No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
The effect of contaminated snow reflectance using hyperspectral remote sensing – a review
Abstract
Snow is composed of small crystalline ice particles consisting of multitude of snowflakes that fall from clouds. This review paper highlights the scope and nature of the research work done in North West Himalayan region. Spectral signatures were collected for varying snow grain size, contamination, adjacency factors and other ambient objects. The retrieval of snow parameters such as grain size, contamination, spectral albedo using high resolution imaging data at different wavelength are discussed in this paper. Wavelengths 550, 1240 and 1660 nm are found to be useful wavelength for discriminating different snow feature. Spectral unmixing (SU) or the disintegration of individual spectra into a mixture of a small number of end members represents the spectra of pure and contaminated components. Many linear SU techniques exploit this notion in a way or another. For instance, where the pixel purity index algorithm projects the spectra of every pixel onto random vectors in spectral space, and tags the extremities. The spectra that got tagged the most is considered as end members. The N-findR algorithm searches for the largest volume simplex via an iterative procedure, and assigns the vertices of this simplex as end members. These algorithms of end member extraction are based on the assumption that spectra of pure pixel exist in data and form the extremes of a simplex embedded in the data cloud. But, in reality this is often not the case as with multiple scattering in wet environments, secondary reflections through vegetation canopies or between fuzzy surface materials. Nonlinear algorithm is made upon an assumption that the pixel reflectance results from nonlinear function of the abundance vectors associated with the pure spectra of snow with ambiguity of unknown spectral signatures of the pure snow and nonlinear function.
... Various researchers have observed that the reflectance recorded by remote sensors has a spectral mixing problem and offered analytical methodologies to address this issue. Thus, spectral unmixing remains difficult and requires further research [153][154][155]. ...
Toxic heavy metals in soil negatively impact soil’s physical, biological, and chemical characteristics, and also human wellbeing. The traditional approach of chemical analysis procedures for assessing soil toxicant element concentration is time-consuming and expensive. Due to accessibility, reliability, and rapidity at a high temporal and spatial resolution, hyperspectral remote sensing within the Vis-NIR region is an indispensable and widely used approach in today’s world for monitoring broad regions and controlling soil arsenic (As) pollution in agricultural land. This study investigates the effectiveness of hyperspectral reflectance approaches in different regions for assessing soil As pollutants, as well as a basic review of space-borne earth observation hyperspectral sensors. Multivariate and various regression models were developed to avoid collinearity and improve prediction capabilities using spectral bands with the perfect correlation coefficients to access the soil As contamination in previous studies. This review highlights some of the most significant factors to consider when developing a remote sensing approach for soil As contamination in the future, as well as the potential limits of employing spectroscopy data.
The field observation data of ice‐snow (the ice covered by snow) is scarce, causing a poor understanding of its characteristics and some uncertainty in ice process simulation. Using a multi‐angle observation instrument, we conducted a field investigation and measured the snow/ice/ice‐snow Bidirectional Reflectance Distribution Function. Both snow and ice have the typical feature of forwarding scattering, and the scattering intensity is governed by their properties (i.g., snow grain size, liquid water content, and transparency). The characteristics of ice‐snow are similar to snow when a small amount of fresh snow is deposited on the ice surface, but will change significantly when the surface snow is metamorphic (melt or refreezing). This change can affect the monitoring accuracy of snow and ice using remote sensing data. In this study, a Normalized Difference Ice‐Snow Index (NDISI) was developed using two bands (i.e., 1,080–1,120 nm and 1,760–1,800 nm). We applied spectra from different regions to verify the ability of NDISI and its application for distinguishing ice‐snow. This work should be helpful for the fine classification between snow, ice, and ice‐snow. It can support a more accurate simulation of ice phenology.
We propose a system of analytical equations to retrieve snow grain
size and absorption coefficient of pollutants from snow reflectance or snow
albedo measurements in the visible and near-infrared regions of the
electromagnetic spectrum, where snow single-scattering albedo is close to
1.0. It is assumed that ice grains and impurities (e.g., dust, black and
brown carbon) are externally mixed, and that the snow layer is semi-infinite and
vertically and horizontally homogeneous. The influence of close-packing
effects on reflected light intensity are assumed to be small and ignored. The
system of nonlinear equations is solved analytically under the assumption that
impurities have the spectral absorption coefficient, which obey the
Ångström power law, and the impurities influence the registered spectra
only in the visible and not in the near infrared (and vice versa for ice grains).
The theory is validated using spectral reflectance measurements and albedo of
clean and polluted snow at various locations (Antarctica Dome C, European
Alps). A technique to derive the snow albedo (plane and spherical) from
reflectance measurements at a fixed observation geometry is proposed. The
technique also enables the simulation of hyperspectral snow reflectance
measurements in the broad spectral range from ultraviolet to the
near infrared for a given snow surface if the actual
measurements are performed at a restricted number of wavelengths (two to four,
depending on the type of snow and the measurement system).
