Feature ArticleAn Airborne Imaging Multispectral Polarimeter
By Dr. Brett A. Hooper
Graduate Student Intern
Dr. J. Zandy Williams
The use of remote sensing for characterizing littoral and estuarine environments has become an increasingly important tool for gaining a better understanding of the hydrodynamic influence of waves, tides, turbulence and currents on the morphology within these environments. Airborne remote sensing that provides time-series optical imaging of the scene, in tune with temporal processing, offers a viable way to estimate these important hydrodynamic parameters at regional scales on the order of 10 to 100 square kilometers.
Areté Associates has constructed, tested and deployed in a variety of venues an airborne multispectral polarimetric time-series imaging system that has enhanced and expanded the ability to obtain accurate measurements of these dynamic environmental parameters.
Areté's Airborne Remote Optical Spotlight System-Multispectral Polarimeter (AROSS-MSP) is a 12-channel sensor system that measures four color bands (RGB and near-infrared) and three polarization states for the full linear polarization response of the imaged scene. Typical single-image footprints provide area coverage on the water surface of about two square kilometers with two-meter ground sample distance. These images can be combined in a mosaic that can cover a region on the order of 100 square kilometers.
Polarimetric remote sensing provides information about imaged scenes that cannot be obtained from luminance and spectral measurements alone. The AROSS-MSP minimizes surface reflections to image the subsurface water column at a greater depth than randomly polarized light, which allows sedimentation fronts to be tracked. Exploiting the polarization state of light reflected from the ocean surface also enables measurement of the 2D slope field of the surface. From this measurement, the significant wave height of the surface wavefield can be calculated.
AROSS-MSP also has the ability to image through marine haze and maximizes wave contrast to improve hydrographic parameter retrievals, such as waterline contours and topography, directional wave spectra, water depth and currents. Color and polarization imagery on this scale and resolution is also capable of distinguishing fronts at saltwater and freshwater interfaces, observing sea grass and other subaquatic vegetation distributions, and determining bottom structure on exposed tidal flats.
AROSS-MSP was designed, constructed and evaluated as part of a U.S. Navy Office of Naval Research (ONR) Small Business Innovative Research (STTR) Phase II program. The sensor was designed to employ ad'vanced, visible-band commercial off-the-shelf technology to support littoral-zone environmental intelligence products, and it is based on the proven technology of Areté's AROSS series of sensor packages.
The AROSS system was specifically designed to provide time-series imagery of ocean waves suitable for retrieval of meteorological and oceanographic parameters over large areas. These areas can be covered by looking out from an aircraft at grazing angles (measured from the horizon, down) of about 30°, providing good wave visibility. The suite of products that can be derived from a sensor of this type ex'tends from near-shore oceanographic bathymetry to river current and sediment transport data.
The system architecture consists of a rack-mounted computer that controls and maintains the aim of the positioner using position and attitude data from an inertial measurement unit (IMU). The computer also controls the imagery through commands to the camera and records all received data, including the attitude, position, global positioning system time, camera parameters, trigger times and imagery. The sensor payload is held in a yoke-style positioner (SPS-500) from Atlantic Positioning Systems (Largo, Florida). This allows the payload to be pointed to an aimpoint and the imagery to be georectified.
In order to maximize spatial resolution while collecting multichannel data, a multicamera approach was chosen. In order to measure four colors and three polarization states, 12 Pixelfly QE cameras (1,392 by 1,024 pixels) from Romulus, Michigan-based Cooke Corp. were used. These cameras combine a small form factor with a high-dynamic range, allowing all 12 to fit into a small payload. Schneider (Hauppauge, New York) COMPACT f/1.9 35-millimeter lenses (15° by 11° field of view) were chosen for their high quality and rugged, compact size.
The 12 cameras were arranged in four modules of three cameras each, with each module featuring a different color filter and three polarization filters.
The color/polarization filters are linear with broadband coatings that select 100-nanometer bands with peak transmissions near 450, 550 and 650 nanometers for blue, green and red color bands and a 200-nanometer band centered near 800 nanometers in the near-infrared. The polarization filters are oriented with respect to the horizontal at 30°, 90° and 150° for stability of estimates of degree and angle of linear polarization (P and ψ), making them well suited for measuring ocean waves.
Use of multichannel data, such as from AROSS-MSP, requires accurate intercamera calibrations in addition to high-quality individual camera calibrations. Each camera in AROSS-MSP is optimally focused and corrected for lens vignetting, nonuniform pixel response, relative radiometry and geometric distortion.
Camera calibration is a fundamental requirement of all photogrammetric systems. The calibrations required by a multisensor system can be roughly grouped into two categories: single-head calibrations and intersensor calibrations. The first category comprises most of the tests that would normally be performed on a standard single sensor (e.g., image quality, flat field, radiometric calibrations) and are the intrinsic parameters of the camera. The second category of measurements is concerned with the relative position and angular aim of the 12 cameras to one another and to the IMU system. A systems approach to the instrument is taken using both laboratory and in-situ measurements to arrive at the best calibration.
