Using technology developed in the 1970s, tropical deforestation on a gross scale has been monitored from satellites for more than 25 years. Known as Landsat analysis, these maps can reveal the widespread destruction of forested areas. Often, these disturbances are the result of a practice known as clear-cutting, the wholesale removal of trees and vegetation for agricultural purposes or cattle-ranching.

But tropical rainforests also face complex threats that are invisible to these satellite studies. Some are natural: tree mortality and regrowth, insect and pathogen outbreaks, small but intense weather events. However, others have human origins. For example, selective timber harvesting has become common throughout humid tropical forests. Sometimes it is done legally for conservation purposes, but when it is poorly managed or takes place illegally the forest structure is weakened, making tropical forests more vulnerable to large-scale disruptions like fire or high winds.

Scientist Greg Asner and his colleagues in the Carnegie Institution’s Department of Global Ecology at Stanford University have developed a tool to extract better data from Landsat images. The Carnegie Landsat Analysis System (CLAS) uses automated image analysis and pattern-recognition to examine satellite photographs pixel by pixel, a scale that corresponds to parcels of land measuring about 100 square feet. Through these minute computations, the program can detect fine differences in vegetation patterns. Run on supercomputers in Asner’s lab, CLAS rapidly analyzes millions of pixels in succession. When aggregated, these calculations produce a detailed map of forest conditions over large areas, provide exact damage measurements, and identify zones of special concern.

Tested in the Brazilian Amazon, CLAS showed that traditional remote sensing methods miss 50 to 80 percent of damage resulting from “diffuse” disturbances like unmanaged timber harvests. Forests that appeared intact turn out to be in great danger. Information produced with CLAS has forced scientists to double their estimates of Brazil’s already alarming deforestation rate. With MacArthur support, Asner and his colleagues are speeding up the processing time, refining the analysis of satellite imagery, and developing simplified desk- and laptop versions for regulators and researchers. The Carnegie team will test the improved CLAS in a swath of Peruvian rainforest, but the tool will enhance forest management worldwide.

Carnegie Landsat Analysis System >>

Department of Global Ecology at Stanford University >>

Conservation Grantmaking>>

Geographic Priorities>>

Newsletter: Global Conservation>>

Grantmaking Team>>