Snow has a very high reflection property when compared with all other natural surfaces on Earth; thus, a small amount of contamination (30 ng/g) can dramatically reduce the reflectance of snow. To quantify the effect of black carbon (BC) concentrations on the directional reflectance factors of snow, we deposited BC concentrations onto snow surfaces at natural levels (98-4095 ng/g) and compared the measurement results with those from a theoretical model. It was found that increasing BC concentrations decreased the reflectance factor of snow and changed its distribution pattern. Moreover, our data provided valuable verification of the snow reflection model, which has previously been used to characterize the reflectance of snow. The model did not well match our measurements; for example, the model found fewer anisotropic results than those from observations. Subsequently, a specular kernel combined with two free parameters was proposed to counter the forward and backward scattering of the optical property of snow. The improved model successfully characterized the observed variability in the reflection measurements of snow with different BC concentration levels under field conditions, and its inverted parameter (M) had the potential to estimate BC concentrations. The improved model also had the ability to simulate the spectral reflectance factor of snow with low BC concentrations (i.e., smaller than 118 ng/g) over a wide range of viewing zenith angles. This paper provides an additional and effective method for studying the angular and spectral reflection properties of snow.
Study of hyper-spectral behavior of snow is important to interpret, analyze and validate optical remote sensing observations. To map and understand response of snow-mixed pixels in RS data, field experiments were conducted for linear mixing of external materials (i.e. Vegetation, Soil) with snow, using spectral-radiometer (350-2500nm). Further, systematic non-linear mixing of snow contaminants (soil, coal, ash) in terms of size and concentration of contaminants is analyzed to imitate and understand spectral response of actual field scenarios. Sensitivity of band Indices along with absorption peak characteristics provide clues to discriminate the type of contaminants. SWIR region is found to be useful for discriminating size of external contaminants in snow e.g. Avalanche deposited snow from light contaminated forms. Present research provide inputs for mapping snow-mixed pixels in medium/coarse resolution remote sensing RS data (in terms of linear mixing) and suitable wavelength selections for identification and discriminating type/size of snow contaminants (in terms of non-linear mixing).
Snow is the part of atmosphere in the climate system of the Earth, and its physical parameters play an important role in hydrological and climate models. The present study concerns with the imaging spectroscopy to produce the snow cover maps and to estimate snow grain size in the NorthWestern Himalayan region. It is necessary to develop an approach to accurately map the snow cover, snow grain size spatially using advance remote sensing data and technique. Remote sensing techniques can provide spatial and temporal information of a large extent, economically and efficiently. In the present study, one of the important snow physical parameters (i.e., snow grain size) has been estimated using Spectral Angle Mapper (SAM) classification method and Grain Index (GI) method. The study has been carried out by using Hyperspectral EO-1 Hyperion sensor data of 12th January and 23rd January, 2016 to map the grain size of snow. The FLAASH (Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes) atmospheric correction model has been used to apply atmospheric correction from satellite images. The spectral reflectance of different types of snow grain size has been collected in the Hyperion image using spectral library. The important wavelengths are found for the retrieval of snow parameters, such as grain size. The snow cover maps were produced using Normalized Difference Snow Index (NDSI) technique. The snow maps were generated for dry snow, small grain size snow, medium grain size snow, large grain size snow and wet snow classes. This study is important for mapping of snow-cover characteristics which can provide valuable input for climatology, hydrology, and mountain hazard applications.
Deposited mineral dust and black carbon are known to reduce the albedo of snow and enhance melt. Here we estimate the contribution of anthropogenic black carbon (BC) to snowmelt in glacier accumulation zones of Central Asia based on in-situ measurements and modelling. Source apportionment suggests that more than 94% of the BC is emitted from mostly regional anthropogenic sources while the remaining contribution comes from natural biomass burning. Even though the annual deposition flux of mineral dust can be up to 20 times higher than that of BC, we find that anthropogenic BC causes the majority (60% on average) of snow darkening. This leads to summer snowmelt rate increases of up to 6.3% (7 cm a⁻¹) on glaciers in three different mountain environments in Kyrgyzstan, based on albedo reduction and snowmelt models.
A spectral reflectance investigation was carried out using radiometer in spectral range from 350 to 2500 nm. The objective was to understand the effect of contamination and mixed objects on the snow reflectance. The experiments were carried out at SASE Observatory Dhundi (HP). In snow and vegetation mixed area, peak of vegetation reflectance in near infrared region is possible to detect, when vegetation is 25 percent. As vegetation proportion increases reflectance curve follows pattern of vegetation and amount of reflectance reduces in visible region. In case of soil mixing, as proportion of soil increased reflectance of visible region reduced and reflectance increased in SWIR region. In case of snow-soil-vegetation mix, peak reflectance in NIR due to vegetation, high reflectance in SWIR region and overall high reflectance in visible and NIR region is possible to detect. This suggests, it may be possible to detect type and amount of mixing in snow pixel, if hyper spectral data is available. Influence of snow, soil and vegetation mixing was also studied on NDSI. It suggests that if proportion of vegetation and soil is less than 25 percent, then pixel can be identified as snow
This article presents highlights of the research work done in hyperspectral remote sensing in the Himalayan cryosphere. Hyperspectral radiometric investigations conducted at different field locations of NW Himalaya and cold laboratory are discussed. Spectral signatures were collected for varying snow grain size, contamination, liquid water content, vegetation/soilmixed snow, glacier ice, moraines and other ambient objects. The important wavelengths for snow applications are found to be 440, 550, 590, 660, 860, 1050, 1240 and 1650 nm. Further, the retrieval of snow parameters such as grain size, spectral albedo and snow contamination using imaging data at the above wavelength channels is discussed. Wavelengths 550, 1240 and 1660 nm are found to be useful for discriminating different glacier features. Limitations in hyperspectral remote sensing such as availability of imaging data, rugged topography and further research issues such as multi-sensor mapping and data fusion, multiangle measurements, 3D adjacency effect and improved algorithms for quantitative retrieval of contaminants are identified.
