysis.^[6] Key components of a spectral signature include the relevant wavelength ranges—such as the visible (0.4–0.7 μm), near-infrared (0.7–2.5 μm), and thermal infrared (8–14 μm)—and the corresponding variations in intensity, which reflect the underlying molecular and atomic structures of the material influencing energy interactions.^[1] For reflectance-based signatures, the core metric is spectral reflectance, defined by the equation

\rho(\lambda) = \frac{E_r(\lambda)}{E_i(\lambda)},

where \rho(\lambda) is the reflectance at wavelength \lambda, E_r(\lambda) is the reflected irradiance, and E_i(\lambda) is the incident irradiance.^[7]

Physical Principles

Spectral signatures arise from the interaction of electromagnetic radiation with matter, primarily through absorption, reflection, and emission processes governed by quantum mechanical principles. Absorption occurs when photons excite electrons, molecules, or atoms to higher energy states, leading to characteristic features in the spectrum. Electronic transitions, involving the promotion of electrons between molecular orbitals, produce broad absorption bands typically in the ultraviolet, visible, and near-infrared regions, as these require higher energy photons. Vibrational modes, where molecules oscillate around equilibrium bond lengths, result in narrower absorption features in the infrared, often as overtones or combinations in the near-infrared. Rotational energy levels contribute finer structure, particularly in gaseous samples, but are less prominent in solid materials due to broadening effects. These mechanisms collectively create unique patterns because each material's atomic and molecular structure dictates specific transition energies, making spectral signatures akin to molecular fingerprints.^[8]^[9]^[10] Kirchhoff's law of thermal radiation establishes a fundamental link between absorption, emission, and reflectance for bodies in thermal equilibrium. For opaque bodies, where transmission is negligible, the law states that spectral emissivity \epsilon(\lambda) equals spectral absorptivity \alpha(\lambda), and since \alpha(\lambda) = 1 - r(\lambda) with r(\lambda) as reflectance, it follows that \epsilon(\lambda) = 1 - r(\lambda). This equivalence implies that materials poor at absorbing radiation at a given wavelength are also poor emitters, while strong absorbers are strong emitters, unifying the physical basis for both reflected and emitted spectral signatures across wavelengths. The law holds for thermal radiation but extends conceptually to non-thermal contexts in remote sensing, where it underpins the interpretation of surface properties from measured spectra.^[11]^[12] The composition of a material determines its diagnostic absorption features, as specific chemical bonds and elements selectively interact with certain wavelengths. In vegetation, chlorophyll pigments exhibit strong absorption in the blue (400–500 nm) and red (600–700 nm) regions due to electronic transitions in porphyrin rings, creating a sharp reflectance increase—the "red edge"—around 700 nm where near-infrared scattering dominates. Minerals display analogous features; for instance, iron oxides like hematite show electronic transitions causing broad absorptions near 870 nm, while hydroxyl-bearing silicates produce vibrational overtones around 1.4 and 1.9 \mum from O-H and Si-O bonds. Organic pigments in soils or paints similarly yield unique bands, such as carotenoid absorptions in the blue-green range, enabling material discrimination based on molecular makeup.^[13]^[14]^[15] Beyond composition, physical properties like particle size and surface texture modulate spectral signatures through scattering effects. Smaller particles enhance forward scattering and broaden absorption bands due to increased surface area and Mie scattering in the visible-near-infrared, altering the overall reflectance curve compared to larger grains. Surface texture influences the angular distribution of reflected light: smooth, diffuse surfaces approximate Lambertian reflectance, where intensity is independent of viewing angle, while rough or specular textures produce non-Lambertian behavior, with hotspots or directional variations that can shift apparent spectral features in directional measurements. These factors complicate but enrich the uniqueness of signatures, as they overlay compositional signals with structural information.^[16]^[17]

Characteristics

Reflectance Spectra

The reflectance spectrum quantifies the proportion of incident electromagnetic radiation reflected by a surface across different wavelengths, serving as a fundamental component of spectral signatures in passive remote sensing. It is typically measured in the visible-near infrared to short-wave infrared (VNIR-SWIR) range, from approximately 0.4 to 2.5 μm, where solar illumination provides sufficient energy for detection without thermal emission dominance.^[13]^[18] This spectrum is expressed as the ratio of reflected to incident flux, revealing material-specific patterns that enable identification of surface compositions in non-thermal wavelengths.^[19] Key features of reflectance spectra include distinct absorption bands, where energy is selectively absorbed by molecular vibrations or electronic transitions, leading to troughs in the curve; continuum trends, which outline the baseline shape influenced by scattering and overall albedo; and slope changes, marking abrupt shifts such as the "red edge" transition from visible to NIR regions. For instance, liquid water exhibits a prominent absorption band near 0.97 μm due to O-H stretching overtones, which deepens with increasing water content and affects spectra of hydrated surfaces.^[20]^[18] These elements collectively form unique signatures, with absorption depths and positions varying by material chemistry and physical structure. Representative examples illustrate the diversity of reflectance spectra. Healthy vegetation displays low reflectance (5-20%) in the visible range (0.4-0.7 μm) due to strong chlorophyll absorption, followed by a steep rise to high NIR reflectance (40-60%) from 0.7 to 1.3 μm, attributed to multiple internal reflections within air-cell interfaces in leaf mesophyll.^[21]^[18] In soils, clay minerals like kaolinite produce diagnostic absorption bands around 2.2 μm from Al-OH combinations, with deeper features indicating higher clay content and influencing soil classification in arid or exposed terrains.^[18] Urban materials, such as concrete, typically show a relatively flat, high-reflectance profile (20-40%) across the VNIR with minimal absorption features, gradually declining in the SWIR due to minor hydration effects, making them distinguishable from vegetated or mineral-rich surfaces.^[18] The shape and amplitude of reflectance spectra are modulated by external factors, including illumination angle, which alters the incident energy distribution, and viewing geometry, affecting the observed scattering paths. These variations are encapsulated by the bidirectional reflectance distribution function (BRDF), a model that parameterizes reflectance dependence on solar and sensor directions to normalize observations across different acquisition conditions.^[22] For example, forward-scattering surfaces like vegetation exhibit higher reflectance at off-nadir views, while backscattering materials like soils show angular sensitivity primarily in the NIR.^[23]

