PARENT SESSION
Special Session 8: Spatial statistics at multiple scales
Organized by: P Legendre and M Fortin
Tuesday, August 9, 8:00 AM - 11:30 AM, Exhibit Hall 210a-e, Level 2, Palais des congrès de Montréal
Comparing and classifying multi-scale spatial patterns.
Ollier, Sébastien*,1, Pavoine, Sandrine1, Couteron, Pierre2, 3, 1 Université Claude Bernard, Lyon, France2 Institut français de Pondichery, Pondichery, India3 CIRAD - Amis, Montpellier, France
ABSTRACT- Concepts of spatial pattern and scale play a central role in the study of both ecological and land surface processes. Indeed, as both biological and physical processes generally exert an influence on a broad range of spatial scales, pattern characterization may valuably contribute to an understanding of natural phenomena by pointing out for which scales a given set of variables display its highest variability. Several methods have been developed to investigate scales of heterogeneity in ecology. All these methods characterize a pattern observed in a particular sampled area, for instance along a transect or within a grid, by studying the relationship between variance and scale. What has been largely missing until now is a general framework allowing multi-scale comparisons between many sampling units in which spatial patterns are observed and quantified via a particular method of pattern analysis. By revisiting several well-established methods for pattern quantification, namely the Two-Terms Local Variance analysis, the Mean-Square-Block-Size analysis, spectral and wavelet analysis, we propose a multivariate approach allowing consistent comparison and classification of multi-scale patterns. The need of a standardization of pattern intensities for different scales is emphasised. Indeed, as large-scale features are likely to have a higher variance than small-scale features, information at small scale is likely to remain blurry in the absence of any preliminary standardization. Two illustrations are provided on 1-D and 2-D sampling units. We first illustrate our approach by comparing and classifying 257 laser altimeter profiles, with a length of 64 measure points, which were collected on three main landscape units with distinct geomorphologic and ecological characteristics. The second illustration deals with a set of 192 multi-temporal digitized aerial photographs (315m by 315m) used to quantify diachronic changes in spatial patterns displayed by semi-arid periodic vegetation.
Key words: multi-scale spatial pattern, multivariate classification, laser latimeter profile, digitized aerial photograph