Hyperspectral endmember extraction is to estimate endmember signatures (or material spectra) from the hyperspectral data of an area for analyzing the materials and their composition therein. The presence of noise and outliers in the data poses a serious problem in endmember extraction. In this paper, we handle the noise- and outlier-contaminated data by a two-step approach. We first propose a robust-affine-set-fitting algorithm for joint dimension reduction and outlier removal. The idea is to find a contamination-free data-representative affine set from the corrupted data, while keeping the effects of outliers minimum, in the least squares error sense. Then, we devise two computationally efficient algorithms for extracting endmembers from the outlier-removed data. The two algorithms are established from a simplex volume max-min formulation which is recently proposed to cope with noisy scenarios. A robust algorithm, called worst case alternating volume maximization (WAVMAX), has been previously developed for the simplex volume max-min formulation but is computationally expensive to use. The two new algorithms employ a different kind of decoupled max-min partial optimizations, wherein the design emphasis is on low-complexity implementations. Some computer simulations and real data experiments demonstrate the efficacy, the computational efficiency, and the applicability of the proposed algorithms, in comparison with the WAVMAX algorithm and some benchmark endmember extraction algorithms.
We describe spectral reflectance measurements of snow containing the snow alga Chlamydomonas nivalis and a model to retrieve snow algal concentrations from airborne imaging spectrometer data. Because cells of C. nivalis absorb at specific wavelengths in regions indicative of carotenoids (astaxanthin esters, lutein, β-carotene) and chlorophylls
a and b, the spectral signature of snow containing C. nivalis is distinct from that of snow without algae. The spectral reflectance of snow containing C. nivalis is separable from that of snow without algae due to carotenoid absorption in the wavelength range from 0.4 to 0.58 μm and
chlorophyll a and babsorption in the wavelength range from 0.6 to 0.7 μm. The integral of the scaled chlorophyll a and b absorption feature (I
0.68) varies with algal concentration (Ca
). Using the relationshipCa
= 81019.2 I
0.68+ 845.2, we inverted Airborne Visible Infrared Imaging Spectrometer reflectance data collected in the Tioga Pass region of
the Sierra Nevada in California to determine algal concentration. For the 5.5-km2 region imaged, the mean algal concentration was 1,306 cells ml−1, the standard deviation was 1,740 cells ml−1, and the coefficient of variation was 1.33. The retrieved spatial distribution was consistent with observations made in the
field. From the spatial estimates of algal concentration, we calculated a total imaged algal biomass of 16.55 kg for the 0.495-km2 snow-covered area, which gave an areal biomass concentration of 0.033 g/m2.
Spectral unmixing is a crucial processing step when analyzing hyperspectral
data. In such analysis, most of the work in the literature relies on the widely
acknowledged linear mixing model to describe the observed pixels.
Unfortunately, this model has been shown to be of limited interest for specific
scenes, in particular when acquired over vegetated areas. Consequently, in the
past few years, several nonlinear mixing models have been introduced to take
nonlinear effects into account. These models have been proposed empirically,
however without any thorough validation. In this paper, the authors take
advantage of two sets of real and physical-based simulated data to validate the
accuracy of various nonlinear models in vegetated areas. These physics-based
and analysis models, and their corresponding unmixing algorithms, are evaluated
with respect to their ability of fitting the measured spectra and of providing
an accurate estimation of the abundance coefficients, considered as the spatial
distribution of the materials in each pixel.
Snow is a highly reflecting object found naturally on the Earth and its albedo is highly influenced by the amount and type
of contamination. In the present study, two major types of contaminants (soil and coal) have been used to understand their
effects on snow reflectance in the Himalayan region. These contaminants were used in two categories quantitatively – addition
in large quantity and addition in small quantity. Snow reflectance data were collected between 350 and 2500 nm spectral ranges
and binned at 10 nm interval by averaging. The experiment was designed to gather the field information in controlled conditions,
and radiometric observations were collected. First derivative, band absorption depth, asymmetry, percentage change in reflectance
and albedo in optical region were selected to identify and discriminate the type of contamination. Band absorption depth has
shown a subtle increasing pattern for soil contamination, however, it was significant for small amounts of coal contamination.
The absorption peak asymmetry was not significant for soil contamination but showed a nature towards left asymmetry for coal.
The width of absorption feature at 1025 nm was not significant for both the contaminations. The percentage change in reflectance
was quite high for small amount of coal contamination rather than soil contamination, however, a shift of peak was observed
in soil-contaminated snow which was not present in coal contamination. The albedo drops exponentially for coal contamination
rather than soil contamination.
The airborne ESA-APEX (Airborne Prism Experiment) hyperspectral mission simulator is described with its distinct specifications to provide high quality remote sensing data. The concept of an automatic calibration, performed in the Calibration Home Base (CHB) by using the Control Test Master (CTM), the In-Flight Calibration facility (IFC), quality flagging (QF) and specific processing in a dedicated Processing and Archiving Facility (PAF), and vicarious calibration experiments are presented. A preview on major applications and the corresponding development efforts to provide scientific data products up to level 2/3 to the user is presented for limnology, vegetation, aerosols, general classification routines and rapid mapping tasks. BRDF (Bidirectional Reflectance Distribution Function) issues are discussed and the spectral database SPECCHIO (Spectral Input/Output) introduced. The optical performance as well as the dedicated software utilities make APEX a state-of-the-art hyperspectral sensor, capable of (a) satisfying the needs of several research communities and (b) helping the understanding of the Earth’s complex mechanisms.
