Building a Methodological Reference Framework for Quantifying Tropical Deforestation with Remote Sensing

Deforestation is a major threat to the sustainability of natural resources. Thus, adequate estimates of deforestation are crucial for evaluating how sustainable programs are implemented. Still, there is controversy in estimating deforestation, as different estimates often produce contrasting or even conflicting results. It is known that variation in estimates depends on a wide diversity of variables that modify the methods for measuring deforestation, such as scale, types and complexity of vegetation, the definition used, and available inputs of information. This study developed a methodological tool to select the most suitable remote sensing method to measure deforestation in tropical forests. We conducted a systematic review of peer-reviewed publications quantifying deforestation with remote sensing and field data. The information was analyzed and synthesized to build a methodological framework of reference. The methodological and descriptive information of the selected publications served to construct four decision rules (excluding factors, classifier options, elements for choosing a classification, and additional information) for selecting a method for quantifying deforestation. We tested the functionality of this methodological framework of reference by quantifying the deforestation of tropical rainforests in southern Mexico. Based on the decision rules of the framework, two deforestation quantification classifiers were used in the study area (Maximum likelihood and spectral angle mapper (SAM)). We observed that Maximum likelihood had higher values of accuracy than SAM, although both values of accuracy were acceptable. This framework facilitates the selection of remote sensing methods for measuring deforestation by considering the characteristics of each study area and the available inputs. The use of this framework reduced the uncertainty in the estimates of deforestation by controlling a greater number of variables and provided a robust approach for adequately implementing sustainable programs in these threatened rainforests.