Lam, DK, Remmel, TK (2010). Effect of configuration on spectral signatures. CANADIAN JOURNAL OF REMOTE SENSING, 36(1), 48-54.
A 4-month controlled laboratory experiment was conducted to determine the impact of spatial configurational change on spectral characteristics of grasses. Eight different planting configurations were examined of two grass seed mixtures (sun and shade grass and Kentucky bluegrass) of equal composition (50% each). Spectral properties were quantified using (i) leaf area index (LAI), (ii) normalized difference vegetation index (NDVI) and soil-adjusted vegetation index (SAVI), and (iii) an unsupervised iterative self-organizing data analysis (ISODATA) classification. Analysis of variance (ANOVA) tests were performed for LAI, NDVI, and SAVI; Tukey's post-hoc tests showed significant differences between random and all other treatments for LAI means, no significant differences for NDVI means, and significant differences between irregularly patchy and regularly patchy treatments for SAVI means. Chi-square tests on the four-class aggregation results for ISODATA indicated that configuration does influence the spectral characteristics that drive image classification. This study reflects upon the importance of configuration in spatial pattern analysis for remote sensing of similar surface features with internally heterogeneous properties.