Reflectance data for contaminated and different grain-size snow were collected using a spectroradiometer ranging from 350 to 2500 nm. Contamination was predominantly due to soil. The radiometer data were binned at 10 nm intervals by averaging, and then principal component analysis, shape, size and strength of the absorption peak, first and second derivatives were computed, providing information about the effect of grain size and contamination on snow reflectance. Relative strength for contamination and grain size showed a distinct reverse pattern at 1025 nm after continuum removal. Band absorption depth at 1025 nm showed an increase with increasing snow grain size, whereas the band depth was found to decrease with increased soil contamination. The curve shape was right asymmetric and showed a change to left asymmetry with increase in contamination. The first derivative of reflectance in the visible region showed a shift of peak due to contamination. Soil contamination significantly reduced the albedo of snow at a low level of contamination but showed little influence at higher level. Relative strength, shape of curve and reflectance characteristics have shown the potential to identify the influence of contamination and grain-size based metamorphism using satellite-based hyperspectral remote sensing.
Satellite observations of snow in the near-infrared wavelengths can be
used to roughly estimate snow grain size. When the grain size is large,
it is possible to use measurements in the visible wavelengths to
estimate snow water equivalence below some threshold value of around 100
mm. While sufficient data to fully evaluate these possibilities are not
available, model calculations, selected satellite observations, and
limited ground truth are in qualitative agreement. Complications arise
because the effect of contamination by atmospheric aerosols is similar
to that of finite depth, and because the near-infrared channel on the
NOAA TIROS-N series satellites is not in the wavelength region where
snow reflectance is most sensitive to grain size.
It is a long-standing problem to find an appropriate small set of parameters that best describes snow micro structure as observed on sec tions or thin slices, for the purposes of relating physical properties of snow to its electromagnetic, thermal, or mechanical properties. The stereological parameters point density and intercept length (uniquely related to surface area per unit volume) are likely members of the minimal set of descriptors, but a measure of their distribution, such as the variance, is also needed. In addition a topological classifier of ice- or pore-connectivity, such as the first Betti number, is possibly useful to describe thermal or mechanical proper ties, but that parameter is only retrieved after tedious preparation and analysis of serial sections. For models of microwave emission or metamor- phism at weak temperature gradients a geometric description from orthog onal sections provides useful parameters.
To understand the effect of contamination and mixed objects on snow reflectance, field experiments were carried out in Indian Himalaya during the winters of 2004–2005 and 2005–2006. The observations suggest, as the amount of contamination increases, the peak of snow reflectance in the visible region shifts toward the higher wavelength range. For snow and soil cover mix, as the proportion of soil cover increases, reflectance increases in SWIR region. For snow and vegetation mix areas, a red edge can be observed if vegetation cover is more than 25%. A significant change in the reflectance is observed in the visible region for snow thickness below 5 cm. Normalized Difference Snow Index (NDSI) and S3 based snow indexes can map soil and ash contaminated snow. In snow–soil mixing (patchy snow), NDSI underestimates the snow cover area and S3 index overestimates. Both the snow indexes can map the snow under vegetation and mixed snow–soil–vegetation areas. Thin snow cover areas can also be mapped using both the indexes. This study will be helpful in developing new indexes and selection of threshold value for snow cover mapping.
In the summer of 2016, fire broke out in the forests of Uttarakhand state. The effect of forest fire was thought to have severe impact on nearby glaciated region in terms of faster melting. It is understood that contamination and heat enhance the snowmelt, which reduces the reflectance in different part of EM spectrum. In order to assess the effect, AWiFS and IMAGER data from Resourcesat-1 and INSAT-3D, respectively, for the months of April and May of the years 2012, 2013, 2014 and 2016 were used to compare the reflectance of snow. It was observed that pre- and post-fire data of 2016 show drop in reflectance in comparison with earlier years. The change in reflectance for locations near to forest fire was significantly high in comparison with previous year images, whereas far-away locations did not show much change. This drastic drop may be attributed to deposition of black carbon on nearby locations in snow-covered area. AWS data of nearby Joshimath observatory were also analysed to avoid any anomalous change in temperature for the same duration.
Radiative transfer theory based retrievals of effective snow grain size (SGS), albedo were widely explored using hyper-spectral sensors. However, investigation of linkage between diagnostic absorption peak characteristics (PCs) and snow surface parameters remained scarce, though these are significant to determine composition and state of material. In present study, field data acquired during three winter seasons was analyzed to study variability of effective SGS with PCs (i.e. relative depth, peak position and asymmetry). Effective SGS was found positively correlated with relative depth, with correlation of 90% and 86% for absorption peaks around 1.24 µm and 1.03 µm respectively. Asymmetry was found in good correlation with SGS (74%) at 1.24 µm than at 1.03 µm (21%). Further, PCs were derived using Hyperion data in ENVI software on pixel basis. Significance of PCs is described and compared with effective SGS, which was retrieved using widely accepted Asymptotic Radiative Transfer model. This work provide an alternate approach to infer state of snow metamorphism using peak characteristics, in scarcity of reference spectral database of snow classes.
