South Patagonia, Chile, is very isolated and difficult to reach.
In spite of the far reaching claims of satellite providers, technology providers and others who promote the vision of a shrinking world, there are parts of this planet that haven't been mapped, or have been mapped insufficiently. There are regions so dangerous to enter, due to turf wars or other hostilities, that to fly over them at a low altitude to capture aerial data is out of the question. Some areas are covered by clouds most of the time, making aerial photography less than ideal. Satellite imagery is still very expensive.

In our time of great advancements, we have explorers and mapmakers working in remote regions to provide accurate maps of places that have great importance to the entire planet — as sources of ecological data, environmental change, riparian study, etc. Several factors contribute to the success of any of these expeditions — the lifting of Selective Availability by the U.S. Department of Defense, GPS ground control points, digital cameras and GPS and Inertial Navigation System (INS) (INS measures the orientation of an airborne system in three axes), increased accuracy in satellite and aerial imagery. There are situations where the difference of a few meters can mean life or death, such as in the case of mountaineering. If you plan your route up K2 with one meter imagery that is precisely rectified, you can plan better and more safely than if your data is ten meters off and not georeferenced very accurately.

The lifting of the Selective Availability (SA) by the DOD allows GPS to be far more accurate, but by itself does not offer any insurance of accuracy (the accuracy budget from DOD is currently 28.5m). Rather, it opens the doors to a new era of data collection. "We can now get within 5 meters with uncorrected GPS, and no longer require a base station in the area. This eliminates the need for differential corrections in some situations," says geographer Kyle Bohnenstiehl.

Data Collection Tools

PCI Geomatics' senior scientist Philip Cheng and Vice President David Stanley spoke about the difficulties of mapping for a remote region. PCI's two major products, OrthoEngine and Geomatica, allow people to correct images themselves.

In order to create maps for a remote area, Cheng says you need the following: satellite images or aerial photos, some ground control points (GCPs), a digital elevation model (DEM) and a remote sensing software.

For regions such as South America, guerrilla warfare and cloud cover are big deterrents to data collection. There is nowhere to land a helicopter, and flying low can be too dangerous. Recently an oil company wanted to have an area in South America mapped. "We have one customer who wanted us to do the mapping in South America, and wanted the oil lines layered all the way from the mountains to the shore," explains Cheng. "We couldn't use optical SPOT images because the area is cloudy, so they used stereo RADARSAT images, which can penetrate through clouds, to extract the elevations and to create a map of the area."

The advent of satellite imagery has allowed explorers to gather imagery anywhere in the world, which has not been possible with aerial photography. With SPOT images and some DEMs some of PCI's customers can create maps of remote areas, he adds. "Some countries have only one day a year without clouds so you do large scale mapping of the whole country, which is expensive."

In spite of these efforts to make maps, some remote areas are so poorly mapped that GPS information must be collected. "This is where we must stitch five or six images together," says Cheng. "Poor maps can't be used for 10-meter or higher ground control points without being 20-50 meters off. Usually people have to go collect the GCPs," says David Stanley.

LANDSAT TM image of the northern position of the SPI.

In Search of Accuracy

For Kyle Bohnenstiehl, the South Patagonia area that he wanted to explore had no good maps. The Chilean military had 1:500,000 scale maps, pretty coarse and limiting for hikers or explorers. In fact the region, located about two-thirds of the way down Chile, is less populated and "wired" than Antarctica. "There's more of a communication infrastructure in Antarctica," says Bohnenstiehl. "More planes fly overhead between McMurdo U.S. base at Ross Island and the South Pole. South Patagonia is very isolated and quite expensive to reach."

What he found when he returned to the U.S. as leader of the GEOCAP 98-99 team was the 1986 Landsat imagery was the least expensive imagery, but there was a lack of cloud-free imagery for this area. The team wanted to see how much the ice cap had changed over the years, if at all. As Landsat was the best available for the money at the time, Bohnentstiehl and his team decided to go back down to Chile and collect GPS ground control points to rectify the imagery into a coordinate system to produce a map.

In addition to this plan, Bohnenstiehl knew they needed some up to date imagery. He entered a drawing and won a RADARSAT image for anywhere in the world. They could use the 1986 Landsat imagery and the RADARSAT imagery.

"The RADARSAT imagery was very difficult to correct for terrain," explains Bohnenstiehl. "The way the radar sensor works there's a lot of topographic layover. The radar can't look straight down, it has to view at a look angle." These look angles are at an angle of 40-45 degrees so it's like looking at a landscape from a 45 degree angle, thereby introducing a large amount of distortion to the images.

The preparation for the return trip to Chile was predicated by the generation of maps. Bohnenstiehl went to ERDAS to borrow a license for their orthoradar tool. This led to the importation of an existing DEM into the system, using the ephemeris data of the image. The ephemeris data contains the location of the satellite at the time of acquisition, plus other parameters. Using the DEM it will automatically geo-reference the imagery to within about 30 meters.

"We ended up with a nice base map and were then able to rectify the Landsat imagery to the RADARSAT imagery. With the dataset we could to some degree compare changes in the ice, and we could merge the two datasets to get more information."

As with most expeditions, the reality did not match the dream exactly. The team went down during what turned out to be the driest winter in Chile. "The dryness prevented us from getting the ground control points we wanted for the imagery," says Bonnenstiehl. "As a result we decided to put the scientific goals on the side for now, and maybe go down in January with an inflatable boat with a motor on it." With a boat, they can get up into some of the lakes and fjords, capturing ground control points from the water rather than from the ice. "The ice is slow going," he ventures, "in a boat, we hope to be able to hop from fjord to fjord and get a good distribution for ground control points."