Emission Spectra

The emission spectrum of a material quantifies the intensity of electromagnetic radiation it emits as a function of wavelength, primarily observed in the thermal infrared region (approximately 3–15 μm) for remote sensing applications. This emitted radiation arises from the thermal energy of the material and follows Planck's law, which describes the spectral radiance B(\lambda, T) of a blackbody at temperature T as

B(\lambda, T) = \frac{2hc^2}{\lambda^5} \frac{1}{e^{hc / \lambda kT} - 1},

where h is Planck's constant, c is the speed of light, k is Boltzmann's constant, and \lambda is the wavelength.^[24] For non-ideal materials, the actual emitted radiance is scaled by the spectral emissivity \varepsilon(\lambda), defined as the ratio of the energy radiated by the material to that of a blackbody under identical conditions: \varepsilon(\lambda) = L(\lambda) / B(\lambda, T), where L(\lambda) is the material's emitted radiance. Emissivity values range from 0 to 1, with variations reflecting the material's surface properties and molecular structure.^[25] A prominent feature in emission spectra of many minerals and rocks is the Reststrahlen bands, regions of low emissivity (high reflectivity) caused by resonant vibrations of molecular bonds in the mid- to long-wave infrared. For instance, quartz exhibits Reststrahlen bands around 9–10 μm due to Si–O stretching vibrations, resulting in sharp emissivity minima that enable mineral discrimination in thermal remote sensing.^[26] Similarly, feldspar minerals display characteristic emission features in the 8–12 μm range, attributed to Si–O–Si bond vibrations, which produce distinct spectral contrasts useful for geological mapping.^[27] Water bodies, in contrast, show a broad emission peak with high emissivity (typically 0.96–0.98) across 8–14 μm, reflecting their strong absorption and uniform thermal response, which aids in distinguishing aquatic from terrestrial surfaces. Atmospheric gases like CO₂ contribute absorption-emission features at approximately 15 μm, influencing downward and upward radiative transfer in the troposphere and enabling remote sensing of trace gas concentrations.^[28] Emission spectra are inherently temperature-dependent, with the peak emission wavelength shifting according to Wien's displacement law: \lambda_{\max} = \frac{2898}{T} μm·K, where T is in Kelvin. This relation implies that cooler surfaces (e.g., Earth's ambient temperatures around 300 K) peak near 10 μm, while hotter sources shift toward shorter wavelengths, affecting the detectability of spectral features in remote sensing data.^[24] Such shifts underscore the need to account for thermal variations when interpreting emission signatures for material identification.

Measurement Techniques

Ground-Based Methods

Ground-based methods for acquiring spectral signatures involve direct, in-situ measurements using portable instruments positioned on or near the Earth's surface to capture high-fidelity data from targets such as rocks, soils, vegetation, and minerals. These techniques emphasize close-range observations, typically within meters, to obtain detailed reflectance or radiance spectra without the distortions introduced by distance or elevation. Primary tools include handheld spectrometers and spectroradiometers, such as the ASD FieldSpec series from Malvern Panalytical, which operate across the visible, near-infrared, and short-wave infrared regions from approximately 350 nm to 2500 nm.^[30]^[31] These devices achieve spectral resolutions of 3 nm in the VNIR (around 700 nm) and 8 nm in the SWIR (around 1000 nm and 2200 nm), enabling the detection of fine absorption features critical for material identification.^[31] The measurement procedure begins with field calibration to ensure accuracy, typically using a calibrated white reference panel made of Spectralon, a diffuse reflector with known reflectance properties close to 100% across the measured wavelengths.^[32] Operators first collect a dark current measurement by covering the foreoptic to subtract instrument noise, followed by a white reference scan under the same illumination conditions as the target.^[33] Measurements can be taken in contact mode, using a specialized probe that presses directly against the sample and incorporates an internal light source to exclude ambient light interference, ideal for solid materials like rocks or powders.^[34] Alternatively, non-contact mode employs a foreoptic attachment, such as a 25-degree field-of-view lens, positioned 0.5 to 1 meter from the target to measure larger areas like vegetation canopies or soil surfaces under natural or artificial illumination.^[34] To enhance signal-to-noise ratio, multiple scans—often 25 to 50—are averaged for the dark current, white reference, and sample spectra, with integration times adjusted based on ambient light levels.^[33]^[35] These methods offer distinct advantages, including spectral resolutions as fine as 1-10 nm, which allow for precise characterization of narrow absorption bands not resolvable in coarser remote sensing data.^[31] Additionally, the proximity of the instrument to the target minimizes atmospheric interference, such as absorption by water vapor or aerosols, providing near-laboratory-quality data under field conditions.^[36] Historically, ground-based spectral measurements gained prominence in the 1970s through efforts by the U.S. Geological Survey (USGS), where researchers like G.R. Hunt and J.W. Salisbury compiled foundational reflectance spectra of minerals and rocks to support mineral mapping initiatives.^[37]^[38] Their work, including systematic surveys in the visible and near-infrared ranges, established spectral libraries that remain integral to geological applications. In modern developments, portable hyperspectral cameras, such as those from Specim, extend these capabilities by integrating imaging with spectroscopy for ground-based acquisition of spatial-spectral data cubes in the field.^[39] These instruments complement remote sensing platforms by providing validation data at finer scales.^[39]

