Our ongoing work with Conservation International has led to the first regional forest cover change map that has been derived using a single consistent methodology. The details of the study have not yet been published. However as we work on validation and interpretation of the results a pattern has become very clear.
Previous deforestation studies in Chiapas have usually concentrated attention on well defined study areas. This could produce the impression that deforestation is a homogeneous process over the whole state. In fact study areas selected for deforestation analysis are usually those with the highest rates when compared to the rest of the region. This is quite natural. Many of the areas where deforestation is no longer occurring have already lost a large proportion of their forest cover. However the overall regional deforestation we have quantified is considerably lower than previous studies imply.
Where deforestation has followed the classic pattern of forest conversion to permanent agriculture or pasture the CI methodology, which is based on Landsat imagery, has provided a remarkably good match with high resolution imagery. However where chronic, low level forest disturbance takes place the overall impact of human activities are much less easily quantified at the resolution of Landsat imagery. The difficulty in accurately evaluating forest cover change increases in areas of dry forest. Nevertheless regional patterns are robust.
The clearest deforestation hotspot in the state of Chiapas remains the Marques de Comillas area in the Southern Lacandon. Deforestation and carbon sequestration in this area has previously been studied in some detail by De Jong et al 2000
The two images below are animated gif files which change when clicked on to enlarge them to full size. The show clearly how Landsat based deforestation analysis coincides in this area with the conclusions drawn from visual analysis of recent high resolution imagery. The visual analysis is produced by overlaying our change analysis in Google Earth using Geoserver, and through the use of QGis.