This first one-meter resolution image of Sydney, Australia was collected by Space Imaging's IKONOS satellite. It is the first high-resolution satellite image ever taken of an Olympic site. The IKONOS satellite travels four miles-per-second in an orbit 423 miles above Earth.

Better Accuracy Equals Greater Safety

The study has resulted so far in a good base map suitable for climbers, explorers and others who want to visit the area. The group gave out several digital maps to various people who were going down there, plus to the Chilean military and other Chilean government agencies. "The demand for this type of information by the police and Chilean military was very high. Providing this data and imagery to the agencies became more of a focus after the trip," Bonnenstiehl explains.

Steps toward even greater detail and accuracy involve getting a base station established at the farm of local villager Don Juan Nahuel, who owns a seaside cabin in the Patagonia region. This base will allow them to create differential GPS information.

When they go back to Chile, Bohnenstiehl adds, they will establish the permanent base station and tie into the International Global Spatial Reference Network, which will then establish very precise geodetic control close to the ice cap.

"For a long term monitoring project, if you have to orthorectify imagery-correct for terrain, take a satellite image and five control points you pull off a quad sheet map, this is not going to be good enough," Bohnenstiehl claims. "You have to get down there and get the GPS data to rectify it with the GPS coordinate system."

What about the heavily touted "maps of the world" generated by some of the imaging companies? "They have great maps of the world, but if you don't have GPS ground control on it, if it's not tied to a Geodetic control network it is not very useful, particularly for remote areas."

This sentiment is echoed by J.D. Main, system engineer at CompassCom Inc., Englewood, Colorado, in a recent article in Imaging Notes. "Higher quality satellite imagery doesn't eliminate the need for ground control, in fact, as imagery resolution increases, more and more customers are counting on higher spatial accuracy as well."

Collection of ground control points (GCPs) is critical in the creation of IKONOS orthorectified products for high accuracy. Collecting ground control for IKONOS is very different from collecting data for Landsat imagery. IKONOS is higher resolution and requires field surveys conducted with GPS technology. Landsat imagery can be orthorectified using points taken directly from 1:24,000 USGS maps.

CARTERRA Precision by Space Imaging produces images from IKONOS 1-meter resolution panchromatic data that have been radiometricaly corrected and fully orthorectified to meet 1:4,8000 map scale standards, according to Main. This product was developed specifically for state, provincial and municipal government applications.

The product can be used also for international projects. Users can collect ground control points themselves.

Space Imaging requires written reports about how each point has been measured, as well as a sketch and description of the GCP location, in addition to the point file. Points must be chosen and measured carefully so that technicians can find the points in the image and orthorectify the image correctly.

Hi Meadows Fire, Colo. Credit: ""

Mapping Long Term with the National Spatial Reference System

Back home in the U.S., on Hopi Indian Reservation land in Arizona, Bohnenstiehl considers mapping in the long term as well. "The Hopi have been here for thousands of years. When we give someone a property lease, that property lease might be passed down for the next 500 years. We didn't want to use section corners, for example, because they can be moved and destroyed easily. The National Spatial Reference System monuments are set in rock. So it's a long term investment for the tribe."

The mapping office at Hopi has invested over $100,000 in GPS equipment to manage the 2 million acres of land. "we established a network of 24 first order control stations, accurate within a centimeter or so, across the reservation and tied those into the National Spatial Reference System. It's a network of National Geodetic Survey high accuracy control points that exist across the country."

The data collected in the field with GPS is mapped and differentially corrected against their base station, which is a point on that Survey map system. The system is network adjusted at all those points to get a good fit. Any GIS data you get within that system will fit with other GIS data collected by other people using the same system. "The system will be accurate now and a thousand years from now," Bohnenstiehl says.

Before this national infrastructure was in place, geographers were dependent upon manually digitized maps provided by the Bureau of Indian affairs and put on USGS 7 ½ maps. Since then they have bought 18 SPOT images for the entire reservation, collected GPS ground control points across the entire project area (160 kilometers x 160 kilometers), tied it into National Spatial Research System. They then did an aerial triangulation to get the best fit and to make the imagery fit on top of the network reference system. This project was just completed a few months ago.

New Technologies

As technology becomes cheaper and more accessible, the problems with mapping remote regions will hopefully decrease. Software has become easier to use and more accessible, says David Stanley, plus hardware requirements are not as great and the price of both hardware and software has dropped. "Now people can operate software within a few days rather than training on something for awhile," he explains.

One satellite image that does hold some promise is from the space shuttle that recently flew a DEM for the entire planet. Says Kyle Bonnenstiehl: "When that high resolution DEM becomes available to us then we'll really be able to correct for terrain distortion."

Aerial photography will become easier with digital cameras. GPS together with INS can eliminate the need in some cases for GPS ground control points. GPS gives you your location and INS determines the heading of the aircraft. "Sometimes you can just feed the digital camera image into the software system and it corrects it, as long as you have a DEM," says Stanley.

Digital cameras are rapidly evolving with higher resolution and throughputs, and GPS with INS is also growing. Says Stanley, "You can already get good results with GPS and INS and a digital camera."

"The quest for more accurate maps is going to be driven by the consumer," states Kyle Bohnenstiehl. "There are new methods for rectifying imagery now. Earthsat has a proprietary algorithm for rectifying Landsat imagery using an aerial triangulation algorithm, and is also developing a global Landsat mosaic that will be accurate within a pixel."

This can be done anywhere in the world with accuracy to within 30 meters. He adds, "But to get down to a meter and a half we still require a man in the field."

Article reprinted with permission from GISCafe.