|
Remapping
the
Forest
Industry
Using Digital Airborne Multispectral Imagery
Mark W. Brennan, Rosemary Brutico,
Ken Murray, and Joe Pelham
The use of current imagery is essential to support
Geographic Information Systems (GIS) that are used by
major forestry products firms for quantifying the temporal
and spatial aspects of managed forests. This information
allows decision makers to address environmental
compliance, wildlife management, recreational use, harvest
scheduling, and in-field surveillance requirements.
J.D. Irving is a 100-year-old, family-owned company
that manages millions of acres in Canada and Maine. The
firm uses GIS in conjunction with computer simulation
models to “grow” the forest 80 years into the future.
This process involves identifying which areas of the
forest, called blocks, are to be cut each year, over a
25-year period, while making sure that the current forest
diversity is preserved 80 years from now. According to Joe
Pelham, J.D. Irving’s GIS manager, “Our challenge is
to map and keep an up-to-date current state of the forest
so that we know where every stand is and how much volume
is growing on every acre that J.D. Irving Woodlands
manages.”
Imagery Is Essential
Extensive use of imagery in GIS technology is not a
new practice in forest management. Scanned hardcopy photos
have served as a backdrop for more than 20 years. The
sources of imagery now available are varied and include
metric cameras (a camera which creates products that
provide accurate distance measurements), scanners, and
digital airborne and satellite sensors. Furthermore, the
tools for exploiting multispectral imagery have evolved
from stand-alone research environments to integrated
desktop applications. Bands that are commonly collected in
multispectral applications are blue, green, red, and
near-infrared, although custom filters can be specified to
accommodate particular spectral bands of interest.
J.D. Irving has been using a mature GIS system for
at least 15 years to manage the company’s forest
inventory. Using this system, a forester has access to
stand-specific information, contained within a relational
database, which is displayed as an overlay with a digital
orthorectified background image (Figure 1). J.D. Irving
represents its stands and delineates changes in its
forests using a mapping technique that uses a vector array
tied to individual polygons. Its system provides a
querying capability for rapidly accessing information for
a subset of stands.
Harvest Block Mapping
Until recently, J.D. Irving used conventional
film-based aerial photography for mapping its land.
Extracting information from this source using traditional
methods required a number of labor-intensive and costly
steps. J.D. Irving first had to create and deliver paper
maps to an aerial photography contractor for flight
planning. The contractor then collected the photographs
using a mapping camera and produced 9x9 hardcopy prints of
the harvest operations. Photographing every acre where
harvest operations were taking place resulted in thousands
of photographs that had to be processed every year.
Harvest block boundaries were delineated on the
photographs using a stereoscope, and stereo pairs of
photographs were then compiled using a Sketchmaster (a
mapmaking instrument that enables the projection of one
image onto a flat surface or another map) to transfer
polygons onto a Mylar base map. J.D. Irving created the
final GIS layer by digitizing the Mylar base maps and
registering to ground control.
New Technologies: Combining
GPS, IMU, and Multispectral Digital Cameras
A relatively new departure from traditional image
collection and production methods is the integration of
commercial digital cameras, inertial measurement units (IMUs),
and Global Positioning System (GPS) technology. These
technologies, taken together, allow for the rapid
production of digital orthorectified, georegistered,
tonally balanced and mosaicked imagery. GeoVantage, an
information technology company located in Swampscott,
Massachusetts, has commercialized this new approach and
currently fields and operates 22 low-cost, digital
airborne 4-channel (B, G, R, NIR) imaging sensors (Figure
2). These sensors, in conjunction with an accompanying
software suite, speed up and simplify the entire process
from planning flight operations to delivering digital
orthorectified imagery.
GeoVantage’s imagery collection and production
system combines position and attitude data from the GPS
and IMU, respectively, with the imagery and allows for
georegistration of the imagery without pre-surveyed ground
control. With traditional photogrammetry, installing and
maintaining “ground control” over a large area could
cost as much as the rest of the imagery collection
activity. GeoVantage has streamlined the process by
replacing countless manual procedures (often requiring a
specially trained staff) for mission planning, mission
execution, and post-processing with automated procedures
that have proven to be cost effective for many mapping
applications.
The traditional use of color infrared imagery
provides insight into vegetation growth and health that is
otherwise not possible using natural color (B, G, R)
imagery alone. While multispectral imagery offers the
advantage of color, which helps interpret the image, such
data has not been widely used in the past because it has
been cost prohibitive. GeoVantage’s orthorectified
imagery solution, however, offers significant cost
benefits to forest managers who require quick turnaround
and detailed information.
