Richard B. Gomez

Preliminary Investigation of Submerged Aquatic Vegetation Mapping using Hyperspectral Remote Sensing

The use of airborne hyperspectral remote sensing imagery for automated mapping of submerged aquatic vegetation (SAV) in the tidal Potomac River was investigated for near to real-time resource assessment and monitoring. Airborne hyperspectral imagery and field spectrometer measurements were obtained in October of 2000. A spectral library database containing selected ground-based and airborne sensor spectra was developed for use in image processing. The spectral library is used to automate the processing of hyperspectral imagery for potential real-time material identification and mapping. Field based spectra were compared to the airborne imagery using the database to identify and map two species of SAV (Myriophyllum spicatum and Vallisneria americana). Overall accuracy of the vegetation maps derived from hyperspectral imagery was determined by comparison to a product that combined aerial photography and field based sampling at the end of the SAV growing season. The algorithms and databases developed in this study will be useful with the current and forthcoming space-based hyperspectral remote sensing systems.

Hyperspectral imaging: a useful technology for transportation analysis

We address hyperspectral imaging (HSI) technology and its attendant key issue of spectral libraries to enable the exploitation of hyperspectral images for transportation applications. Five key applications are reviewed here: detection/identification of submerged aquatic vegetae tion in navigable waterways, detection/tracking of oil spills, extracting/ assessing road characteristics, mapping impervious surfaces, and the detection/identification of vehicles. Central to all these applications is the need for a comprehensive spectral library in which various reflective spectra are correlated with physical surfaces and environments encountered in transportation. Much of this critical work is being funded by the Department of Transportation through four university consortia, each specializing in one of the key transportation areas of: transportation flows; infrastructure; environmental assessment; and safety, hazards, and disaster assessment for transportation lifelines.

Wavelet-based hyperspectral and multispectral image fusion

Different research groups have recently studied the concept of wavelet image fusion between panchromatic and multispectral images using different approaches. In this paper, a new approach using the wavelet based method for data fusion between hyperspectral and multispectral images is presented. Using this wavelet concept of hyperspectral and multispectral data fusion, we performed image fusion between two spectral levels of a hyperspectral image and one band of multispectral image. The reconstructed image has a root mean square error of 2.8 per pixel and a signal-to- noise ratio of 36 dB. We achieved our goal of creating a composite image that has the same spectral resolution as the hyperspectral image and the same spatial resolution as the multispectral image with minimum artifacts.

Hyperspectral image assessment of oil‐contaminated wetland

In exploring the nature of hyperspectral data, this study has focused on one of its most challenging applications—oil spill detection—in order to uncover the potential limits of such data. The classification performance of conventional techniques can be improved by testing the accuracy of the existing classifiers using a ground data image as a reference. Moreover, a created prototype demonstrates how hyperspectral data can supplement information on environmental deterioration due to oil pollution, specifically the Patuxent River wetland at the Chesapeake Bay in Maryland. The data allow an assessment of the current state of wetland losses and habitat changes due to oil pollution of local waters and associated wetlands. Airborne Imaging Spectro‐Radiometer for Applications (AISA) hyperspectral imagery was used for this study and the results were derived using the Environment for Visualizing Images (ENVI) software. The use of different classifiers showed low accuracy and class overlap for many classes. Therefore, a ground data image was created using maximum likelihood (ML) classification to compare the results of several classifiers and to assess the accuracy of each technique. Using 2D scatter plots for selecting regions of interest yielded more accurate results than digitizing polygons for training samples. It allowed precise identification of grass stress and soil damaged by polluted water.

Hyperspectral imaging applied to medical diagnoses and food safety

This paper analyzes the feasibility and performance of HSI systems for medical diagnosis as well as for food safety. Illness prevention and early disease detection are key elements for maintaining good health. Health care practitioners worldwide rely on innovative electronic devices to accurately identify disease. Hyperspectral imaging (HSI) is an emerging technique that may provide a less invasive procedure than conventional diagnostic imaging. By analyzing reflected and fluorescent light applied to the human body, a HSI system serves as a diagnostic tool as well as a method for evaluating the effectiveness of applied therapies. The safe supply and production of food is also of paramount importance to public health illness prevention. Although this paper will focus on imaging and spectroscopy in food inspection procedures -- the detection of contaminated food sources -- to ensure food quality, HSI also shows promise in detecting pesticide levels in food production (agriculture.)