Remote Sensing Platforms

Remote sensing platforms enable the acquisition of spectral signatures over extensive areas, facilitating scalable monitoring of Earth's surface from aerial and orbital vantage points. These systems, mounted on aircraft and satellites, capture multispectral and hyperspectral data to discern material properties based on their unique reflectance and emission patterns across electromagnetic wavelengths. Unlike localized ground measurements, such platforms provide synoptic views, supporting applications in large-scale environmental and geological assessments by balancing spatial coverage with spectral detail.^[40] Airborne systems, such as the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) developed by NASA, exemplify early advancements in hyperspectral remote sensing. Deployed on high-altitude aircraft like the ER-2 jet, AVIRIS acquires calibrated radiance images in 224 contiguous spectral bands spanning 400 to 2500 nm, with a spectral sampling interval of approximately 10 nm. This configuration allows for detailed spectral signature capture over targeted regions, achieving spatial resolutions that vary with flight altitude, such as approximately 4 m at lower altitudes (e.g., 4 km with the Twin Otter) or 20 m at 20 km (ER-2 jet), though coverage is limited by flight paths and duration. AVIRIS has been instrumental in validating spaceborne data and mapping vegetation and mineral distributions since its first flights in the early 1990s.^[41]^[42] Spaceborne platforms extend this capability globally, with satellites hosting both multispectral and hyperspectral sensors. The Landsat series, initiated by NASA in the 1970s, represents foundational multispectral systems, such as Landsat 8, which images in 11 bands across visible, near-infrared, and thermal wavelengths at a 30-meter spatial resolution for most bands, a 185 km swath width, and a 16-day revisit time. These parameters enable consistent, repeat coverage of continental scales, though with coarser spectral resolution compared to hyperspectral instruments.^[43]^[44] Pioneering hyperspectral spaceborne efforts include NASA's Hyperion instrument on the Earth Observing-1 (EO-1) satellite, launched in November 2000 and operated until March 2017, which provided 220 contiguous bands from 400 to 2500 nm at 30-meter resolution over a 7.5 km swath, yielding valuable data for mineral and vegetation mapping. More recent missions, like Italy's PRISMA satellite launched in March 2019, advance this with over 240 bands in the 400-2500 nm range, 30-meter resolution, a 30 km swath, and a 29-day revisit cycle, enhancing global hyperspectral data availability for precision agriculture and disaster response.^[45]^[46] The evolution of these platforms traces from multispectral sensors in the 1970s, driven by Landsat's launch in 1972 to support broad land-use monitoring, to hyperspectral systems emerging in the 1990s with airborne prototypes like AVIRIS, culminating in orbital demonstrations by the early 2000s. This progression reflects improvements in sensor technology, from broad-band filters to narrow contiguous channels, enabling finer discrimination of spectral signatures. Upcoming missions, such as Germany's EnMAP launched in April 2022, continue this trend with 242 bands across 420-2450 nm, 30-meter resolution, a 30 km swath, and flexible revisit times down to 4 days for targeted areas, promising enhanced scalability for environmental analysis. As of 2025, recent launches include India's Pixxel Firefly satellites in January 2025 and Pakistan's HS-1 in October 2025, further expanding access to hyperspectral data for environmental and resource applications.^[47]^[48] Key operational parameters—spatial resolution, swath width, and revisit time—trade off to optimize coverage versus detail; for instance, Landsat's wider swath suits regional surveys, while hyperspectral satellites like PRISMA prioritize spectral fidelity for specific features.^[49]^[50]

Applications

Environmental Monitoring

Spectral signatures play a crucial role in environmental monitoring by enabling the detection of ecological changes through the analysis of reflectance and emission patterns across electromagnetic wavelengths. In vegetation health assessment, the Normalized Difference Vegetation Index (NDVI), calculated as (NIR - Red)/(NIR + Red), leverages the strong absorption of red light by chlorophyll in healthy plants and high reflectance in the near-infrared (NIR) band due to leaf cell structure.^[51]^[52] This index quantifies vegetation vigor, with values typically ranging from 0.2 to 0.8 for dense, healthy canopies, and has been widely applied to monitor stress conditions such as drought, where reduced chlorophyll content leads to decreased red absorption and thus lower NDVI values.^[53] For instance, during drought events, NDVI drops signal early physiological stress in crops and forests before visible symptoms appear, allowing for timely interventions in precision agriculture and ecosystem management.^[53] In water quality monitoring, spectral signatures facilitate the identification of pollutants and biological indicators, particularly in inland and coastal waters. Algal blooms, often dominated by cyanobacteria, are detected via the distinct absorption feature of phycocyanin pigment at approximately 620 nm in the red-orange spectrum, which differentiates cyanobacterial biomass from other phytoplankton.^[54] This absorption trough enables remote sensing algorithms to map bloom extent and intensity, supporting early warning systems for harmful algal blooms (HABs) that can release toxins and disrupt aquatic ecosystems.^[55] Turbidity, arising from suspended sediments, is assessed through increased backscattering of light in the visible and NIR bands, where higher particle concentrations elevate reflectance and reduce water clarity, providing a proxy for erosion and pollution levels in rivers and lakes.^[56] Land cover mapping relies on spectral libraries—comprehensive collections of reference signatures for various surface types—to classify and track changes in ecosystems like forests and wetlands. These libraries, such as the USGS Spectral Library, contain calibrated reflectance spectra for vegetation, soils, and water bodies, allowing multispectral or hyperspectral sensors to distinguish between land covers based on unique absorption and reflection patterns.^[57] For example, in the Sahel region of West Africa, monitoring desertification since the 1980s has utilized AVHRR satellite data and NDVI-derived signatures to quantify vegetation loss from overgrazing and drought, revealing shifts from woody savannas to barren lands over decades.^[58] Such applications have informed restoration efforts by mapping wetland shrinkage and forest fragmentation, highlighting areas vulnerable to climate-induced degradation.^[57] A prominent case study involves the MODIS instrument on NASA's Terra and Aqua satellites, which detects global fire activity through thermal anomalies in the mid-infrared (4 μm) and thermal infrared (11 μm) bands, where active fires exhibit elevated brightness temperatures compared to surrounding surfaces.^[59] The enhanced contextual algorithm processes these signatures to identify fire pixels while minimizing false alarms from sun glint or industrial sources, enabling near-real-time mapping of burned areas and smoke plumes that affect air quality and biodiversity.^[59] This capability has supported international fire management since 2003, tracking events like widespread Sahelian bushfires and their role in carbon emissions.^[59]

