Spatial analyses of logging impacts in Amazonia using remotely sensed data

Performances of selected spatial methods are investigated for characterizing canopy disturbance in a reduced-impact logging operation in central Amazonia using Landsat-7 ETM+ and Ikonos visible, near-infrared, and normalized difference vegetation index data. Texture, fractal dimension (D), and Moran's I index of spatial autocorrelation were calculated for (1) 10-ha plots representing logged (LF), logged excluding major roads and patios (L), and old-growth (OG) forest; and (2) 335-ha plots representing LF and OG. Ikonos data were sensitive to roads, patios, and some logging gaps, whereas ETM+ data were only sensitive to major logging features. The spatial methods were effective at characterizing the different logging feature treatments at both plot sizes; D_TPSA and Moran's I were most sensitive to fine-scale surface details. The spatial methods show potential for monitoring and management of logging activities over landscape scales. The importance of scale, given the ever-increasing choice of remotely sensed data, is emphasized.