After meeting with GeoVantage, J.D. Irving
immediately saw the benefits of adopting this new digital
image collection and processing technology and was quick
to integrate the process into its forest management
operations. According to Pelham, “Processing imagery is
totally different using GeoVantage’s new technique.
GeoPost, GeoVantage’s software suite, produces
terrain-corrected, orthorectified digital mosaics in a
GIS-ready format in just a few hours after collection. We
bring the image up on a computer screen and it fits
directly over our GIS database. It has allowed us to
eliminate two steps of the process from our old system,
thus saving us a third to half the time that we’ve
previously spent. The result is a huge increase in
productivity.”
Flight Operations
This new approach simplifies flight planning and
flight operations. The compact camera system can be
installed in less than an hour on a Cessna 172 aircraft
(Figure 3). A cockpit LCD mission display shows the
location of the pre-planned sites over a map presentation
and aides the pilot in flying to the correct locations.
Sites to be imaged are established into “groups”
composed of several (up to 10) individual rectangular
imagery blocks with a single imagery block requiring from
2 to 10 flight lines. The actual imagery collection and
management of GPS/inertial data are fully automated. While
on a flight line, which is clearly indicated on an LCD
display, the pilot uses a “steering bar” mounted on
the dash to control the aircraft’s lateral motion to
within several meters of the desired fight line. Flights
are typically conducted at 8,000 feet above the ground to
provide 1.0m-resolution imagery. Missions can be flown
with a solid cloud deck immediately above the flight
altitude. Clouds between 7,500 and 10,000 feet would
preclude traditional collections that operate at higher
altitudes. Upon landing, the pilot can easily de-install
the GeoVantage sensor if the aircraft is required for
other rental services.
Image Processing Simplified
The post-mission processing begins by
off-loading the data from the flight drive onto an
archival storage device. The software suite provides all
utilities for performing the GPS/IMU data processing that
tags each collected image with the precise position and
attitude of the camera. This information is then used to
georegister each of the collected frames. The
georegistration process consists of tracing a “ray”
from the camera center through each of the camera color
cells (pixels) until the ray impacts a 3D height model of
the ground. This represents about 5 million ray tracings
per image that must be computed. The entire computational
process can be completed for each image in about 10
seconds.
Following the georegistration process, each
individual image is mosaicked into a seamless composite
image. This is relatively easy because the intense
computational effort is performed in the single-image
georegistration step. The most critical step at this point
is to address the “tonal balance” of the composite
mosaic and to remove artifacts resulting from taking
images at slightly different times under slightly
differing light conditions. Different solar illumination,
as well as different observation angles, can produce edge
artifacts that are corrected by correlating the respective
“color” intensities in the overlapped regions of the
image set. Figures 4a and 4b show a natural color
composite (R, G, B) and a color infrared (G, R, NIR) after
tonal balancing in the overlapped regions. The additional
contrast and the information it represents shown in the
color infrared images clearly demonstrate the advantage of
including this information in the image set.
J.D. Irving is currently using GeoVantage’s
digital imagery to update the boundaries of its annual
harvest program. The imagery is included in a GIS layer,
as shown in Figures 5a (natural color) and 5b (color
infrared), wherever forest operations have taken place.
GeoVantage’s digital orthorectified imagery helps
J.D. Irving improve its operations and meet its business
objectives while preserving the forests’ eco-system for
generations to come.
Conclusion
The use of multiband digital cameras, in
conjunction with GPS and IMU technology, has resulted in
an unprecedented degree of automation and simplification
of a complex earth observation process. This includes
replacing manual procedures for mission planning, mission
conduct, and post-processing with computerized steps that
have been proven to be effective in field operations. Four
bands are collected simultaneously (R, G, B, NIR) and
allow for producing natural color and color infrared
digital orthorectified imagery at standard resolutions of
.25 meter per pixel to 1.5 meters per pixel.
Orthorectified, georegistered, tonally balanced, and
mosaicked imagery can be entered into the GIS with no
additional processing in order for technicians who are
knowledgeable with forestry practices to immediately
classify the spatial extent and nature of the forest
activities.
About the Authors
Mark W. Brennan is former Director of the Forestry
Programs at GeoVantage Inc.
Joe Pelham is GIS Manager of J.D. Irving,
Limited—Woodlands Division, Saint John, New Brunswick.
He may be reached at [email protected].
Ken Murray is a GIS Forestry GIS Technician at J.D.
Irving, Limited—Woodlands Division, Saint John, New
Brunswick. He may be reached at [email protected].
Rosemary Brutico is Principal of Quintessence
Communication, a marketing research and communications
firm. She may be reached at [email protected].
Visit Geovantage at www.geovantage.com.
Back
|
