ITRES Recommends ENVI for Hyperspectral Analysis

ITRES* Research Limited builds and supports state of the art airborne hyperspectral imaging systems for studying the environment. They provide complete data analysis solutions to industries like forestry, agriculture, water quality mapping, military ordinance detection and emergency response. The data collection tool central to the imaging systems is casi (Compact Airborne Spectral Imager), the ITRES-developed, push-broom remote sensing device. ITRES' remote sensing systems allow researchers to collect data in a cost-effective, programmable manner. When ITRES' customers need to learn all they can from their hyperspectral data, ITRES suggests incorporating ENVI (Research Systems' Environment for Visualizing Images) as the software for remote sensing data analysis.

"The ease of use, ability to handle casi hyperspectral data, powerful algorithms and flexible parameters make ENVI an excellent choice for hyperspectral analysis," says Steve Mah, vice president of commercial applications at ITRES. In the data analysis process, factors such as the sensor and attitude alignments, accuracies of the attitude measurement system, GPS (Global Positioning System), ground control points and the digital elevation model all need to be taken into consideration. This ancillary data, in combination with the hyperspectral data (in the 400 nm to 1000 nm band range, depending on the application requirements), adds up to more than two gigabytes of data amassed during each operational remote sensing flight. In addition, multiple flights are routinely required to scan the entire area under study, further compounding the volume of data until complete images, sometimes as large as 50 gigabytes, are created. ENVI software provides all the tools and data handling capabilities researchers need to determine the accuracy of the data and assess forest silviculture and aquatic characteristics.

"Users usually rely on ENVI's spectral analysis tools, for example the spectral angle mapper and unmixing algorithms, to examine the data that's important to their research," says Mah. ENVI also includes an extensive spectral library, as well as masking and classification routines which complement ENVI's many other image processing and data analysis routines. "Additional forest analysis algorithms are developed using IDL and incorporated with ENVI," says Mah, referring to how ITRES has extended ENVI to meet some of their client's needs.

CSIRO Studies Australia with casi & ENVI

In 1993, CSIRO (Commonwealth Scientific Research Organization) first used casi's calibrated hyperspectral scanning abilities to quantitatively map aspects of water quality, such as the turbidity and chlorophyll concentration, in the key rivers and lakes of Australia. Analyzing the data required CSIRO to spend time building their own specialized models of radiative transfer through the atmosphere and water.

According to Dr. Alex Held of CSIRO's Remote Sensing Group, CSIRO recently began using ENVI to perform "advanced spectral analysis" of the enhanced spectral data gathered with casi over the Daintree Mangrove Estuary, in Far North Queensland. "We are interested in evaluating the casi data to determine the mangrove diversity of estuaries along a 'diversity tract' that spans from Daintree to Western Somoa and Fiji," says Held. "ENVI has been used to help us construct a spectral library for various mangrove species. We are in the process of using this library for spectral feature fitting and unmixing of the sixteen band, fifty-square kilometer mosaic of the area at a two-point-five meter resolution." ENVI's built-in spectral libraries provide quick reference to USGS spectral data, giving users an extensive, well-characterized reference set of mineral and some vegetative spectral values.

Because ENVI's underlying language is IDL¨, the Interactive Data Language, extending its functionality and linking to other languages is quick and easy. "We expect to develop and add our own set of tools to the ENVI 'tool box' thanks to the ease of incorporating C, Fortran or IDL routines into the program," adds Held.

Seeing the Trees in the Forest

MacMillan Bloedel, the largest forestry company in Canada, goes beyond just harvesting trees and processing timber. MacMillan Bloedel's Remote Sensing Analyst Dennis Paradine is conducting operational research for forestry management in conjunction with ITRES and the Pacific Forestry Center of the Canadian Forest Service. The team hopes to develop a more cost-effective method for identifying diseased trees within dense forests. The research is funded by Forest Renewal British Columbia.

One of the aims of the project is to provide more cost-effective data for forest management. "The casi has the potential to provide an alternative data source at a lower cost than conventional timber cruising," says Paradine. Timber cruising requires researchers to make physical observations while hiking through the forest. In the long run, Paradine's work will contribute to a project called "Development of Certified Applications using Compact Airborne Spectrographic Imager (casi) data." The project's goal is to build a complete system that can be used by foresters and engineers throughout British Columbia for forest management.

For the past year, Paradine has been analyzing the costal rain forests along the west coast of British Columbia. Paradine focuses on near infrared, red, green and blue spectra to classify forest species. He also looks for evidence of Phellinus sp. root rot in the eight-band hyperspectral data. "Within the imagery, we first look for gaps in the crown closure of the forest canopy," Paradine explains. "Then we concentrate on finding 'snags,' places where there's a lot of standing dead trees. After that, we look for infested trees on a tree-by-tree basis." After determining the root rot "epicenter," MacMillan Bloedel's foresters are alerted and the trees are removed before the root rot spreads, saving money and timber volume. "Also, we use ENVI's spectral angle mapping functions for characterizing river features such as depth, river-bottom composition, and the location of boulders and sand bars," Paradine says. This provides information on different watershed and ecological processes for both monitoring and change detection purposes.

Advances in remote sensing technology will help us better understand and manage our environment, and preserve the Earth's resources. By providing cost-effective, easier to use tools, ITRES stands to help make the benefits of new technology available to a wider group of researchers and commercial users.

*ITRES is an acronym that stands for "Innovative imaging Technology and leading edge Research to meet the needs of customers with emphasis on scientific Excellence and Service"