|
Satellite Remote Sensing/GIS
A Portland, Ore. forest management company uses high resolution satellite imagery
to manage its properties more effectively.
By Kevin P. Corbley Hampton Tree Farms Inc. is among the first to put high-resolution satellite imagery to the test in an operational timber mapping project, and the Portland, Ore., forest management company says it is impressed with the results.
"It's accurate and current, and that's what we need," said Dennis Creel, timber manager for Hampton. "We need to have an up-to-date base map with accurate acreage of timber types in order to better manage our properties."
Hampton has not limited its use of satellite imagery to counting acreage. The firm, which owns several tree farms, lumber companies and mills in the Pacific Northwest, relies on the images to distinguish tree species, estimate age, and map transportation routes in its land holdings. Soon, the company may also employ the images in watershed analysis projects.
Satellite images are not new technology to Hampton. Since 1990, it has contracted Atterbury Consultants, a remote sensing/GIS firm in Beaverton, Ore., to build and regularly update a GIS of Hampton's timberlands in Washington and Oregon. After using Landsat and SPOT images for years in the inventory projects, Atterbury decided to order a panchromatic scene from the new Indian IRS-1C satellite in 1996.
IRS-1C is the one of the latest remote sensing satellites launched by the Indian Space Research Organization (ISRO). It went into orbit late in 1995. Worldwide marketing and distribution of IRS imagery are handled by Space Imaging EOSAT of Thornton, Colo., which operates Landsat and will launch the first 1-meter resolution satellite later this year.
IRS-1C grabbed the attention of Atterbury and other long-time image users with its 5-meter panchromatic imagery, the highest resolution now available on the commercial market. The panchromatic sensor is one of a three-instrument suite of scanners on the satellite which also acquires multispectral and wide-area images.
"We received the second commercial image sold by Space Imaging EOSAT in the United States," said Bob Wright, Atterbury's information systems forester. "From what we have seen so far, [high-resolution imagery] has great potential in forestry." Updating Forest Maps with Imagery
Atterbury planned to integrate the Indian imagery directly into a 95,000-acre mapping project on a property Hampton was managing near Longview, Wash. The timber company wanted to use the imagery to straighten out a mapping problem that had arisen with the purchase of the property.
When Hampton assumed management of the property, its owner was in the process of creating a new digital GIS base map for the tract. Unfortunately, the owner's base map did not match well with digital maps the Washington Department of Natural Resources (DNR) was encouraging timber companies to use.
"The problem was marrying the two maps," said Creel. "The roads were not in the same places, and some of forest type boundaries were not located correctly."
The recently acquired IRS-1C imagery would be used to settle any inconsistencies between the two data sets. But Atterbury was not able to move ahead with the project as quickly as planned. The Indian imagery was so new that commercial image processing packages did not yet have functions to import it.
Atterbury regularly uses TNTmips image processing software in its forest mapping projects and called upon its developer, MicroImages of Lincoln, Neb., for assistance.
"MicroImages really kept us on schedule," said Jeff Jenkins, Atterbury's support services manager. "They figured out a work-around solution so we could use the imagery right away, and then they sent a new import routine within a couple weeks."
The IRS-1C imagery had been delivered in a modified version of Space Imaging EOSAT's Fast Format, which MicroImages determined could be imported into the image processing software as a simple array file. The solution worked, and Atterbury was able to perform all of the same processing functions on the Indian image in TNTmips that it usually applies to Landsat and SPOT images.
After applying the enhancements to the panchromatic imagery, Atterbury converted the image to a TIF file and exported it to ForestVIEW, a desktop mapping and GIS software that it developed specifically for forest inventory applications. Atterbury overlaid the satellite image on the digital maps from the land owner and the DNR to rectify and correct both sets.
The image confirmed that some of the inconsistencies between the two maps resulted from differences in their dates, such as one showing a timber stand where the other indicated a clear cut. But most of the errors related to the position of vector delineations of timber stands.
"We discovered errors in the GIS vectors on the digital maps," said Jenkins. "There was a systematic positional shift of 300 to 500 feet that we attributed to projection shift, and there were variable errors related to radiometric distortions or topographic displacement transferred from the aerial photos that had been used to make the original timber type vectors."
