Remote River Reconnaissance: Space Shuttle Observations Help Conservation Biologists

by Michon Scott · design by Robert Simmon and Michon Scott · February 9, 2007

Hydrologist Bernhard Lehner is from Bavaria—the birthplace, he quips, of beer. College dreams of starting a brewery with friends prompted his choice to study water. While his friends wound up in the brewing business, Lehner stuck with science. After graduation, he went to work as a hydrologist with the conservation group World Wildlife Fund. There, he encountered a paradoxical problem: often, the places he most wanted to study had the poorest maps.

In remote parts of the world, detailed maps are rare at best. Yet, precisely because they haven’t been extensively populated and paved, remote regions contain some of the most intact biodiversity on Earth. With development looming in even Earth’s most remote regions, conservation biologists want to recommend the best places to protect, but without thoroughly knowing an area, they struggle to give governments and policymakers good advice. Lehner’s efforts to solve this problem for a remote corner of the Amazon eventually became a stepping stone for a much more ambitious project: a detailed, digital map of water channels for the entire globe.

  Page 2

Image in title graphic courtesy

  Amazon wildlife

Amazon Stepping Stone

In 2004, Lehner, Robin Abell (his manager at World Wildlife Fund), and their colleagues faced a predicament as they examined the Madre de Dios River in Peru and Bolivia, in the southwest Amazon Basin. Abell, a freshwater conservation biologist, was trying to characterize the habitats in the region to figure out what sort of networks of protected and managed areas would best represent the area’s wide variety of freshwater ecosystems.


Remote tropical forests, including the Amazon, are home to much of the Earth's remaining intact biodiversity. Some examples of the Amazon’s diverse wildlife include numerous species of (left to right) butterflies (© 2005 Tina Carlson), frogs (© 2006 Steve Makin), primates (© 2006 radiospike), trees (© 2006 Shannon Roy), and fish (© 2006 Big-E-Mr-G).

  Madre de Dios photograph

The area was far too big for personal inspection, and existing maps weren’t much help. Detailed maps of the Amazon Basin varied by country, with differing scales and map projections. But wildlife doesn’t care about national borders, and Abell and her collaborators needed a consistent map format. “The best product we had was a USGS [United States Geological Survey] product of 1-kilometer spatial resolution,” Lehner explains. “There was nothing wrong with it, but to derive high-quality river maps, it was just not good enough.”


Abell, Lehner, and their colleagues initially studied watersheds around the Madre de Dios River in South America. (© 2005 damclean.)

  Madre de Dios in South America

Lehner and Abell needed a more detailed data source, and for that Lehner turned to topographic maps based on radar data from an 11-day mission flown by NASA’s Space Shuttle Endeavour in February 2000. Called the Shuttle Radar Topography Mission, or SRTM, for short, the mission produced the most complete high-resolution, digital topographic map of the Earth. The map was a hundred times more detailed than the satellite data Lehner had considered using before.


The Madre de Dios region lies just east of the Andes Mountains in Peru and Bolivia. Attempts to map freshwater habitats in the Madre de Dios watershed laid the foundation for efforts to create a global, digital hydrologic map. (Image by Jesse Allen, NASA Earth Observatory, using data from Natural Earth.)

  Hydro1K versus SRTM

Lehner used SRTM data as the foundation for a digital map of the headwaters of the Madre de Dios, letting the topographic data reveal where rivers would be if he traced the path of water flowing downhill. After combining the digital river map with information on land surface characteristics like geology and vegetation, Lehner, Abell, and their collaborators were able to recommend protecting two continuous corridors from the mouth of the Madre de Dios to its headwaters in the Andes.

Perhaps more significantly, the project showed how valuable space-based topography data could be for understanding remote, poorly mapped regions.

The case study of the Madre de Dios quickly drew admiration from other conservation scientists. “After people saw it,” Lehner recalls, “they kind of pushed me to map rivers and watersheds for the entire Amazon, and even all of South America.”

Pressed to expand his work, however, Lehner gave a surprising answer. He said no.


