Abstract : While French Guiana forms a part of the biologically diverse Guiana Shield ecoregion, information on the distribution of forest types within the territory is limited, despite a number of global-scale remote sensing studies conducted over the past 2 decades. Those studies essentially depict French Guiana's mass of forests as one large " green carpet, " instead of identifying the various forest types present within the territory. Nonetheless, with forest monitoring a key element of the UNFCCC's Reduced Emissions from Deforestation & forest Degradation (REDD+) initiative, renewed attention is placed on assessing French Guiana's forest ecosystems. Building on earlier efforts, this study drills down into French Guiana's " green carpet, " making use of monthly reflectance estimates derived from 15-years of data from the MODerate resolution Imaging Spectroradiometer (MODIS). With French Guiana as a case study for other tropical forest territories, this paper thus documents some of the challenges to using remote sensing for the purposes of inventorying and monitoring forests. It is shown that both the separability of identified forest types and the number of spectrally distinct forest types varies month to month, indicating that certain months may be more optimal than others for mapping forest types
