Remotely sensed data with different spatial resolutions are being used as the primary information source for the analysis of forest fragmentation. However, there is currently a lack of appropriate methods that allow for the comparison of forest fragmentation estimates across various
spatial scales. To provide insights into this problem we analyzed a forested study area in central Spain and a set of 10 widely used fragmentation indices. Forests were mapped from two simultaneously gathered satellite images with different spatial resolutions, 30 m (Landsat-TM) and 188 m
(IRS-WiFS). TM forest data were transferred to WiFS resolution through different aggregation rules and compared with actual WiFS data. We found that incorporating sensor point spread function (which replicates the real way in which remote sensors acquire radiation from the ground) greatly
improved comparability of forest fragmentation indices. We found a poor performance of power scaling laws for estimating forest fragmentation at finer spatial resolutions, and suggest that the true accuracy and practical utility of these scaling functions may have been overestimated in previous
literature. Finally, we report an unstable behavior of three cell-based fragmentation indices (clumpiness, aggregation, and patch cohesion indices), for which spuriously high values can be obtained by resampling forest data to finer spatial resolutions. We believe that the results and guidelines
provided may significantly contribute to an adequate analysis and comparison across scales of forest fragmentation estimations. FOR. SCI. 51(1):51–63.
No Supplementary Data
No Article Media