Previously available hydrological maps of the Madre de Dios watershed (top) were not detailed enough for freshwater-habitat conservation planning in a place as biologically diverse as the Amazon. The Shuttle Radar Topography Mission (SRTM, bottom) provided roughly 100 times the level of detail, allowing scienitsts to map many more rivers and streams (blue lines). (NASA images by Jesse Allen, based on SRTM and Hydro1K data from the U.S. Geological Survey.)


Going Global


Lehner didn’t refuse the requests for more maps because he doubted their importance. He just didn’t feel that World Wildlife Fund had the resources to handle such a big project. Instead, he asked his contacts at U.S. Geological Survey and elsewhere when they would make SRTM-based elevation maps that included stream and river channels. No one had immediate plans. “But at World Wildlife Fund, we still needed this product,” he says.

  Page 1Page 3

Lehner realized that if he and his colleagues were going to get the maps they needed, he would have to assemble them himself. Lehner began making plans to scale up his Amazon Basin case study, adapting geographic information system software and writing his own computer scripts to refine the shuttle topography data. The whole process required significant computing power. Ideally, he would have used a supercomputer but, on a limited budget, he made do with personal computers (PCs) instead.

“To process the globe, my estimate today is that it would take one year of pure processing time on four computers,” he says. As it was, he ran about six PCs, all of them working in parallel. Lehner had to cut the data into pieces for each machine to work on, but the computers still “had to know each other” and occasionally check one another’s data. At one point, the six-computer arrangement went awry. The master drive serving all PCs failed. “I always had backups, but the one key drive was destroyed,” he said. He spent two weeks recovering the data.

Lehner’s final, crucial step involved manually inspecting and correcting the maps, making sure rivers appeared where they should be—and not where they shouldn’t. For example, in flat, partially deforested areas, “rivers” formed along deforested areas and roads because of the difference in elevation between the bare ground and the top of the adjacent forest canopy. He and a small support team made nearly 100,000 manual corrections to improve the maps.

  Bernhard Lehner

Bernhard Lehner had mapped watersheds with elevation data before while studying the Amazon’s Madre de Dios region, but scaling that project up to the global level was a much more significant challenge. (Photo courtesy World Wildlife Fund.)

  Road in rainforest

World Wildlife Fund is calling this global hydrological mapping project “HydroSHEDS.” After hundreds of hours of computing time and thousands of manual corrections, Lehner unveiled a digital map of water channels for South America in April 2006. The other continents should follow in 2007.


To make HydroSHEDS, Lehner and his support team made approximately 100,000 manual corrections of the elevation-based river maps. For example, the difference in elevation between the top of a forest canopy and a road would have appeared to be a potential river channel. (© 2006 rocketdude.)


HydroSHEDS’ Future


With global coverage and continued refinements, Lehner hopes that future releases of HydroSHEDS will serve a wide variety of scientific and practical applications. For example, biologists sampling fish species often use GPS (Global Positioning System) coordinates to locate habitats on maps. If their maps aren’t completely accurate, they might place fish in the wrong water body. That could be a serious problem if, for example, they are trying to identify rivers that support populations of threatened or endangered species. When the scientists Lehner worked with wondered whether this problem could be solved, he recalls, “Everyone said, ‘Not really. We don’t have good enough maps.’ But now they suddenly do.” Using HydroSHEDS, conservation biologists can pinpoint species habitats with unprecedented accuracy.

  Page 2
  Madre de Dios HydroSHEDS image

One of the scientists eager to see HydroSHEDS cover the globe is Ned Gardiner of the American Museum of Natural History, who wrote a user guide for the product. Working with the museum’s ichthyology (fish) department, Gardiner plans to use HydroSHEDS in his own research on Africa’s Congo River. “We’re looking at the evolutionary history of species,” he explains. “We know that every time we go back there, we find more fishes that seem to only be found in the Congo. There used to be a giant inland lake in Africa’s interior during the Pleistocene, but eventually the lake was captured by rivers draining to the coast.” All the fishes that evolved in the lake environment radiated into river environments over time. Though geographically separated now, they share common ancestry. Once HydroSHEDS maps for the Congo become available, museum scientists will use them to analyze river networks in conjunction with DNA data about fishes to trace their evolutionary history.