Spatial information on snow wetness content(SWC) is important for hydrology, climatology applications. Limited work is available on estimation of SWC using optical sensors. In present work, spectral signature characteristics of snow (~145 samples) acquired in winters of three years, using field spectral-radiometer (350-2500nm) were correlated with synchronized SWC measurements. Correlation is found stronger in Near-Infra-Red(NIR) and Short-Wave-Infrared(SWIR) regions than Visible(VIS). Spectral peak width at 905nm and 1240nm is found negatively correlated with SWC, while positively correlated at 1025nm. Asymmetry tends towards right as SWC increases and has stable positive correlations as compared to other characteristics. Sensitivity of widely used snow related indices to SWC is also analysed. Based on analysis, new ratio method at selected wavelengths is proposed to discriminate dry and wet snow zones using air/ground borne sensors. Proposed methodology is evaluated on air-borne hyper-spectral(AVIRIS-NG) data and 88% overall accuracy with kappa coefficient 77.6 observed after validation with reference observations.
Cryosphere studies are very critical for understanding the behaviour of the global climate system. The cryosphere components such snow cover, glacier ice, ice sheets, icebergs, etc. need to be studied, from Earth’s energy and water balance point of view. However, their quantification and mapping is tedious on field using tradition survey methods. This review paper summarizes the major research work done in field of remote sensing based cryosphere studies in Indian Himalaya and Polar Regions. The traditional survey methods using terrestrial photogrammetry and aerial photography for snow and glacier studies are presented, initially. Then, remote sensing methods of snow and glacier mapping using visible, infrared and thermal spectral bands are presented. Later, the application of these techniques and method in Indian conditions is discussed. The advanced methods of microwave sensors based inversion models for snow physical properties retrieval are also explained. The separate sub-sections are given to highlight the use of remote sensing in glacier and polar studies. In each section, the current status and advances of the technology are discussed. The future prospects of remote sensing technology for each cryosphere theme are highlighted with emphasis to Indian scenario.
Although support vector machine (SVM) has shown powerful capability to support supervised classification for hyperspectral data, such model does not have the capability to define semantically the classification and reasoning rules which can potentially contribute to a more accurate supervised classification. In this article, we present an ontology-driven framework to support SVM-based hyperspectral data classification. First, we proposed a dimension reduction algorithm to allow automatic selection of prominent spectral characteristics for each land cover class included in a hyperspectral image. The result of the prominent spectral characteristics includes ranking and weights, used to signify the importance of a subset of all wavebands in distinguishing a particular land cover class from others. Then, we developed an ontology named HIC-Ontology to formally represent the extracted spectral characteristics to support the final training and classification process. The experimental results show that the proposed technique achieves better performance in classifying hyperspectral data than using the classic classification algorithm alone. We expect this work to contribute significantly in hyper-spectral image processing by introducing this knowledge-based approach.
Mainly due to the limited spatial resolution of the data acquisition devices, hyperspectral image pixels generally result from the mixture of several components that are present in the observed surface. Spectral mixture analysis (or spectral unmixing) is a key processing step which aims at identifying the spectral signatures of these materials and quantifying their spatial distribution over the image. The main purpose of this chapter is to introduce the spectral unmixing problem and to discuss some linear and nonlinear models and algorithms used to solve it. We will show that, capitalizing on several decades of methodological developments in the geoscience and remote sensing community, most of the unmixing algorithms proposed to unmix remotely sensed images can be directly applied in the chemometrics field to process hyperspectral data arising from various scanning microscopic techniques such as scanning transmission electron microscopy and Raman imaging.
Measurements of the dependence of snow albedo on wavelength, zenith angle, grain size, impurity content, and cloud cover can be interpreted in terms of single-scattering and multiple-scattering radiative transfer theory. Snow albedo is therefore much lower in the near IR. The near-IR solar irradiance thus plays an important role in snowmelt and in the energy balance at a snow surface.-from Author
With increasing emphasis being placed upon global measurements in an effort to understand global change, more sophisticated techniques must be developed to provide the kinds of information that are important to the development of predictive models. The definition of imaging spectrometry is the acquisition of image data in many contiguous spectral bands. The ultimate goal is to produce laboratory-like reflectance spectra for each pixel in an image. -from Author
Anomaly detection is an important task for hyperspectral data exploitation. A standard approach for anomaly detection in the literature is the method developed by Reed and Yu, also called RX algorithm. It implements the Mahalanobis distance, which has been widely used in hyperspectral imaging applications. A variation of this algorithm, known as kernel RX (KRX), consists of applying the same concept to a sliding window centered around each image pixel. KRX is computationally very expensive because, for every image pixel, a covariance matrix and its inverse has to be calculated. We develop an efficient implementation of the kernel RX algorithm. Our proposed approach makes use of linear algebra libraries and further develops a parallel implementation optimized for multi-core platforms, which is a well known, inexpensive and widely available high performance computing technology. Experimental results for two hyperspectral data sets are provided. The first one was collected by NASA’s airborne visible infra-red imaging spectrometer (AVIRIS) system over the World Trade Center (WTC) in New York, five days after the terrorist attacks, and the second one was collected by the hyperspectral digital image collection experiment (HYDICE). Our anomaly detection accuracy, evaluated using receiver operating characteristics (ROC) curves, indicates that KRX can significantly outperform the classic RX while achieving close to linear speedup in state-of-the-art multi-core platforms.
