Infodoc : Sciences, Techniques et Culture

The formation of gaps in the forest canopy is a crucial process in maintaining biodiversity. Some research bas already been dalle conceming gap ecology, but very few have attempted to map the distribution of gaps on a larger sc ale and to relate this to determining environmental factors. ln this research, the study of gap structure and ecology in a tropical rain forest was combined with the development of a remote sensing tool for large scale gap detection. A SPOT 4 satellite image (resolution 20 m) was used. ln the surnrner of 2002 (July 12th - September 25th), a total of Il gaps of various size and age were analyzed and geo-iocated. This thesis was part of a cooperation between the University of Antwerp on one band, and the French CIRAD and IRD institutes on the other band. The fiel[...]

The formation of gaps in the forest canopy is a crucial process in maintaining biodiversity. Some research bas already been dalle conceming gap ecology, but very few have attempted to map the distribution of gaps on a larger sc ale and to relate this to determining environmental factors. ln this research, the study of gap structure and ecology in a tropical rain forest was combined with the development of a remote sensing tool for large scale gap detection. A SPOT 4 satellite image (resolution 20 m) was used. ln the surnrner of 2002 (July 12th - September 25th), a total of Il gaps of various size and age were analyzed and geo-iocated. This thesis was part of a cooperation between the University of Antwerp on one band, and the French CIRAD and IRD institutes on the other band. The field work was performed in the humid tropical Tain forest ofCounami (12 000 ha), French Guiana, in the northem Amazon (5°21' N, 53°15' W). The surface areas of Il gaps were determined (both at canopy level and at ground level), gap undergrowth was classified (density and height) and gap leaf area index was measured. The gap creating tree species were identified, gap age was estimated, and the topographic and geologic gap environment were described. Mapping and image processing software were combined and allowed for canopy gaps to be accurately reconstructed. An image filter was developed to distinguish canopy gaps from surrounding dense forest. It was found that the gap area at canopy level had to potentially cover 25 to 50% of pixel area to be detected. ln absolute figures, this corresponded to a minimal gap size-threshold was of 100 to 200 m2. The number of "gap- pixels" held back by the image filter was a function of gap undergrowth density and height (thus of gap age), of topographic disturbances, of gap shape and its position relative to image pixel positions. Several confusions with non-gap elements occurred (human disturbances, hilly areas and groups of Arecaceae), and research was further hampered by GPS imprecisions, image geo-iocation imprecisions and a time lag ofnine months between the moment the satellite image had been taken and the actual fieldwork was carried out. Due to these problems, no reliable estimation of the spatial distribution of canopy gaps throughout the forest could be made. However, a useful tool was developed which can be fine-tuned in the future research on tropical forest dynamics.