Robert A. Fusina

Exploiting manifold geometry in hyperspectral imagery

A new algorithm for exploiting the nonlinear structure of hyperspectral imagery is developed and compared against the de facto standard of linear mixing. This new approach seeks a manifold coordinate system that preserves geodesic distances in the high-dimensional hyperspectral data space. Algorithms for deriving manifold coordinates, such as isometric mapping (ISOMAP), have been developed for other applications. ISOMAP guarantees a globally optimal solution, but is computationally practical only for small datasets because of computational and memory requirements. Here, we develop a hybrid technique to circumvent ISOMAPs computational cost. We divide the scene into a set of smaller tiles. The manifolds derived from the individual tiles are then aligned and stitched together to recomplete the scene. Several alignment methods are discussed. This hybrid approach exploits the fact that ISOMAP guarantees a globally optimal solution for each tile and the presumed similarity of the manifold structures derived from different tiles. Using land-cover classification of hyperspectral imagery in the Virginia Coast Reserve as a test case, we show that the new manifold representation provides better separation of spectrally similar classes than one of the standard linear mixing models. Additionally, we demonstrate that this technique provides a natural data compression scheme, which dramatically reduces the number of components needed to model hyperspectral data when compared with traditional methods such as the minimum noise fraction transform.

Cross-Scale Patterns in Shrub Thicket Dynamics in the Virginia Barrier Complex

A bstractTo interpret broad-scale erosion and accretion patterns and the expansion and contraction of shrub thickets in response to sea level rise for a coastal barrier system, we examined the fine-scale processes of shrub recruitment and mortality within the context of the influence of ocean current and sediment transport processes on variations in island size and location. We focused on Myrica cerifera shrub thickets, the dominant woody community on most barrier islands along the coastline of the southeastern USA. Observations suggest that M. cerifera, a salt-intolerant species, is increasing in cover throughout the Virginia barrier islands, yet rising sea level in response to climate change is increasing erosion and reducing island area. Our objective was to explain this apparent paradox using pattern–process relationships across a range of scales with a focus on ocean currents and sediment transport interacting with island characteristics at intermediate scales. Multi-decadal comparisons across scales showed a complex pattern. At the scale of the entire Virginia barrier complex, modest decreases in upland area were accompanied by large increases in shrub area. Responses were more variable for individual islands, reflecting inter-island variations in erosion and accretion due to differences in sediment transport via ocean currents. Several islands underwent dramatic shrub expansion. Only for within-island responses were there similarities in the pattern of change, with a lag-phase after initial shrub colonization followed by development of linear, closed canopy thickets. Understanding the fine-scale processes of shrub seedling establishment and thicket development, in conjunction with the influence of ocean currents and sediment transport, provides a framework for interpreting island accretion and erosion patterns and subsequent effects on shrub thicket expansion or contraction across scales of time and space.

Bathymetric Retrieval From Hyperspectral Imagery Using Manifold Coordinate Representations

In this paper, we examine the accuracy of manifold coordinate representations as a reduced representation of a hyperspectral imagery (HSI) lookup table (LUT) for bathymetry retrieval. We also explore on a more limited basis the potential for using these coordinates for modeling other in water properties. Manifold coordinates are chosen because they are a data-driven intrinsic set of coordinates, which naturally parameterize nonlinearities that are present in HSI of water scenes. The approach is based on the extraction of a reduced dimensionality representation in manifold coordinates of a sufficiently large representative set of HSI. The manifold coordinates are derived from a scalable version of the isometric mapping algorithm. In the present and in our earlier works, these coordinates were used to establish an interpolating LUT for bathymetric retrieval by associating the representative data with ground truth data, in this case from a Light Detection and Ranging (LIDAR) estimate in the representative area. While not the focus of the present paper, the compression of LUTs could also be applied, in principle, to LUTs generated by forward radiative transfer models, and some preliminary work in this regard confirms the potential utility for this application. In this paper, we analyze the approach using data acquired by the Portable Hyperspectral Imager for Low-Light Spectroscopy (PHILLS) hyperspectral camera over the Indian River Lagoon, Florida, in 2004. Within a few months of the PHILLS overflights, Scanning Hydrographic Operational Airborne LIDAR Survey LIDAR data were obtained for a portion of this study area, principally covering the beach zone and, in some instances, portions of contiguous river channels. Results demonstrate that significant compression of the LUTs is possible with little loss in retrieval accuracy.

A credit assignment approach to fusing classifiers of multiseason hyperspectral imagery

A credit assignment approach to decision-based classifier fusion is developed and applied to the problem of land-cover classification from multiseason airborne hyperspectral imagery. For each input sample, the new method uses a smoothed estimated reliability measure (SERM) in the output domain of the classifiers. SERM requires no additional training beyond that needed to optimize the constituent classifiers in the pool, and its generalization (test) accuracy exceeds that of a number of other extant methods for classifier fusion. Hyperspectral imagery from HyMAP and PROBE2 acquired at three points in the growing season over Smith Island, VA, a barrier island in the Nature Conservancys Virginia Coast Reserve, serves as the basis for comparing SERM with other approaches.