Published by World Metreological Organisation, Geneva, Switzerland
Given a mixed hyperspectral data set, linear unmixing aims at estimating the
reference spectral signatures composing the data - referred to as endmembers -
their abundance fractions and their number. In practice, the identified
endmembers can vary spectrally within a given image and can thus be construed
as variable instances of reference endmembers. Ignoring this variability
induces estimation errors that are propagated into the unmixing procedure. To
address this issue, endmember variability estimation consists of estimating the
reference spectral signatures from which the estimated endmembers have been
derived as well as their variability with respect to these references. This
paper introduces a new linear mixing model that explicitly accounts for spatial
and spectral endmember variabilities. The parameters of this model can be
estimated using an optimization algorithm based on the alternating direction
method of multipliers. The performance of the proposed unmixing method is
evaluated on synthetic and real data. A comparison with state-of-the-art
algorithms designed to model and estimate endmember variability allows the
interest of the proposed unmixing solution to be appreciated.
The adjacency effect is a well-known phenomenon of creating path interferences between the reflectances from different ground-cover materials. The effect is caused by atmospheric scattering, hence a typical approach to its detection has been the modeling of radiation transfer and spectral correspondence at particular wavelengths. In this paper, we investigate the detection of adjacency effect as being a general unmixing problem. This means that we opt to use unmixing to separate the true signature of a pixel from the background scatter reflected from its large neighborhood. Here, we concentrate on the prevalent linear mixing, and compare this with a specialized approach for detecting the adjacency effect in turbid waters surrounded by vegetation.
Snow is among the most "colorful" materials in nature, but most of the variability in snow reectance occurs beyond 0.8 µm rather than in the visible spectrum. In these wavelengths, reectance decreases dramatically as the snow grains evolve and grow, whereas in the visible spectrum snow reectance is degraded by contaminants such as dust, algae, and soot. From imaging spectrometer data, we can estimate the grain size of the snow in the surface layer, and thereby derive spectral and broadband albedo. We can also estimate the fraction of each pixel that is covered by snow, the liquid water content in the surface layer, and the amount of radiative forcing caused by absorbing impurities. Estimates of fractional snow-covered area and albedo dramatically improve the performance of spatially distributed snowmelt models that include net solar radiation as an input value, most signicantly in locations and at times where incident solar radiation is high and temperatures low. Experience with imaging spectrometer data has allowed extension of the fractional snow-cover and albedo estimates to multispectral sensors, particularly MODIS, the Moderate-Resolution Imaging Spectroradiometer.
Spectral unmixing is a very important task for remotely sensed hyperspectral data exploitation. It amounts at identifying a set of spectrally pure components (called endmembers) and their associated per-pixel coverage fractions (called abundances). A challenging problem in spectral unmixing is how to determine the number of endmembers in a given scene. Several automatic techniques exist for this purpose, including the virtual dimensionality (VD) concept or the hyperspectral signal identification by minimum error (HySime). Due to the complexity and high dimensionality of hyperspectral scenes, these techniques are computationally expensive. In this paper, we develop new fast implementations of VD and HySime using commodity graphics processing units. The proposed parallel implementations are validated in terms of accuracy and computational performance, showing significant speedups with regards to optimized serial implementations. The newly developed implementations are integrated in a fully operational unmixing chain which exhibits real-time performance with regards to the time that the hyperspectral instrument takes to collect the image data.
An algorithm is being developed to map global snow cover using Earth Observing System (EOS) Moderate Resolution Imaging Spectroradiometer (MODIS) data beginning at launch in 1998. As currently planned, digital maps will be produced that will provide daily, and perhaps maximum weekly, global snow cover at 500-m spatial resolution. It will also be possible to generate snow-cover maps at 250-m spatial resolution using MODIS data, and to study snow-cover characteristics. Preliminary validation activities of the prototype version of the snow-mapping algorithm, SNOMAP, have been undertaken. SNOMAP will use criteria tests and a decision rule to identify snow in each 500-m MODIS pixel. Use of SNOMAP on a previously mapped Landsat Thematic Mapper (TM) scene of the Sierra Nevada`s has shown that SNOMAP is 98% accurate in identifying snow in pixels that are snow covered by 60% or more. Results of a comparison of a SNOMAP classification with a supervised-classification technique on six other TM scenes show that SNOMAP and supervised-classification techniques agree to within about 11% or less for nearly cloud-free scenes and that SNOMAP provided more consistent results. About 10% of the snow cover, known to be present on the 14 March 1991 TM scene covering Glacier National Park in northern Montana, is obscured by dense forest cover. Mapping snow cover in areas of dense forests is a limitation in the use of this procedure for global snow-cover mapping. This limitation, and sources of error will be assessed globally as SNOMAP is refined and tested before and following the launch of MODIS.
The classification of remote sensing data based on the exploitation of spatial features extracted with morphological and attribute profiles has been recently gaining importance. With the development of efficient algorithms to construct the profiles for large datasets, such methods are becoming even more relevant. When dealing with hyperspectral imagery, the profiles are traditionally built on the first few principal components computed from the data. However, it needs to be determined if other feature reduction approaches are better suited to create base images for the profiles. In this article, we explore the use of profiles based on features derived from three supervised feature extraction techniques (i.e. Discriminant Analysis Feature Extraction, Decision Boundary Feature Extraction and Non-parametric Weighted Feature Extraction) and two unsupervised feature-extraction techniques (i.e. Principal Component Analysis (PCA) and Kernel PCA) in classification and compare the classification accuracies obtained by using different techniques for two different classification methods. The obtained results indicate significant improvements in the accuracies using the supervised feature extraction methods. However, the choice of the method affects the quality of the results for different datasets depending on the availability of the training samples.