Geological and Mineral Identification

Spectral signatures play a crucial role in geological and mineral identification by enabling the detection of specific absorption features associated with rock-forming minerals and alteration products. For instance, iron oxides exhibit a prominent crystal field absorption band centered around 0.9 μm in the visible-near infrared (VNIR) region, arising from electronic transitions in ferric iron (Fe³⁺), which is commonly used to map iron-rich ore deposits during prospecting efforts.^[60] Similarly, carbonate minerals display a characteristic absorption feature near 2.3 μm due to overtone vibrations of C-O bonds, facilitating the identification of limestone, dolomite, and related sedimentary rocks in lithological mapping for resource exploration.^[61] These diagnostic bands allow hyperspectral remote sensing to distinguish mineral assemblages over large areas, supporting targeted drilling in mining operations.^[62] In hydrothermal alteration mapping, spectral signatures of clay minerals provide indicators of fluid-rock interactions linked to ore formation. Kaolinite, a common alteration product, shows a distinctive doublet absorption in the shortwave infrared (SWIR) around 2.17 and 2.21 μm (often approximated as 2.2 μm), resulting from Al-OH stretching and bending overtones, which signals silicification and argillization zones.^[63] This feature has been applied to delineate alteration halos in Carlin-type gold deposits, such as those in Nevada's Carlin trend, where hyperspectral imaging identifies kaolinite and associated clays along fault structures to vector toward hidden mineralization.^[64]^[65] Such mapping enhances exploration efficiency by highlighting prospective areas without extensive ground surveys. Spectral analysis also aids in assessing soil composition, particularly through bands sensitive to organic matter content, which influences soil stability and agricultural productivity. Organic matter in soils produces absorption features near 1.7 μm from C-H stretching overtones, allowing quantification of soil organic carbon levels via near-infrared spectroscopy.^[66] These measurements support erosion prediction models in agriculture, as lower organic matter correlates with increased soil erodibility and vulnerability to degradation under cropping practices.^[67] A notable case study is the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) aboard the Mars Reconnaissance Orbiter, operational since 2006, which uses hyperspectral data from 0.362 to 3.92 μm to identify planetary mineralogy. CRISM detects signatures of iron oxides, carbonates, and clays on Mars' surface, mapping ancient hydrothermal systems and aiding in the reconstruction of the planet's geological history.^[68]^[69]

Analysis Methods

Spectral Matching Algorithms

Spectral matching algorithms are essential for comparing observed hyperspectral spectra against reference spectral libraries to identify and classify materials, particularly in remote sensing applications where precise spectral similarity assessment is required. These methods quantify the resemblance between target and reference spectra, often treating them as vectors in high-dimensional space, to enable automated material detection while accounting for variations in illumination and noise. Common approaches include geometric, statistical, and encoding-based techniques, with performance evaluated through metrics like overall accuracy in classification tasks.^[70] One widely adopted method is the Spectral Angle Mapper (SAM), which measures the angular difference between two spectral vectors to assess similarity, making it insensitive to changes in illumination intensity. Introduced in 1993, SAM treats spectra as directions in n-dimensional space and computes the angle θ as follows:

\theta = \cos^{-1} \left( \frac{\mathbf{t} \cdot \mathbf{r}}{||\mathbf{t}|| \ ||\mathbf{r}||} \right)

where \mathbf{t} is the target spectrum vector, \mathbf{r} is the reference spectrum vector, \cdot denotes the dot product, and || \cdot || is the Euclidean norm. Smaller angles indicate higher similarity, allowing thresholds to classify pixels; this approach has been foundational for hyperspectral analysis since its development for imaging spectrometer data.^[71] Binary encoding provides a computationally efficient alternative for spectral matching by converting continuous spectra into binary bit patterns based on spectral features, such as absorption bands or shape summaries, enabling rapid cross-correlation with library spectra. Developed in the early 1990s for imaging spectrometry systems, this technique compresses data into discrete codes (e.g., 1 for presence of a feature, 0 otherwise) to facilitate fast searching and matching, particularly useful for large datasets.^[72] Since the late 1990s, machine learning techniques, particularly supervised classifiers like Support Vector Machines (SVM), have been integrated with spectral features for enhanced matching and classification of hyperspectral signatures. SVMs map spectral vectors into higher-dimensional spaces using kernel functions to separate classes with maximum margins, effectively handling the high dimensionality and nonlinearity of hyperspectral data. This integration allows for robust discrimination of subtle spectral differences, outperforming traditional methods in scenarios with limited training samples.^[73] Performance of these algorithms is often assessed using metrics such as overall accuracy (OA), producer's accuracy (PA), and user's accuracy (UA) in mixed-pixel scenarios, where sub-pixel variability challenges pure spectral matching. These metrics underscore the algorithms' effectiveness, particularly when preprocessing steps like atmospheric correction are applied to mitigate noise.

Hyperspectral Data Processing

Hyperspectral data processing encompasses a series of preprocessing steps designed to transform raw sensor data into reliable, physically meaningful spectral signatures suitable for extraction and analysis. This pipeline addresses inherent distortions and external influences in the acquired data, ensuring accuracy in subsequent applications such as material identification. Key stages include radiometric calibration, atmospheric correction, geometric correction, and dimensionality reduction, each tailored to the high-dimensional nature of hyperspectral imagery. Radiometric calibration is the initial step, converting the digital numbers (DNs) recorded by the sensor into physical units of radiance or reflectance. This process utilizes sensor-specific coefficients derived from laboratory measurements or onboard calibrators to account for instrumental response variations across spectral bands. For instance, in the case of the GF5-02 hyperspectral imager, calibration coefficients are applied to map DNs to top-of-atmosphere (TOA) radiance, enabling quantitative analysis independent of sensor artifacts.^[74] Accurate radiometric calibration is essential, as uncorrected data can introduce biases in spectral feature detection.^[75] Following radiometric calibration, atmospheric correction removes the effects of scattering and absorption by atmospheric constituents, such as aerosols, water vapor, and gases, to retrieve surface reflectance spectra. Widely adopted models include the Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH), which employs MODTRAN radiative transfer simulations for moderate-resolution atmospheric compensation across visible to shortwave infrared wavelengths, and the Second Simulation of the Satellite Signal in the Solar Spectrum (6S), a vectorized radiative transfer code that handles vector effects like polarization for enhanced precision.^[76]^[77] These methods effectively mitigate path radiance and adjacency effects; for example, FLAASH has demonstrated significant reductions in water vapor absorption errors in hyperspectral datasets from airborne platforms. Without such corrections, spectral signatures can be shifted or attenuated, compromising signature fidelity. Geometric correction then aligns the processed data to a geographic reference frame through orthorectification, correcting for terrain relief, sensor orientation, and platform motion to produce map-projected images. In pushbroom hyperspectral sensors, this step also addresses optical distortions like the "smile" effect—spectral curvature causing wavelength shifts across the field of view—and the "keystone" effect—spatial misalignment of spectral lines due to varying slit projections.^[78] Orthorectification typically involves applying ground control points or digital elevation models, achieving sub-pixel accuracy in georeferencing for applications requiring precise spatial-spectral correspondence.^[79] These corrections ensure that spectral signatures remain tied to accurate locations, preventing misalignment errors that could reach several pixels in uncorrected airborne data. Finally, dimensionality reduction techniques such as principal component analysis (PCA) compress the high-bandwidth hyperspectral data while retaining essential variance for signature extraction. PCA transforms the original spectral bands into a smaller set of uncorrelated principal components, ordered by explained variance, often reducing hundreds of bands to 5–10 components that capture over 95% of the data's information.^[80] In hyperspectral classification tasks, this method has been shown to preserve spectral discriminability, with the first few components highlighting dominant features like vegetation indices or mineral absorptions, as demonstrated in early applications on AVIRIS data.^[81] By mitigating the curse of dimensionality, PCA facilitates efficient downstream processing, such as spectral matching, without significant loss of signature integrity.

