Characterizing Landscape Dynamism Using Pane ' '-Pattern Metrics

Pattern metrics were generated from four land-usellend-cover (LULC) classifications derived from Landsat MSS scenes in 197212973, 197511 976, 1979, and 1985. Patch-level statistics were longitudinally assembled into a "life history " of patches in order to assess structural changes over time in landscape composition, and further to infer process in an area reported to have undergone rapid deforestation. Patches were defined using the 197211973 LULC classification and then applied as landscape partition boundaries for calculating pati ern metrics throughout the remaining time series. Pattern metrics tested were double log fractal dimension, interspersionljuxtaposition index, percentage area, and mean patch fmctd dimension. The paneled metrics produced mixed results: some panels exhibited strong association with ecological gradients and anecdotal information regarding landscape processes in the area, while others offered no insight into LULC dynamics at work in the area due to their invariance. Recommendations are made for further assessing the potential of a aneled-pattern metric approach for characterizing lan scape dynamics.