The ENVISAT mission with a suite of high performance sensors offers some opportunities for mapping snow cover at regional and catchment scales. The spatial resolution of the Medium Resolution Imaging Spectrometer Instrument (MERIS) data and the spectral characteristics of the Advanced Along Track Scanning Radiometer (AATSR) data are suitable for these purposes. A new approach has been developed for the generation of snow cover products in Alpine regions, based on the combined use of ENVISAT optical data and topographic information. The Alpine region is selected as a test area to demonstrate the potential and the limitations of the novel approach. In particular, attention is focused on three regions of northern Italy (Valle d'Aosta, Piemonte, Lombardia). The first results obtained by the application of this new method to Earth Observation data will be presented and analysed.
Imaging spectrometry, or hyperspectral imaging as it is now called, has had a long history of development and measured acceptance by the scientific community. The impetus for the development of imaging spectrometry came in the 1970's from field spectral measurements in support of Landsat-1 data analysis. Progress required developments in electronics, computing and software throughout the 1980's and into the 1990's before a larger segment of the Earth observation community would embrace the technique. The hardware development took place at NASA/JPL beginning with the Airborne Imaging Spectrometer (AIS) in 1983. The airborne visible/infrared imaging spectrometer (AVIRIS) followed in 1987 and has proved to this day to be the prime provider of high-quality hyperspectral data for the scientific community. Other critical elements for the exploitation of this data source have been software, primarily ENVI, and field spectrometers such as those produced by Analytical Spectral Devices Inc. In addition, atmospheric correction algorithms have made it possible to reduce sensor radiance to spectral reflectance, the quantity required in all remote sensing applications. The applications cover the gambit of disciplines in Earth observations of the land and water. The further exploitation of hyperspectral imaging on a global basis awaits the launch of a high performance imaging spectrometer and more researchers with sufficient resources to take advantage of the vast information content inherent in the data.
Dense media radiative transfer (DMRT) equations based on quasicrystalline approximation (QCA) for densely distributed moderate size particles are developed. We first compute the effective propagation constant and coherent transmission into a dense medium on the basis of the generalized Lorentz-Lorenz law and the generalized Ewald-Oseen extinction theorem. The absorption coefficient of the dense media is then calculated. The distorted Born approximation is next applied to a thin layer to determine the bistatic scattering coefficients and the scattering coefficient. The phase matrix in DMRT is then obtained as bistatic scattering coefficient per unit volume. The model is applied to multiple sizes and for sticky particles. Numerical results are illustrated for extinction and brightness temperatures in passive remote sensing using typical parameters in snow terrain. The QCA-based DMRT is also used to compare with satellite Special Sensor Microwave Imager (SSM/I) brightness temperatures for four channels at 19 and 37 GHz with vertical and horizontal polarizations and for two snow seasons. It shows reasonable agreement to snow depth of 1 m.
The compact airborne spectrographic imager (casi) is a pushbroom imaging spectrograph intended for acquisition of VNIR multispectral imagery from light aircraft. An ongoing development program has resulted in improvements to the radiometric calibration procedures, and the capability for roll correction and geocorrection of imagery acquired with casi. A variety of monitoring and research missions have been undertaken for aquatic and terrestrial applications and development of remote sensing methodologies.
In contrast with traditional remote sensing, hyperspectral remote sensing has the characteristics of high spectral resolution, combination of graph and spectrum, and large quantity of imaging bands. It has been used in many fields. According to the research results of other scholars, building surface has become a thermal active surface which has an important effect for forming city three-dimensional climate. In the 863 research project of relieving and analyzing of city heat island, based on summarizing now available spectrum identification methods, we selected frequently-used spectral angle match (SAM) identification method for the research, then we realized a program of identifying building surface material with IDL, and we did a simulation experiment for the program by using an AVIRIS image. Finally, we got a good result. It is proved that identifying building surface material based on hyperspectral remote sensing is feasible. So it makes a solid basis for further research.
N-finder algorithm (N-FINDR) is probably one of most popular and widely used algorithms used for endmember extraction. Three major obstacles need to be overcome in its practical implementation. One is that the number of endmembers must be known a priori. A second one is the use of random initial endmembers to initialize N-FINDR, which results in inconsistent final results of extracted endmembers. A third one is its very expensive computational cost caused by an exhaustive search. While the first two issues can be resolved by a recently developed concept, virtual dimensionality (VD) and custom-designed initialization algorithms respectively, the third issue seems to remain challenging. This paper addresses the latter issue by re-designing N-FINDR which can generate one endmember at a time sequentially in a successive fashion to ease computational complexity. Such resulting algorithm is called SeQuential N-FINDR (SQ N-FINDR) as opposed to the original N-FINDR referred to as SiMultaneous N-FINDR (SM N-FINDR) which generates all endmembers simultaneously at once. Two variants of SQ N-FINDR can be further derived to reduce computational complexity. Interestingly, experimental results show that SQ N-FINDR can perform as well as SM-N-FINDR if initial endmembers are appropriately selected.
The end-to-end calibration plan for the Hyperion EO-1 hyperspectral payload is presented. The ground calibration is traceable to a set of three high quantum efficiency p-n silicon photodiode trap detectors the responsivities of which are traceable absolutely to solid state silicon diode physical laws. An independent crosscheck of the radiance of the Calibration Panel Assembly used to flood the Hyperion instrument in field and aperture was made with a transfer radiometer developed at TRW. On-orbit measurements of the sun's irradiance as it illuminates a painted panel inside the instrument cover are compared to the radiance scale developed during pre-flight calibration. In addition, an on-orbit calibration lamp source is observed to trace the pre-flight calibration constants determined on the ground to the solar calibration determination.
