Sampling Global Deforestation Databases: The Role of Persistence

Abstract

One of the concerns regarding transformation of land cover in tropical areas has been the large degree of uncertainty associated with both rates of deforestation over time and total deforestation. Special monitoring mechanisms must be taken into consideration if a program toward deforestation control is going to be implemented at the national or regional scale. The premise of this paper is that any attempt to quantify tropical deforestation and deforestation rates — at regional level, by randomly selecting sites within a population of satellite scenes — would require an overwhelming number of samples. This paper suggests a methodological approach for sampling remote sensing databases to be used as part of land use/cover change or joint implementation projects. This paper uses the concept of stratification and persistence as main tools.