Challenges and Limitations

Atmospheric Interference

The Earth's atmosphere distorts spectral signatures acquired by remote sensing instruments through scattering and absorption processes that modify the radiance reaching the sensor. Rayleigh scattering, which is more pronounced at shorter wavelengths such as those in the blue and visible spectrum, causes preferential attenuation of high-frequency light, leading to a hazy appearance in imagery and reduced contrast in spectral features. Absorption by atmospheric gases further alters signatures by selectively removing energy at specific wavelengths; for instance, water vapor (H₂O) exhibits strong absorption around 1.4 μm, carbon dioxide (CO₂) around 2.0 μm and 4.3 μm, and ozone (O₃) in the ultraviolet and near-visible regions, creating dark bands in transmission spectra that obscure underlying surface reflectances.^[82]^[83] Path radiance, an additive component of scattered light within the atmosphere, contributes to these distortions by introducing background illumination that dilutes the target signal, particularly in the visible bands where it reduces the dynamic range and spectral contrast of surface features. This effect is a combination of Rayleigh molecular scattering (varying as λ⁻⁴) and Mie scattering from aerosols, exacerbating haze in low-visibility conditions.^[84] These interferences degrade the signal-to-noise ratio (SNR) of remote sensing data, as the added atmospheric noise overwhelms the surface signal; for example, in hazy Landsat imagery, this complicates feature discrimination.^[85] Standard mitigation strategies include empirical line correction, which calibrates sensor radiance to surface reflectance using linear fits derived from in-situ measurements of spectrally stable ground targets, and physics-based approaches employing radiative transfer models like MODTRAN to simulate and subtract atmospheric contributions based on input parameters such as aerosol optical depth and gas concentrations.^[86]^[87]

Spectral Variability

Spectral variability refers to the inherent fluctuations in the spectral signatures of materials caused by intrinsic properties and environmental influences on the target itself, rather than external propagation effects. These variations can significantly alter the shape, amplitude, and position of reflectance curves across wavelengths, complicating identification and analysis in remote sensing applications. Factors such as moisture, scale, temporal dynamics, and illumination geometry contribute to this variability, often requiring adaptive modeling to accurately interpret signatures.^[88] Intrinsic factors like moisture content profoundly influence spectral signatures by modifying absorption and scattering properties within the material. For instance, increasing soil moisture content reduces overall spectral reflectance, particularly darkening the near-infrared (NIR) region (0.8–2.5 μm) due to water's strong absorption bands, which can mask underlying soil chromophores and alter feature depths by up to several percent in reflectance values.^[89] This effect is most pronounced in the 2–2.4 μm range, where sensitivity to moisture reaches correlations (R²) around 0.80, as demonstrated in laboratory and field studies using hyperspectral sensors like HYMAP.^[89] At the scale of remote sensing pixels, sub-pixel mixing introduces variability by combining multiple endmembers within a single pixel, resulting in averaged or composite spectral signatures that do not represent any pure material. In hyperspectral imagery, this mixing—driven by limited spatial resolution—leads to proportion indeterminacy and errors in abundance estimation, as the observed spectrum reflects fractional contributions from diverse surface components like vegetation, soil, and shadows.^[88] Advanced unmixing models, such as the generalized linear mixing model, address this by estimating pixel-specific endmember variations, improving accuracy over traditional averaged approaches.^[88] Temporal changes, particularly phenological shifts in vegetation, cause dynamic alterations in spectral signatures over seasons, with notable movements in the red edge position (680–760 nm). During growth phases, such as in winter wheat, the red edge peak shifts to longer wavelengths (e.g., from ~735 nm toward higher values until flowering) due to increased chlorophyll and leaf area index, enhancing NIR reflectance; conversely, senescence induces a blue shift and reduced amplitude as biomass declines.^[90] Similar patterns occur in crops like sugar beets, where seasonal vitality indicators like derivative peaks at 703 nm and 736 nm vary with development stages, enabling monitoring of phenological transitions.^[90] Illumination variability, primarily from changes in solar zenith angle (SZA), affects spectral signatures through the bidirectional reflectance distribution function (BRDF), which modulates reflectance based on incident and viewing geometry. As SZA increases from nadir (0°) to higher angles (e.g., 60°), diurnal patterns emerge, with vegetation indices like NDVI showing increases (up to several days' shift in phenology estimates) due to enhanced backscattering, while EVI decreases from forward scattering effects; these variations can alter reflectance by 0.02–0.06 across visible to NIR bands.^[91] BRDF modeling, such as using MODIS products, quantifies these impacts, revealing root mean square errors in phenology timing of 5–33 days depending on SZA normalization.^[91]

