Imaging systems using the short-wave infrared (SWIR) wavelength bands offer unique remote sensing capabilities, such as material detection and smoke penetration overcoming challenges in measurement, inspection, and process-monitoring applications often impossible with other technologies. This extraordinary capacity springs from the convergence of the phenomenology's unique absorption properties, atmospheric transmission windows that allow the Earth to be imaged from space, and particle scattering physics that enable smoke penetration.
To better understand this technology, consider the full spectrum of light. SWIR refers to nonvisible light falling roughly between 1400 and 3000 nanometers (nm) in wavelength. Visible light, on the other hand, typically corresponds to the 400 to 700 nm range. Immediately adjacent to visible light is near infrared, or NIR, within the 700 to 1400 nm range, and SWIR is adjacent to NIR.
Collecting satellite imagery in SWIR wavelengths has unique benefits, including improved atmospheric transparency and material identification. Because of their chemistries, many materials have specific reflectance and absorption features in the SWIR bands that allow for their characterization from space. Examples include minerals used in mineral exploration; urban features, such as roofing and construction materials; vegetation; petroleum (e.g. an oil spill); and a variety of other man-made chemical compounds. Snow and ice display distinctive variations in some SWIR bands, and SWIR-based imaging can even penetrate some types of smoke, such as from a forest fire.
Atmospheric aerosols affect shorter wavelengths, primarily in the visible range, and have a lesser impact on SWIR bands. Water molecules, however, absorb light in some SWIR wavelengths, rendering the atmosphere nearly opaque in these ranges. Remotely sensed data must therefore be collected in atmospheric windows between these water absorption wavelengths, limiting the placement of sensor bands.
SWIR sensing opportunities occur in three atmospheric windows. Starting with shorter wavelengths, the first window is centered near 1250 nm. Bands here are useful for bracketing iron absorption features (at shorter wavelengths). Vegetation indices sensitive to leaf moisture content, such as the Normalized Difference Water Index, also use bands in the 1250 nm window.
The second SWIR window is roughly between 1500 and 1750 nm. Man-made materials and chemicals present multiple absorption features in this range. Examples include plastics, fiberglass, and petroleum. Also, snow and ice can be differentiated from clouds in this window.
The third atmospheric window—between 2000 and 2400 nm—offers unique opportunities because of its mineral absorption features. With sufficient sensor radiometric resolution, observers can make mineral identifications and chemical measurements in this window unavailable in other reflected wavelengths.
Leveraging Absorption Features
These absorption characteristics are a powerful phenomenology that can be exploited in SWIR wavelengths. In general, the interaction of light waves (electromagnetic energy) with matter involves the transfer of energy. Energy from light is either transferred to molecules of matter (absorbed) or reflected away from them. This is why things get hot when the sun shines on them. In addition, light is not absorbed uniformly. For example, red paint appears red because more red light is reflected, while blue and green light is absorbed.
The various wavelengths of light are absorbed differently because each wavelength (e.g. blue, red, and SWIR) has unique energy content. Using concepts from quantum mechanics, scientists describe electrons in atoms as existing only in specific energy states, or orbitals. Accordingly, electrons can absorb only discrete amounts, or packages, of energy. This means that different substances absorb discrete and unique wavelengths, or energy levels, of light. This is called an electronic absorption.
A second type of absorption is called a vibrational absorption. Atomic bonds in molecules function like little springs. As electrons interact with atoms, the bonded atoms are drawn to or repulsed from each other, creating a vibrating motion. These vibrations have a frequency, and specific wavelengths (or frequencies) of light can stimulate the vibration. As a result, the wavelength of light is absorbed. In the SWIR spectrum, absorption features are primarily due to molecular vibrations.
Electronic absorptions at wavelengths less than about 1000 nm allow for identification of materials containing iron (Fe+3). Molecular vibrational features at wavelengths between 1000 and 2500 nm are diagnostic of materials containing anion groups, such as Al-OH, Mg-OH, Fe-OH, Si-OH, CO3, NH4, and SO4. Small differences in the absorption band wavelength position and shape are correlated with both material composition and variability.
Hyperspectral imagers provide the most effective method of remotely sensing SWIR absorption features. These systems have many narrow contiguous spectral bands and can accurately discriminate absorption features' wavelength position and shape.
Scientists have deployed a variety of these imagers on airborne platforms. For example, the Airborne Visible/Infrared Imaging Spectrometer, or AVIRIS, has been used on four aircraft platforms with an assortment of pixel sizes, ranging from 2 to 20 m, depending on the altitude at which they operate. On a satellite platform, the Hyperion hyperspectral imager has 30 m pixels. At much lower spectral resolution, multi-spectral satellites, such as Landsat, have single bands in the SWIR windows. With single bands, scientists can make only general inferences about broad material categories.
Super-spectral satellite-based instruments, on the other hand, have multiple noncontiguous bands in the SWIR atmospheric windows. Examples include MODIS and VIIRS with spatial resolutions of 500 m and 750 m, respectively, for NIR-SWIR bands; ASTER with 30 m pixels; and the upcoming WorldView-3 satellite with 3.7 m pixels. These sensors permit an intermediate level of material characterization between multi-spectral and hyperspectral.
Greg Hammann is Senior Director of Image Analysis, Product Development and Labs, Next Generation Products, Advanced Geospatial Products, DigitalGlobe.