Atterbury technicians found the satellite resolution so clear that they were able to manually correct and update the vector lines on the digital maps on screen in ForestVIEW. Using the satellite overlay as their reference, they identified boundaries of timber stands, noted the position of roads, and snapped the appropriate vector lines into their correct place on the digital maps.
Hampton believes many of the mistakes found on the land owner's maps are symptomatic of problems related to traditional mapping practices still prevalent in the industry. Traditional mapping cannot keep up with the degree of change that occurs every year in a forest, said Creel.
"Unless a forester counts every unit at harvest time, it's easy for the acreage maps to get bit off," he said. "If you lose an acre here or there, it really adds up after awhile with the value of an acre running $25,000."
Creel said he hopes to use imagery to update land holdings once a year in active areas and every four years on less active properties. Viewing Watersheds in 3D
Atterbury has recently been working with Landsat and SPOT imagery to create three-dimensional views of forested lands in support of watershed analysis studies. The company expects these analyses will be much more detailed and accurate when 5-meter IRS-1C imagery is used.
"A watershed analysis tries to identify existing conditions and predict potential problems associated with various forest related activities," said Jenkins.
Atterbury has created some of these 3D perspectives by purchasing USGS digital elevation models (DEM) and importing them into TNTmips where the satellite image is draped over the terrain points. These merged files can be exported and used in the 'fly-through' software that has now become popular in GIS modeling projects.
TNTmips is equipped with automated elevation routines which Atterbury prefers to use in its watershed analysis projects. The elevation option performs surface computation, viewshed processing and watershed analysis. The surface computation includes automated slope and aspect analysis, as well as shading functions.
"It essentially runs a classification on the image and the DEM to classify the image pixels by elevation and aspect instead of by land cover," explained Jim Bender, an Atterbury image processing technician.
Slope is calculated to determine the steepness of a given surface by comparing its elevation value with adjacent cells, while aspect is a measure of which direction the surface slopes. Shading determines where shade will cover the landscape based on topography and various sun positions.
TNTmips' watershed process uses elevation information to calculate the area of a watershed above a given point that drains into that point. It also determines pour points between neighboring watersheds and flow paths from any point in the watershed.
The viewshed process is popular among foresters as they attempt to determine exactly how harvesting will alter the landscape. It essentially allows the forester to choose a point in the image and view the boundaries of sight from that vantage.
Slope, aspect and watershed analysis is becoming more common in forest management practices because timber companies are able to predict how rivers and streams will be impacted by changes in runoff that often accompany tree removal in drainage basins.
Atterbury predicts that forest management companies will be taking advantage of this image processing capability more frequently in the future. Slope and aspect analysis is a required part of the Bureau of Land Management's (BLM) six-step watershed analysis process, and both the BLM and Forest Service are requiring more of these projects conducted on managed forest lands. Resolution Expands Role of Imagery
Hampton and Atterbury see the role of high-resolution satellite imagery expanding into all aspects of forest inventory as well as into general forest management functions.
In a typical inventory project, such as the one underway on the property managed by Hampton Tree Farms, foresters classify the acreage into type units composed of trees that are the same species and age. That information enables the timber company to calculate accurate board-feet volumes and dollar values.
"We can figure a sustainable harvest level based on that information and better determine the effects of different management strategies due to markets, regulatory or internal constraints," said Creel.
Differentiation of timber types with satellite imagery has traditionally been accomplished by applying automated classification routines to multispectral image scenes, but the IRS-1C panchromatic images have been used to distinguish timber stands through manual interpretation. Atterbury technicians have easily differentiated conifer from deciduous stands simply by viewing the image on screen in TNTmips.
"We have also been able to determine 10-year age breaks for the timber stands by determining the crown size of the trees," said Atterbury's Wright. "With 30-meter resolution imagery, we could only divide the stands at 40-year breaks."
Both firms are anxious to begin using 5-meter Indian imagery and the 1-meter imagery that will soon follow in forest management projects that previously have required ground surveying or aerial photography.
"Imagery assists us in accurately assessing our properties at present so we can do a better job of planning our future," said Creel. About the Author:
Kevin Corbley is freelance writer specializing in remote sensing, GIS and GPS. He may be reached in Denver, Colo., at 303-987-3979 or by e-mail at [email protected]
Back
|
