|
WATCHING MOUNT ST. HELENS RECOVER by Phil Druker Wind is blowing hard out of the south, rain falling horizontally. Ed Rykiel and I are holding a plastic poncho over George He, who huddles over his laptop computer. In mid-October, we are at the base of Mount St. Helens on the Pumice Plain, the moonscape north of the volcano’s crater left by the 1980 eruption. Despite our efforts, rain spatters the laptop screen and keyboard. Rykiel, an ecologist in the Department of Environmental Science and Regional Planning at WSU Tri-Cities, wants to develop a holistic picture of ecological changes that have occurred on St. Helens since the catastrophic eruption. So in collaboration with Karen Steinmaus and George He from the Pacific Northwest National Laboratory, he plans to evaluate how plant communities become established and evolve on the mountain by using advanced remote sensing technology. This technology uses the principle that objects on the Earth’s surface reflect sunlight, and this reflected light can be analyzed in terms of the spectrum. Using spectral analysis, light—whether in the visible or invisible range—can be measured as wavelengths representing specific energy levels. Thus, each object on the Earth’s surface reflects light in a slightly different way. We see this difference as color. Using imaging techniques, researchers can divide the spectrum and measure it with different degrees of precision. For example, aerial photos use black and white imagery. Other techniques use a range of the spectrum humans cannot see: infrared light. This makes it possible to divide and measure light over a wide range of wavelengths represented as bands. Landsat satellite imagery, spectral imagery first introduced in the 1970s as a means of evaluating landscapes, involves multispectral imagery. It uses from a few to a couple dozen bands and has a spatial resolution on the order of a kilometer. The military has been developing and using this kind of remote sensing technology for surveillance. Now, hyperspectral imagery provides spatial resolution of one square meter, and its spectral resolution allows researchers to work with 210 bands of data. Each picture element of the hyperspectral image, called a pixel, defines the location and a measure of the brightness of each fundamental unit. When spectral information is measured at all the pixel locations, a “data cube” can be obtained to calculate the spectral reflectance of specific objects on the ground. Thus, using hyperspectral imaging, researchers can identify the specific bands of light each type of material reflects, and this gives each object on the Earth’s surface, including individual species of plants, its own spectral signature. Rykiel and the multidisciplinary team plan to use this kind of data to identify individual species of plants. This remote sensing technology is becoming available for environmental assessment. In the summer of 1996, the Naval Research Laboratory sent a research airplane over Mount St. Helens equipped with the hyperspectral digital imagery collection experiment (HYDICE) sensor for taking hyperspectral images of a 2,100-foot-wide swath of the Pumice Plain. And those images became available to Rykiel’s group in 1998. The question is how to interpret all this data. And that’s why we are here on Mount St. Helens. With images of the swath of the Pumice Plane, we are trying to match the image with the landforms and what is growing on the ground. Finally, the wind abates, and the sun begins to burn through the clouds rising on the mountain, a curtain rising on a moonscape turning green with fall moisture. Waterfalls cascade from the crater above, and snow covers the peak. My hands are numb with cold. George He presses more keys and places a sample willow leaf in the spectroradiometer, an instrument that measures spectral reflectance. He gets a good reading and saves the data in the computer. The field data he is gathering show the specific spectral signatures for the main types of vegetation growing on Mount St. Helens: lupine, willow, moss, and grass. Initial results look promising. The technology works, but the data is extremely complex and varied. For example, the tops and bottoms of willow leaves are different colors, which causes each side to reflect a different spectral signature. Thus, the spectral signature for willows on a still day will differ from their signature on a windy day. When Mount St. Helens erupted in May 1980, the blast blew out the side of the volcano, scoured the mountainside, and left barren layers, in places over 100 feet deep, of light tan volcanic rock and ash to create the Pumice Plain. Now, as we hike through the rain, wind, and fog—just 18 years after the eruption—we pass lupine colonies and other areas covered with dense thickets of willows. For evolutionary ecologist Rykiel, these colonies of plants raise questions about how they got there and how the colonies grew. Over the years, various researchers have gathered data to answer these questions. This work has focused mainly on individual plots, making it difficult to assemble data that offer a complete picture showing how plant communities evolve on the landscape scale. Some of the plots we pass exemplify plant colonization by diffusion. In this case plants spread slowly from the edge to the center. Other plots show how plant colonies spread by what Rykiel calls “salation”—when plants reach a certain area by skipping over other vegetated or bare areas. Studying the individual plots then shows that colonization appears to follow no particular pattern. Rykiel’s preliminary results, however, suggest a different interpretation. While colonization appears locally random in individual plots, it may be non-random on a global scale. So what appears locally unpredictable may be predictable when the spatial scale changes to include the whole landscape, the scale that hyperspectral images offer. Rykiel and other researchers hope this landscape-level analysis will help explain how ecosystems become established and evolve. Phil Druker teaches writing at the University of Idaho. |
