Are FAO estimates accurate enough? The question is best answered on a case-by-case basis. FAO remote sensing estimates are useful for inference at pan-tropical scales, but they are not adequately precise for small regions or individual nations. FAO estimates are accurate, even in the presence of deforestation hot spots, but a patchy sample from a broad region can easily miss a single given hot spot (Tucker and Townshend 2001). A sampling intensity that is sufficiently accurate to estimate the rate of deforestation might not be sufficiently accurate to detect a small change in that rate. More intense sampling would improve FAO estimates, and 100% coverage would be ideal; however, core questions can be well addressed with a 10% sample.
FAO followed proper statistical principles for scientific inference with sampling. This allows construction of confidence intervals and tests of hypothesis, which help assure that conclusions by FAO are reasonable. Rigorous scientific review of those conclusions is necessary and valuable, but the assertion by Tucker does not qualify as such. FAO estimates remain the only scientifically credible information about the global extent of tropical deforestation since 1980.