INSAR imagery of surface currents, wave fields, and fronts

The authors demonstrate the ability of interferometric radar imagery to determine both relative and absolute surface velocities in the open ocean. Absolute phase calibration is accomplished by noting the azimuthal displacement of range-travelling targets-demonstrating for the first time that under favourable circumstances phase calibration can be achieved in open-ocean in the absence of ground truth. The high resolution of radar imagery permits observation of sharp velocity discontinuities, e.g. the Gulf Stream boundary and the wave field. The recent SIR-C/X-SAR shuttle missions dramatically emphasize the experimental and observational aspects of space-based radar. The combination of absolute velocities, high spatial resolution, and wide-area coverage suggest that interferometric radar imagery can provide a unique and powerful aid both for studies of global circulation patterns and detailed analysis of slope/shelf water interactions with ocean currents. In particular, the authors employ this measurement of the surface currents and wave field near a velocity front to help refine and bound results of their modeling of calculated radar images of the front. The results of this paper are compared with available ground truth. >

Radar backscatter from breaking waves in Gulf Stream current convergence fronts

Bright linear features have been observed in radar imagery taken near the Gulf Stream (GS) boundary on two separate occasions. In each case, these have been observed directly over strong current convergences. Progress has been made in understanding the origin of these signatures through simulations that incorporate environmental forcing from the winds and currents. These simulations significantly underestimate the backscatter unless wave-breaking (WB) effects are included at least approximately. Using a new, quasistatistical procedure that generalizes and quantifies earlier procedures for including WB effects, the authors have been able to successfully simulate the magnitude and behavior of these signatures. The approach combines the statistically based, composite model of radar backscatter with a deterministic feature model that relates backscatter from breaking waves to a particular geometrical model of a spilling breaker. This is accomplished using localized criteria, defined by local wave crest acceleration, to determine the probability of breaking, and by extending the feature model so that its unknown parameters may be evaluated directly from wave-current interaction calculations. The new approach provides an estimate of the critical crest acceleration of a potentially breaking wave, as a function of wind speed, that agrees with independent measurements.

Retrieval of Substrate Bearing Strength from Hyperspectral Imagery during the Virginia Coast Reserve (VCR’07) Multi-Sensor Campaign

Hyperspectral imagery (HSI) derived from remote sensing can delineate surface properties of substrates such as type, moisture, and grain size. These are critical parameters that determine the substrate bearing strength. Although HSI only sees the surface layer, statistics can be derived that relate surface properties to the likely bearing strength of soils in particular regions. This information can be used to provide an initial map estimate on large scales of potential bearing strength. We describe an initial validation study at the Virginia Coast Reserve relating airborne HSI to in situ spectral and geotechnical measurements through a spectral-geotechnical lookup table (LUT).

Bathymetry Retrieval from Hyperspectral Imagery in the Very Shallow Water Limit: A Case Study from the 2007 Virginia Coast Reserve (VCR'07) Multi-Sensor Campaign

We focus on the validation of a simplified approach to bathymetry retrieval from hyperspectral imagery (HSI) in the very shallow water limit (less than 1–2 m), where many existing bathymetric LIDAR sensors perform poorly. In this depth regime, near infra-red (NIR) reflectance depends primarily on water depth (water absorption) and bottom type, with suspended constituents playing a secondary role. Our processing framework exploits two optimal regions where a simple model depending on bottom type and water depth can be applied in the very shallow limit. These two optimal spectral regions are at a local maximum in the near infra-red reflectance near 810 nm, corresponding to a local minimum in absorption, and a maximum in the first derivative of the reflectance near 720 nm. These two regions correspond to peaks in spectral correlation with bathymetry at these depths.

Wavelength dependence of the bidirectional reflectance distribution function (BRDF) of beach sands.

The wavelength dependence of the dominant directional reflective properties of beach sands was demonstrated using principal component analysis and the related correlation matrix. In general, we found that the hyperspectral bidirectional reflectance distribution function (BRDF) of beach sands has weak wavelength dependence. Its BRDF varies slightly in three broad wavelength regions. The variations are more evident in surfaces of greater visual roughness than in smooth surfaces. The weak wavelength dependence of the BRDF of beach sand can be captured using three broad wavelength regions instead of hundreds of individual wavelengths.

A dual-spectrometer approach to reflectance measurements under sub-optimal sky conditions

This paper presents a practical method for the development of spectral reflectance libraries under sub-optimal sky conditions. Although there are commercially available spectrometers which simultaneously measure both downwelling and upwelling radiance to mitigate the impact of sub-optimal sky conditions, these spectrometers only record in the visible and near infra-red. There are presently no commercially available spectrometers with this capability that can record the visible through short-wave infra-red. This paper presents a practical method of recording and processing data using coordinated measurements from two full-range spectrometers and discusses potential pitfalls and solutions required to achieve accurate reflectance spectra. Results demonstrate that high-quality spectral reflectance libraries can be developed with this approach.