The intrinsic camera parameters were measured using standard grid patterns and techniques developed at AretÉ Associates, based on work done by R. Y. Tsai. These calibrations allow for the correct mensuration of the remotely sensed scene and include flat field, radiometric, polarimetric and geometric distortion calibrations.
Radiometric calibrations allow for accurate spectral measurements and for the correction of the irradiance roll-off that is present in all cameras. Production of polarimetric image products, such as P and ψ, requires accurate mutual alignment of the cameras to avoid dilution of the photon polarization information in any pixel by dissimilar information in adjacent pixels.
After each of the individual camera heads is calibrated, the cameras are mounted in three-camera modules and each of the modules is aligned to a reference camera. Following this coarse, mechanical alignment, the angular and position offsets of the cameras are determined using flight data and in-situ measurements of a number of fiducial targets placed throughout the system's field-of-view when it is at its nominal flight altitude.
Using these same targets and other tie points in the scene, the intercamera alignment can be measured such that composite images are generated with subpixel fidelity, on the order of 1/20th of a pixel. The imagery is then georectified to a patch on the ground, providing a common, fixed geodetic surface for time-series collections. The mapped, corrected image data are then analyzed for production of single-frame data products, such as color and polarization imagery, degree and angle of linear polarization imagery, and time-series products, such as currents and bathymetry.
Imagery and Data Products
With camera calibration corrections applied, the combination channels of color and polarization can be viewed and quantitative analyses performed. Imagery over the ocean was collected at the U.S. Army Corps of Engineers Field Research Facility's (FRF) eight-meter ocean directional wave spectrum array—an in-water instrument array that measures the directional wave spectrum in coastal waters at the FRF in Duck, North Carolina—in order to ground truth AROSS-MSP's measurements. Composite color and polarization imagery was generated over the eight-meter array on a particularly windy day, with lots of whitecaps.
A sensor performance model that includes path, sky and upwelling radiance, as well as polarization, was developed to compare the P and ψ values measured by AROSS-MSP, and excellent agreement was found between the model and AROSS-MSP values. Work is ongoing to measure sea slopes and significant wave heights from this polarization imagery.
Color and polarization imagery has also been collected over the Cape Fear River in North Carolina. Suppression of surface-reflected light and enhancement of the upwelling light from the river in the vertical polarization color composite can be noted in this imagery. In the horizontal polarization composite imagery, the surface-reflected light enhances the signature of the waves from boats in the scene. Near-infrared imagery has a bright signal on land due to reflectance from green vegetation. Whitewater, created both by boats and salt-fresh water frontal boundaries, is evident in the RGB and near-infrared images and depicts how this Lambertian scattering surface depolarizes light compared to more highly polarized light reflected from the water in the P image.
A field of current vectors is also retrieved from a time-series of images for this region of the river. Imagery is collected in each of the 12 channels simultaneously at two hertz, and a two-minute section of an orbit is processed for each channel to retrieve the current. The current measurement, one meter per second in this case, is derived in a grid of boxes (100 meters on a side) that fill the region of interest, approximately two square kilometers. The image-derived current values agree well with in-situ current measurements made with an acoustic Doppler current profiler river raft, with root mean square deviation of seven centimeters per second in magnitude and less than 5°.
The airborne image data collection also generates image mosaics from stripmap data collections that cover large areas on the order of 100 square kilometers. Imagery and data products on this regional scale, such as 2D surface currents, are of interest to modelers who are striving to understand the complex hydrodynamics of the near-shore ocean and its interplay with nearby tidal flat, estuarine and riverine environments.
A prototype imaging multispectral polarimeter, AROSS-MSP, was designed, constructed and evaluated as part of an ONR STTR Phase II program to investigate simultaneous multispectral and polarimetric time-series imagery, and the sensor can enhance oceanographic, estuarine and riverine characterization. The sensor package is also capable of accurately geolocating these products from modest grazing angles, which provide large-area coverage and good visibility when imaging ocean waves.
AROSS-MSP is optimally calibrated and aligned with subpixel fidelity, which is vital for generation of composite color and polarization imagery. The mapped, corrected image data is analyzed for production of single-frame data products, such as color and polarization imagery, and time-series products, such as currents. Color and polarization imagery have been analyzed over ocean, tidal flat and riverine environments and compare well with a sensor-performance model that describes the color and linear polarization as measured at the sensor.
The authors thank ONR for its support of this work.
Dr. Brett A. Hooper, senior scientist at Areté Associates since February 2004, has led the development of the Airborne Remote Optical Spotlight System-Multispectral Polarimeter and a sensor-performance model that models the passive RGB and near-infrared polarimetric signature as measured by an airborne sensor.
Becky Baxter is a graduate student pursuing her Ph.D. in physics from Georgetown University. With support from Aret' Associates as an intern, her research is focused on polarimetric remote sensing of ocean waves.
Dr. J. Zandy Williams has served as a senior scientist at Aret' Associates since 1995. He has developed many of the airborne remote optical spotlight systems that make up the AROSS family of sensors and developed algorithms for this time-series imagery.