Brazil's clear-cut deforestation rate led the world just five years ago. And between 1995 and 2006 an average of 19,497 square kilometers of forest was cleared in the Amazon annually, or an area equal to that covered by roughly 3.5 million American football fields.*

Starting in 2006, however, this trend reversed, and the figures started plummeting. Within four years the Brazilian government recorded a 60 percent drop in the nation's annual rate of deforestation; in 2010 it reached its lowest level since 1988. The 2010 figure—6,498 square kilometers—raised hopes that Brazil would fulfill the pledge in its 2009 National Law on Climate Change to reduce Amazon rainforest destruction 80 percent by 2020.

Evidence now suggests that continued reductions in deforestation will be hard to come by. Why? Because clear-cutters have figured out how to defeat detection efforts by targeting small parcels that are less likely to be observed by the satellites currently in use to monitor illegal logging. "The challenge has become much more difficult," says engineer Gilberto Camara, head of INPE, Brazil's National Institute for Space Research.

Camara has repeatedly noted that patterns of Amazon deforestation have changed dramatically in recent years. In 2002 35 percent of the total area cleared of trees consisted of land swaths smaller than 0.5 square kilometer, whereas by 2010, the contribution of these small areas to the overall deforestation number increased to 80 percent. In other words, although the total number of felled trees was reduced, the smaller areas that are still being cleared have increased in number.

Smaller areas are more difficult for monitors to detect, but new space-based, remote-sensing technology can help—something Brazil is quite familiar with. Since the 1980s the country has relied on satellite images to measure environmental changes in the Amazon. In 2004 the Real-Time Deforestation Detection System (or DETER, after the name in Portuguese) came on line. Using daily imagery obtained via the MODIS (Moderate-Resolution Imaging Spectroradiometer) instruments on both NASA's Terra and Aqua satellites, and combining that with data collected by the WFI (Wide-Field Imager) sensor on board the China–Brazil Earth Resources, or CBERS, satellite, INPE manages to issue every fortnight an assessment of areas at risk for deforestation.

Those bi-weekly alerts have guided the nation's environmental police, and enforcement operations have doubled in the region in the past five years. Farmers and loggers cutting down areas as large as 2,000 hectares have received hefty fines, and their properties, machinery, timber and/or cattle have been confiscated. "No one dares to chop down large areas anymore because they know they will get caught," says geologist and geographic information system (GIS) researcher Britaldo Soares at the Federal University of Minas Gerais's Center for Remote Sensing. The result, however, is that deforestation has become much more diffuse and difficult to halt.

The smaller parcels also mean that farmers who engage in slash-and-burn subsistence agriculture now conduct much of the clearing, rather than industrial-scale outfits. Meanwhile, the "bad guys"—big cattle ranchers and loggers—know that satellites' optical instruments rarely detect small patches of destruction or even see them at all beneath cloud cover. So, loggers are adapting their methods to avoid detection.

Camara says that the country now needs a new real-time system with 20 times more spatial resolution than the current one to spot areas down to 0.2 square kilometer. "We are missing a great deal of degradation going under the canopy," he explains. Both sensors used today—Modis and WFI—resolve objects down to 250 meters across. The ideal would be something between 20 and 70 meters. Some existing satellite technology can provide images with such resolution, but they take longer to obtain (Landsat provides images every 16 days) and are too expensive, because they are commercial ventures.

An immediate solution that offers a new real-time system could come soon via a partnership with India. ResourceSAT, flown by the Indian Space Research Organization, offers a sensor called Advanced Wide Field Sensor (AWiFS), capable of yielding images of 56 meters of resolution every five days. These images will be deployed during the next deforestation season, which usually runs during the dry months in the Amazon—May to October.

For a long-term solution Brazil plans to launch a new satellite next year called Amazonia 1, which will supply images for DETER. Moreover, the partnership with China will likely continue, and the two nations are considering the development of a new family of satellites using radar instead of optical instruments, enabling the observation of forests even through clouds.

At times, cloud cover allows just 30 percent of the Amazon to be seen. Radar imaging can overcome this barrier because it beams a radio signal toward Earth's surface and receives a return signal, rather than relying on optical light reflected from the terrestrial surface.

But even the best satellite surveillance will be worthless without changes on the ground as well. Although the Brazilian Amazon has seen a large number of police operations in recent years against illegal logging, very little action has been taken to tackle deleterious agricultural practices. Slash-and-burn cattle ranching and farming requires the land be prepared for cultivation with fire in lieu of proper pasture and crop management that uses tractors to plow the soil. And almost no law enforcement targets those occupying frontier land, which the government grants to poor families and small agricultural producers under land reform programs.

Moreover, the global demand for food (especially animal protein) will increase in coming years, putting pressure for more arable land on the Amazon, seen as one of the last frontiers for industrial agriculture worldwide. "This is the moment to not let the conquest of low deforestation to be lost," says Soares, adding that the government should engage with the agricultural sector to find a common ground around sustainable practices and to limit expansion.

Carlos Souza, a GIS expert and executive director of IMAZON, a charity involved with monitoring the Amazon, thinks that more control over cultivation practices can be achieved with a system that predicts the most vulnerable areas for deforestation. He has developed a computer model that combines remote-sensing information with factors such as climate, soil type, settlement locations, deforestation history and presence of new roads to indicate areas at high risk for being cleared. This data, he says, may be the key to staying one step ahead of the loggers and ranchers who would otherwise tear down the forest. Such foresight, Souza says, is exactly what is currently missing from Brazilian efforts to combat deforestation.

» View the Slide Show

*Correction (4/12/2011): This sentence was edited after posting to correct the area size comparison.