References

[1]
Spectral Signatures - Remote Sensing
Sep 17, 1999 · Remote sensing is the science and art of identifying, observing, and measuring an object without coming into direct contact with it.Missing: definition | Show results with:definition
[2]
ESA - Eduspace EN - Home - Spectral signatures
The radiation reflected as a function of the wavelength is called the spectral signature of the surface. Graphs of spectral signatures of water, soil and ...
[3]
Spectral Signature Definition | GIS Dictionary - Esri Support
The pattern of electromagnetic radiation that identifies a chemical or compound. Unique differences in the reflectance characteristics of materials.<|control11|><|separator|>
[4]
[PDF] Fundamentals of Remote Sensing - NASA Applied Sciences
You can look at the reflected wavelengths detected by a sensor and determine the type of material it reflected from. This is known as a spectral signature. • In ...
[5]
[PDF] 19690018782.pdf - NASA Technical Reports Server (NTRS)
Two aspects are planned and essential. 1. An immediate short-term, ground-based, and field-oriented research project over 5 years devoted to- spectral signature ...<|control11|><|separator|>
[6]
https://www.guilford.com/excerpts/campbell2_old_ch1.pdf
[7]
[PDF] Spectral reflectance - GovInfo
wavelength; spectral reflectance is the fraction of the incident radiant flux that is reflected as a function of wavelength. A related quantity is the ...
[8]
Chapter 3 Physical principles of optical remote sensing
Electronic transitions produce broad absorption features that require higher energy levels than do vibrational processes, and there- fore take place at ...
[9]
[PDF] SPECTRAL SIGNATURES - eGyanKosh
The absorption bands in the spectra of materials are produced by three fundamental physical processes, including electronic transitions, vibrational transitions ...
[10]
[PDF] Spectroscopy of Rocks and Minerals, and Principles of Spectroscopy
What causes absorption bands in the spectra of materials? There are two general processes: electronic and vibrational. Burns (1993) examines the details of ...
[11]
[PDF] CHAPTER 3 ABSORPTION, EMISSION, REFLECTION, AND ...
The reflectances for selected surfaces for the wavelengths of solar radiation are listed in Table 3.1. From Kirchhoff's Law we can also write ελ + rλ + τλ = 1 ,.
[12]
2 Radiation Fundamentals‣ Essential Radio Astronomy
... spectral power absorbed by the opaque body must equal the spectral power emitted, so. e(ν)=a(ν)=1-r(ν). (2.47). This is Kirchhoff's law for opaque bodies.Missing: reflectance | Show results with:reflectance<|control11|><|separator|>
[13]
Spectral Reflectance - an overview | ScienceDirect Topics
Chlorophyll pigments a and b selectively absorb blue (400–500 nm) and red (600–700 nm) wavelengths for photosynthesis. ... chemical composition and function (see ...
[14]
[PDF] The peak absorbance wavelength of photosynthetic pigments ...
Jun 16, 2021 · This sharp contrast in absorbance in the red and reflectance in the NIR is termed the “vegetation red edge” or simply the “red edge.” This ...
[15]
Reflectance Spectral Features and Significant Minerals in ... - MDPI
Under normal solar spectrum (300–2500 μm), several minerals have distinct absorption features because of electronic transition, vibrational overtones, charge ...
[16]
Effects of particle size on bidirectional reflectance factor ...
The bidirectional reflectance factor (BRF) is commonly used to study the structure of a particulate surface based on photometric measurements.
[17]
Top-of-atmosphere reflectance over homogeneous Lambertian and ...
Soil spectral reflectance is controlled by a number of factors, including soil particle size, texture, iron content, soil moisture, surface roughness, etc.<|control11|><|separator|>
[18]
https://speclab.cr.usgs.gov/spectral.lib.html
[19]
Spectroscopy Lab
### Summary of Reflectance Spectra from USGS Spectroscopy Lab
[20]
Visible, near‐infrared spectrometry for simultaneous assessment of ...
Sep 5, 2014 · ... spectra taking into account both the specific absorption band of liquid water at 0.97 µm and the reflectance continuum. 4.2.1 Background on ...
[21]
Physical and physiological basis for the reflectance of visible and ...
A plant leaf typically has a low reflectance in the visible spectral region because of strong absorption by chlorophylls, a relatively high reflectance in the ...
[22]
[PDF] A Bidirectional Reflectance Model of the Earth's Surface for ...
A surface bidirectional reflectance model has been developed for the correction of surface bidirectional effects in time series of satellite observations, ...<|control11|><|separator|>
[23]
Robust estimation of BRDF model parameters - ScienceDirect.com
Spaceborne and airborne remotely sensed data are observed from varying viewing zenith angles, which causes differences in the observed reflectance even though ...Missing: shape | Show results with:shape
[24]
[PDF] Lecture 6: Thermal Radiation
• Thermal remote sensing can distinguish between the sea-ice and “leads”. (open water between the sea-ice) -- useful for navigation and for science. Page 25 ...
[25]
Full article: Land surface emissivity retrieval from satellite data
Definition of emissivity. The emissivity, ϵ, at a given wavelength λ (units, μm) and temperature T (units, K), is defined as the ...
[26]
[PDF] OBSERVATIONS OF SILICATE RESTSTRAHLEN BANDS IN ...
The mi d-infrared region, where thermal emission dominates the lunar spectrum, is potentially useful for mineralogical remote sensing, be- cause the molecular ...
[27]
Spectral analysis of synthetic quartzofeldspathic glasses using ...
Jun 2, 2010 · Si-O-Si bond vibrations within silicate materials produce a prominent absorption feature in the 8–12 μm region of thermal infrared (TIR) ...
[28]
[PDF] Thermal remote sensing of sea surface temperature - CentAUR
The thermal emission from the sea surface comes from a layer whose characteristic depth varies with wavelength (because the complex refractive index of water ...
[29]
Remote Sensing in the 15 µm CO2 Band: Key Concepts and ... - MDPI
We show that the measured 15 µm radiation constrains the population of excited CO2 vibrational levels and the 15 µm radiative flux divergence in the MLT, but ...
[30]
ASD FieldSpec | Field Portable Spectroradiometers
The gold standard field-portable UV/Vis/NIR/SWIR spectroradiometers for the fastest and most accurate spectral field measurements.FieldSpec 4 Hi-Res · FieldSpec 4 Standard-Res... · ASD FieldSpec 4 Hi-Res NG
