Application of 1-M and 4-M resolution satellite data to ecological studies of tropical rain forests

Understanding the current status of the world's tropical rain forests (TRF) can be greatly advanced by global coverage of remotely sensed data at the scale of individual tree crowns. In 1999 the IKONOS satellite began offering worldwide 1-m panchromatic and 4-m multispectral data. Here we show that these data can be used to address diverse aspects of forest ecology and land-use classification in the tropics. Using crowns of emergent trees as control points, we georeferenced a 600-ha subset of IKONOS 1-m and 4-m data from an August 2000 image of the La Selva Biological Station, Costa Rica (root mean square error = 4.3 m). Crown area measured on the image was highly correlated with crown area for the same tree measured from the ground. Using a 1988 aerial photograph as a baseline, all trees >1 m diameter in a long-term study that died over the ensuing 12-year period, and that could be located in the photograph, were detected as missing in the IKONOS image (N = 7). Crown growth for large trees visible on both images averaged 12 m(2)/yr (N = 16). We thus demonstrate that IKONOS imagery can provide data on four variables necessary for doing demographic research: tree size, location, mortality, and growth. Stand basal area, estimated aboveground biomass, and percentage of the canopy >15 m tall for 18 0.5-ha permanent forest inventory plots in old growth were all highly significantly correlated with different indices derived from the IKONOS data. We used summary statistics from the original IKONOS data as well as derived indices to characterize nine areas with well-documented land-use histories. Secondary forests were clearly separable from the other sites. One of the secondary forests was 40 years old, suggesting that IKONOS data can be used to detect significantly older secondary forest than is possible with coarser resolution satellite data. The selectively logged forest was distinguishable by measuring the size of the largest crowns on the 1-m image. This suggests a range of applications for detecting and quantifying biomass degradation due to selective logging and edge effects. Satellite data at 1-m and 4-m resolution make possible a truly global approach to fine spatial resolution remote-sensing studies of TRF ecology and land use.