Sensitivity of landscape measurements to changing grain size for fine-scale design and management

Landscape pattern quantities are affected by issues of scale, namely extent and resolution. The grain size (resolution) of fine-resolution geographic information system (GIS) data for two highly fragmented landscapes in USA and Italy were altered to evaluate the effect of grain size changes on landscape pattern metrics and cost-surface model outputs. Beginning with 3 m resolution data and resampling the data to 300 m resolution, we applied pattern metrics and cost-surface models (available in GIS software) and evaluated the types of behaviors in resulting quantities. Results showed that some pattern metrics are robust to changes in grain size (such as area, cohesion, interspersion and juxtaposition metrics), while others exhibit staircase-like or erratic responses. Compared to previous studies, we identified behavioral responses that differ from grain-size changes at coarser resolutions. Cost-surface models demonstrated robust or consistent responses to grain size changes in most cases. For both types of pattern measurement, however, we found that behaviors could differ contextually; that is, there could be different types of behaviors for different landscapes, classifications, or grain sizes. Results indicate that comparing spatial data collected at different scales (such as historical data to more recent, high-resolution sensed data) is complicated by different types of responses to changes in grain size. This may limit the applicability of tools for the measurement of landscape change over time if landscapes are represented by differently scaled data.