[31]
ASD FieldSpec 4 Hi-Res spectroradiometer - Malvern Panalytical
The 3 nm VNIR, 8 nm SWIR spectral resolution of the ASD FieldSpec 4 Hi-Res spectroradiometer provides superior spectral performance across the full range solar ...
[32]
[PDF] User Manual
When using a calibrated white Spectralon panel as the white reference for a reflectance measurement with the instrument, an even closer reflectance value ...
[33]
ASD FieldSpec3 field measurement protocols
Sep 8, 2019 · The following protocol describes best practices for collecting field spectrometer measurements with an ASD Fieldspec 3 spectroradiometer.
[34]
ASD Instrument Spectroscopy Probes and Accessories
Sep 13, 2018 · Contact Probe. Designed for contact measurements (made using the ASD LabSpec, TerraSpec, or FieldSpec instruments) of solid raw materials ...
[35]
Making Accurate Field Spectral Reflectance Measurements
Mar 23, 2015 · In ASD's RS3 spectroscopy software, set scan average to 40; Set spectrum interval to five seconds, spectrum number to 120 for a 30 x 30 m pixel ...Missing: procedure | Show results with:procedure
[36]
“Ground–Aerial–Satellite” Atmospheric Correction Method Based on ...
Ground measurements conducted using analytical spectral devices (ASD) under rigorous observation protocols and calibration procedures can effectively eliminate ...
[37]
[PDF] USGS Spectral Library Version 7
Jul 23, 2016 · In order to map a mineral using imaging spectroscopy with a diagnostic absorption feature, reference spectra must have high signal-to-noise ...
[38]
Chapter 7 Development of Hyperspectral Imaging for Mineral ...
Introduction. THE PURPOSE of this paper is to provide a historical overview of the development of hyperspectral remote sensing for min-.
[39]
Hyperspectral Cameras, Sensors and Imagers - Specim
Sep 30, 2025 · We offer the broadest range of high-quality hyperspectral cameras covering wavelengths from visible and near-infrared to the thermal range.
[40]
AVIRIS (Airborne Visible/Infrared Imaging Spectrometer) - eoPortal
Spectral range: 360 - 2500 nm with a total of 224 bands. AVIRIS is now an operational instrument with reliable radiometric and spectral calibrations. AVIRIS ...
[41]
AVIRIS - Airborne Visible / Infrared Imaging Spectrometer
May 22, 2024 · ... bands) with wavelengths from 400 to 2500 nanometers (nm). AVIRIS has been flown on four aircraft platforms: NASA's ER-2 jet, Twin Otter ...Missing: specifications | Show results with:specifications
[42]
Concept - AVIRIS - Airborne Visible / Infrared Imaging Spectrometer
May 22, 2024 · The AVIRIS instrument contains 224 different detectors, each with a wavelength sensitive range (also known as spectral bandwidth) of approximately 10 ...
[43]
Landsat 8
These two sensors provide seasonal coverage of the global landmass at a spatial resolution of 30 meters (visible, NIR, SWIR); 100 meters (thermal); and 15 ...Bands · Landsat 8 News · In-Band Spectral Response · Thermal Infrared Sensor
[44]
Landsat 8 | U.S. Geological Survey - USGS.gov
Landsat 8 images have 15-meter panchromatic and 30-meter multi-spectral spatial resolutions along a 185 km (115 mi) swath.Landsat 9 · Landsat 7 · Landsat Level-1 Processing... · Earth As Art 4
[45]
EO-1 (Earth Observing-1) - eoPortal
Built by Northrop Grumman (formerly TRW), Hyperion is a hyperspectral instrument that enables scientists to see the chemical constituents of Earth's surface in ...Spacecraft · Launch · Mission Status
[46]
[PDF] PRISMA, the Italian hyperspectral mission - UNOOSA
Jun 14, 2019 · PRISMA is an Italian hyperspectral mission that detects sunlight reflected from Earth in 240 bands, launched on March 22nd, 2019.
[47]
Three decades of hyperspectral remote sensing of the Earth
The development of hyperspectral ... This paper provides a review of multispectral and hyperspectral remote sensing data, products and applications in geology.
[48]
EnMAP
The Environmental Mapping and Analysis Program (EnMAP) is a German hyperspectral satellite mission that aims at monitoring and characterising the Earth's ...EnMAP-Box · EnMAP Instrument Planning · Data & Access · Geology and Soils
[49]
Monitoring vegetation systems in the Great Plains with ERTS
A method has been developed for quantitative measurement of vegetation conditions over broad regions using ERTS-1 MSS data.
[50]
Red and photographic infrared linear combinations for monitoring ...
These linear combinations of the red and photographic IR radiances can be employed to monitor the photosynthetically active biomass of plant canopies.
[51]
Evaluation of the monitoring capability of various vegetation indices ...
This research evaluated the suitability of 16 commonly used remote sensing vegetation indices for grassland drought monitoring.
[52]
A Machine Learning and Remote Sensing‐Based Model for Algae ...
Feb 28, 2024 · Phycocyanin absorbs strongly at 620 nm, while chlorophyll-a absorbs between 665 and 681 nm (Stumpf et al., 2016). Satellite imagery from several ...
[53]
Retrieval of phycocyanin concentration from remote-sensing ...
It is valuable to remotely estimate the PC concentrations in lakes for predicting algal blooms. In this paper, the characteristics of the specific absorption of ...
[54]
A review of remote sensing-based water quality monitoring in turbid ...
Aug 7, 2025 · Turbidity. Turbidity is a comprehensive indicator of water scattering ability and is linearly related to the backscattering coefficient.
[55]
Spectral Library | U.S. Geological Survey - USGS.gov
The spectral library was assembled to facilitate laboratory and field spectroscopy and remote sensing for identifying and mapping minerals, vegetation, and ...
[56]
30 years of remote sensing data and field observations (Mali, Niger)
Desertification of the Sahel region has been debated for decades, while the concept of a “re-greening” Sahel appeared with satellite remote sensing data ...
[57]
An Enhanced Contextual Fire Detection Algorithm for MODIS
Oct 15, 2003 · In this paper, we present an improved replacement detection algorithm that offers increased sensitivity to smaller, cooler fires as well as a significantly ...
[58]
[PDF] Spectral Variations in Rocks and Soils Containing Ferric Iron ...
All spectra are in continuum-removed format. Larger grain sizes show increased saturation of the crystal field absorption at 0.9 µm, as well as broadening ...Missing: prospecting | Show results with:prospecting