The author's introduction to remote sensing provides coverage of the subject irrespective of disciplines of study or the academic department in which remote sensing is taught. All the ''classical'' elements of aerial photographic interpretation and photogrammetry are described, but equal emphasis is placed on non-photographic sensing systems and the analysis of data from these systems using digital image processing procedures. This text includes coverage of image restoration, enhancement, classification, and data merging, and new sensor systems such as the Large Format Camera, solid-state linear arrays, the Shuttle Imaging radar systems, the Landsat Thematic Mapper, the SPOT satellite system, and the NOAA Advanced Very High Resolution Radiometer. Also covers imaging spectrometry and lidar systems. It contains extensive illustrations.
Snow cover information is an essential parameter for a wide variety of scientific studies and management applications, especially in snowmelt runoff modelling. Until now NOAA and IRS data were widely and effectively used for snow-covered area (SCA) estimation in several Himalayan basins. The suit of snow cover products produced from MODIS data had not previously been used in SCA estimation and snowmelt runoff modelling in any Himalayan basin. The present study was conducted with the aim of assessing the accuracy of MODIS, NOAA and IRS data in snow cover mapping under Himalayan conditions. The total SCA was estimated using these three datasets for 15 dates spread over 4 years. The results were compared with ground-based estimation of snow cover. A good agreement was observed between satellite-based estimation and ground-based estimation. The influence of aspect in SCA estimation was analysed for the three satellite datasets and it was observed that MODIS produced better results. Snow mapping accuracy with respect to elevation was tested and it was observed that at higher elevation MODIS sensed more snow and proved better at mapping snow under mountain shadow conditions. At lower elevation, IRS proved better in mapping patchy snow cover due to higher spatial resolution. The temporal resolution of MODIS and NOAA data is better than IRS data, which means that the chances of getting cloud-free scenes is higher. In addition, MODIS has an automated snow-mapping algorithm, which reduces the time and errors incorporated during processing satellite data manually. Considering all these factors, it was concluded that MODIS data could be effectively used for SCA estimation under Himalayan conditions, which is a vital parameter for snowmelt runoff estimation.
Examination of thematic mapper (TM) data has shown that water spectral reflectances observed at the top of the scattering atmosphere are contaminated not only by an intrinsic atmospheric reflectance but also by the reflectance of the surrounding green vegetated land surface when the water surface has a small size. This last effect is called the adjacency (or background, or environment) effect. A theoretical formalism has been used to explain the effects and is validated by being able to simulate the TM observed spectral reflectances. Further consequences on land surface classification are discussed, as well as the impact of the adjacency effect on the method of using water surface for atmospheric correction.
Measurements of the dependence of snow albedo on wavelength, zenith angle, grain size, impurity content, and cloud cover can be interpreted in terms of single-scattering and multiscattering radiative transfer theory. Ice is very weakly absorptive in the visible (minimum absorption at lambda = 0.46 micrometer) but has strong absorption bands in the near infrared (near IR). Snow albedo is therefore much lower in the near IR. The near-IR solar irradiance thus plays an important role in snowmelt and in the energy balance at a snow surface. The near-IR albedo is very sensitive to snow grain size and moderately sensitive to solar zenith angle. The visible albedo (for pure snow) is not sensitive to these parameters but is instead affected by snowpack thickness and parts-per- million amounts (or less) of impurities. Grain size normally increases as the snow ages, causing a reduction in albedo. If the grain increases as a function of depth, the albedo may suffer more reduction in the visible or in the near IR, depending on the rate of grain size increase. The presence of liquid water has little effect per se on snow optical properties in the solar spectrum, in contrast to its enormous effect on microwave emissivity. Snow albedo is increased at all wavelengths as the solar zenith angle increases but is most sensitive around lambda = 1 micrometer. Many apparently conflicting measurements of the zenith angle dependence of albedo are difficult to interpret because of modeling error, instrument error, and inadequate documentation of grain size, surface roughness, and incident radiation spectrum. Cloud cover affects snow albedo both by converting direct radiation into diffuse radiation and also by altering the spectral distribution of the radiation.
The laboratory procedures, algorithms, measurements, and uncertainties associated with generation of the spectral and radiometric calibration of data acquired by AVIRIS are described. AVIRIS is an airborne sensor that obtains high-spatial-resolution image data of the earth in 224 spectral channels in four spectrometers covering the range from 400 to 2450 nm. The spectral calibration of AVIRIS agrees with the in-flight data to within two nanometers, and the absolute radiometric calibration is consistent with the in-flight verification to 10 percent over the spectral range. In-flight radiometric stability as measured by five consecutive passes over the surface calibration site is reported to be between three and five percent.© (1990) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
The Airborne Prism EXperiment (APEX) is an airborne pushbroom imaging spectrometer for Earth observation. Its products will become available in 2011. APEX is currently prepared for final acceptance configuration completing final hardware upgrades, refined calibration methodologies and test flights. APEX is composed of an airborne dispersive pushbroom imaging spectrometer, a Calibration Home Base (CHB) for instrument calibration and a data Processing and Archiving Facility (PAF) for operational product generation and delivery. A unique In-Flight Characterization (IFC) unit is integrated within the sensor optical head, providing pre- and post- data-acquisition characterization monitoring the instruments spectral and radiometric stability. This paper outlines the activities performed with a special focus on system calibration and validation procedures, as well as preliminary measurement results.