Mapping Impervious Surface Type and Sub‐Pixel Abundance Using Hyperion Hyperspectral Imagery

Abstract Impervious surfaces have been identified as an important and quantifiable indicator of environmental degradation in urban settings. A number of research efforts have been directed at mapping impervious surface type using multispectral imagery. To date, however, no studies have compared equivalent techniques using multispectral and hyperspectral imagery to that end. In this study, data from NASAs 220‐channel Hyperion instrument were used to: a) delineate three types of impervious surface, and b) map sub‐pixel percent abundance for a study site near Washington, D.C., USA. The results were compared with the results of similar methods using same‐spatial‐resolution Landsat ETM+ data for mapping impervious surface type, and with the results of the U.S. Geological Surveys National Land Cover Data (NLCD) 2001 impervious surface data layer, which is derived from Landsat and high‐resolution Ikonos data. The accuracy of discriminating impervious surface type using Hyperion data was assessed at 88% versus Landsat at 59%. The sub‐pixel percent impervious map corresponded well with the NLCD 2001; impervious surface in the study area was calculated at 29.3% for NLCD 2001 and 28.4% for the Hyperion‐derived layer. The results suggest that fairly simple techniques using hyperspectral data are effective for quantifying impervious surface type, and that high‐spectral‐resolution imagery may be a good alternative to high‐spatial‐resolution data.

Detection and monitoring of oil spills using hyperspectral imagery

Oil pollution is a very important aspect in the environmental field. Oil pollution is an important subject due to its capacity to adversely affect animals, aquatic life, vegetation and drinking water. The movement of open water oil spills can be affected by mind, waves and tides. Land based oil spills are often affected by rain and temperature. It is important to have an accurate management of the cleanup. Remote sensing and in particular hyper-spectral capabilities, are being use to identify oil spills and prevent worse problems. In addition to this capability, this technology can be used for federal and state compliance of petroleum related companies. There are several hyper-spectral sensors used in the identification of oil spills. One commonly use sensor is the Airborne Imaging Spectroradiometer for Applications (AISA). The main concern associated with the use of these sensors is the potential for false identification of oil spills. The use of AISA to identify an oil spill over the Patuxent River is an example of how this tool can assist with investigating an oil pipeline accident, and its potential to affect the surrounding environment. A scenario like this also serves as a good test of the accuracy with which spills may be identified using new airborne sensors.

Hyperspectral trace gas detection using the wavelet packet transform

A method for trace gas detection in hyperspectral data is demonstrated using the wavelet packet transform. This new method, the Wavelet Packet Subspace (WPS), applies the wavelet packet transform and selects a best basis for pattern matching. The wavelet packet transform is an extension of the wavelet transform, which fully decomposes a signal into a library of wavelet packet bases. Application of the wavelet packet transform to hyperspectral data for the detection of trace gases takes advantage of the ability of the wavelet transform to locate spectral features in both scale and location. By analyzing the wavelet packet tree of specific gas, nodes of the tree are selected which represent an orthogonal best basis. The best basis represents the significant spectral features of that gas. This is then used to identify pixels in the scene using existing matching algorithms such as spectral angle or matched filter. Using data from the Airborne Hyperspectral Imager (AHI), this method is compared to traditional matched filter detection methods. Initial results demonstrate a promising wavelet packet subspace technique for hyperspectral trace gas detection applications.

Combining MISR, ETM+ and SAR data to improve land cover and land use classification for carbon cycle research

Accurate and reliable information about land cover and land use is essential to carbon cycle and climate change modeling. While historical regional-to-global scale land cover and land use data products had been produced by AVHRR and MSS/TM, this task has been advanced by sensors such as MODIS and ETM since the latter 1990s. While the accuracies and reliabilities of these data products have been improved, there have been reports from the modeling community that additional work is needed to reduce errors so that the uncertainties associated with the global carbon cycle and climate change modeling can be addressed. Remotely sensed data collected in different wavelength regions, at different viewing geometries, usually provide complementary information. Their combination has the potential to enhance remote sensing capabilities in discriminating important land cover components. In this paper, we studied multi-angle data fusion, and optical-SAR data fusion for land cover classification at regional spatial scale in the temperate forests of the eastern United States. Data from EOS-MISR, Landsat-ETM+ and RadarSat-SAR were used. The results showed significantly improved land cover classification accuracy when using the data fusion approach. These results may benefit future land cover products for global change research.

Fast hyperspectral data processing methods

The need for fast hyperspectral data processing methods is discussed. Discussion includes the necessity of faster processing techniques in order to realize emerging markets for hyperspectral data. Several standard hyperspectral image processing methods are presented, including maximum likelihood classification, principal components analysis, and canonical analysis. Modifications of those methods are presented that are computationally more efficient than standard techniques. Recent technological developments enabling hardware acceleration of hyperspectral data processing methods are also presented as well as their applicability to various hyperspectral data processing algorithms.