[59]
Spectral reflectance properties of carbonates from terrestrial ...
Carbonate detection and characterization in the 0.35–2.5-μm region is largely restricted to a single absorption feature in the 2.3-μm region, which can be ...
[60]
ASTER, ALI and Hyperion sensors data for lithological mapping and ...
Hyperion visible and near infrared bands (0.4 to 1.0 μm) and shortwave infrared bands (0.9 to 2.5 μm) allowed to produce image maps of iron oxide minerals, ...
[61]
Remote Detection of Clay Minerals - PMC - PubMed Central
Kaolinite spectra exhibit a sharp pair of bands due to the 2v(Al2OH) stretching overtone at 1.396 and 1.416 μm and a (v + δ)Al2OH doublet at 2.17 and 2.21 μm.
[62]
Spectral properties (0.4 to 25 microns) of selected rocks associated ...
Jan 10, 1986 · Eight characteristics of the disseminated deposits were proposed that could be identified by remote-sensing applications. ... Carlin gold deposit, ...
[63]
None
Summary of each segment:
[64]
REFLECTANCE SPECTROSCOPY OF ORGANIC MATTER IN ...
A number of spectral parameters were evaluated for organic absorption bands observed in the 34 mm wavelength ... The albedo was measured at 1.7 mm, the organic ...
[65]
Spatial and temporal evolution of soil organic matter and its ...
This study revealed that the loss of SOMC increases with soil erosion. Land use also affects change in SOM. The most severe decrease in SOM occurred when ...
[66]
Compact Reconnaissance Imaging Spectrometer for Mars (CRISM ...
May 30, 2007 · CRISM's objectives are (1) to map the entire surface using a subset of bands to characterize crustal mineralogy, (2) to map the mineralogy of ...Science Objectives · Instrument Requirements and... · Instrument Operations
[67]
CRISM Web Site
... Orbiter, which takes measurements of both the surface and the atmosphere of Mars. CRISM data are used to find minerals' spectral signatures on Mars.
[68]
Radiometric Calibration of GF5-02 Advanced Hyperspectral Imager ...
Apr 23, 2023 · Radiometric calibration is the process of converting the imagery digital number (DN) of remote sensors into physical quantities (e.g., radiance) ...
[69]
New approach for the radiometric calibration of spectral imaging ...
Simply stated, radiometric calibration relates the digital counts recorded by the sensor to the true physical units of the signal being sampled. Although ...
[70]
[PDF] ENVI Atmospheric Correction Module User's Guide
This is the user's guide for the Atmospheric Correction Module, version 4.7, for QUAC and FLAASH, which includes an introduction to these.
[71]
Comparison of atmospheric correction models: FLAASH and 6S ...
From the analysis conducted on the spectral response of atmospheric corrected image shows the 6S model has better accuracy for the spectral response from ...Missing: hyperspectral | Show results with:hyperspectral
[72]
[PDF] A Review of Atmospheric Correction Techniques for Hyperspectral ...
Atmospheric correction removes effects from gases and aerosols. Methods include empirical line, flat field, and radiative transfer modeling. ATREM is also ...<|control11|><|separator|>
[73]
[PDF] Preprocessing of hyperspectral imagery with consideration of smile ...
Smile correction should be built into radiometric correction. Furthermore, keystone correction should be processed together with geometric correction. Smile.
[74]
The EnMAP imaging spectroscopy mission towards operations
Aug 15, 2023 · Spectral characterization and smile correction for the imaging spectroscopy mission EnMAP. Proceedings of 2018 IEEE International Geoscience ...<|control11|><|separator|>
[75]
Dimensionality reduction in hyperspectral imaging using standard ...
Oct 3, 2025 · Among the most commonly used dimensionality reduction methods are Principal Component Analysis (PCA), Independent Component Analysis (ICA) ...
[76]
[PDF] Principal Component Analysis for Hyperspectral Image Classification
The contents of the PCA bands are then examined, fol- lowed by a comparison of the classification results obtained using the PCA bands and the original.
[77]
[PDF] Satellite sensor types
Atmospheric transmission windows are separated by absorption bands. ... water molecule: 2v3 near 1.4 µm and 2 + 3 near. 1.9 μm. (From Hunt, 1977 ...
[78]
Effects of Atmosphere - CRISP
Atmospheric absorption affects mainly the visible and infrared bands. Optical remote sensing depends on solar radiation as the source of illumination.
[79]
Path Radiance - an overview | ScienceDirect Topics
The path radiance term is a combination of molecular Rayleigh scattering, which varies with wavelength as λ−4, and aerosol and particulate Mie scattering.
[80]
[PDF] Haze Effects on Satellite Remote Sensing Imagery and their ...
Haze causes visibility to drop and therefore affecting the images acquired for this area using optical sensor such as that onboard the. Landsat-5 TM satellite.
[81]
SNR (Signal-To-Noise Ratio) Impact on Water Constituent Retrieval ...
The objective of this paper is to assess the impact of the SNR on the Chl-a and TSS algorithms in four lakes located at Mamirauá Sustainable Development ...
[82]
The use of the empirical line method to calibrate remotely sensed ...
The empirical line method offers a logistically simple means of generating acceptable estimates of surface reflectance.
[83]
MODTRAN - Radiative Transfer Software - Spectral Sciences
The MODTRAN software computes line-of-sight (LOS) atmospheric spectral transmittances and radiances over the ultraviolet through long wavelength infrared ...
[84]
[PDF] Spectral variability in hyperspectral data unmixing - Hal IMT Atlantique
Jun 18, 2021 · Spectral variability is ubiquitous in hyperspectral images: the pixels in regions composed of a single material (e.g., tree, roof and soil in ...
[85]
Estimation of Soil Moisture Content from the Spectral Reflectance of ...
Angström [17] demonstrated through laboratory measurements that soil moisture content had an impact on the behavior of soil spectral reflectances in the solar ...
[86]
[PDF] red edge shift as vitality indicator for plants
The shift of the red edge in the reflection spectra of vegetation targets is a known phenomenon documenting changes in the biological status of plants.
[87]
Influence of Varying Solar Zenith Angles on Land Surface ... - MDPI
To examine the SZA effect on LSP, the MODIS bidirectional reflectance distribution function (BRDF) product and a BRDF model were employed to derive LSPs under ...