For scientists trying to collate field observations of fish with the correct rivers and streams on maps, HydroSHEDS is a huge improvement over what was available to them before. This image shows part of the Madre de Dios, with detailed information on elevation, river channels, and watershed boundaries. Elevation is color-coded, with dark green for the lowest elevation and white showing the highest. Stream channels, marked in blue-green, are overlaid on elevation data. Watershed boundaries are marked in gray. (NASA image by Jesse Allen, using HydroSHEDS data from the U.S. Geological Survey.)

  Photograph of Cichlid fish in Africa.  

Able to live in a variety of habitats, several hundred species of cichlid fish have evolved in Africa alone, driven by rapidly changing environments. (© 2005 Lars Plougmann.)

  Photograph of cataract on the Congo River  

Today, the lower Congo River is characterized by high discharge and numerous large cataracts. Cataracts can serve as a barrier between different regions, and such barriers can increase species diversification. (© 2007 Bob Schelly, American Museum of Natural History.)

  Photograph of scientists sampling cichlid species in the Congo  

Scientists conduct a fish species survey along the shores of the Congo. (© 2007 Bob Schelly, American Museum of Natural History.)

  Congo map with species distribution

HydroSHEDS digital maps are a big step up in accuracy, completeness, and consistency for scientists and natural resource managers, but Lehner is the first to acknowledge their limitations. One is that, in pinpointing river channels based on elevation, HydroSHEDS may see rivers where there are none. In other words, it identifies potential river channels. “In a dry region like the Sahara, I still get rivers,” he explains. The product also uses vertical exaggeration. “If I detected something that looked like a valley, I lowered it, which is bad for true elevation but great for keeping the water in the channel.” To determine where channels are actually flowing with water and where they may simply be dry streambeds, the HydroSHEDS maps can be layered with other data sets, such as precipitation and soil moisture. But Lehner hopes that future releases of HydroSHEDS will contain data on actual water, in addition to water channels.


In the Congo River watershed, many fish species that are geographically separated today share a common heritage, having radiated outward from a large lake that occupied Africa’s interior during the Ice Age. By combining fish species surveys and DNA samples with HydroSHEDS, researchers can study the evolutionary relationships of fish from different rivers. (NASA image by Jesse Allen, based on HydroSHEDS and species distribution data from Ned Gardiner, American Museum of Natural History.)

  Modifications to SRTM to make HydroSHEDS

Since starting the development of HydroSHEDS, Lehner has moved on to a professorship at McGill University, but he still works closely with Abell and others at World Wildlife Fund, which will be developing the maps for the rest of the globe.

Yet biodiversity isn’t Lehner’s only interest. Human water use interests him, too. According to the United Nations Environment Programme, two out of every three people may live under water-stressed conditions by 2025. As input to hydrological models, HydroSHEDS can inform decisions about human water use as well as wildlife. “At World Wildlife Fund, focus was stronger on the environment,” Lehner says, “but I don’t think I ever lost track of humans. I think they’re very equal.”

  • Thieme, M., Lehner, B., Abell, R., Hamilton, S.K., Kellndorfer, J., Powell, G., Riveros, J.C. 2007. Freshwater conservation planning in data-poor areas: An example from a remote Amazonian basin (Madre de Dios River, Peru and Bolivia). Biological Conservation. 10.1016/j.biocon.2006.10.054
  • Freshwater from the United Nations Environment Programme. Accessed January 26, 2007.

The creators of HydroSHEDS exaggerated small differences in elevation in raw topographic data (left). This vertical exaggeration helped the mapping software better predict potential stream and river channels (right). The digital maps can be combined with data such as soil moisture and rainfall to verify the existence of rivers and streams in different parts of the globe. (NASA image by Robert Simmon; based on data provided by Nikolai Sindorf, World Wildlife Fund.)