A remote sensing technique for assessing plant species diversity
Abstract
Earth science has developed over many centuries as a dynamic and ever evolving research field. In the last decade, earth scientists have postulated many thought provoking theories, tools, models and techniques criticval to ecological problems and plant species preservation. The primary objective of this study was to test whether there is a relationship between variations in 35mm color aerial photographs' pixel brightness and total numbers of plant species in selected prairie preserves In northwest Minnesota, a region dominated by typical tallgrass prairie and bordering the prairie-forest ecotone. Seventeen prairie preserves lie on the Minnesota side of the Red River of the North's drainage basin which roughly coincides with the area once covered by glacial Lake Agassiz and are protected by The Nature Conservancy-Minnesota Chapter in conjunction with the Minnesota Department of Natural Resources. Standard 35mm color aerial photographs were collected to measure pixel brightness variation via density slicing with the Measuronics II LMS system. Regression analyses were performed in an attempt to develop a prediction equation and account for the variability of NSPECIES with various combinations of predictor variables. Partial correlations were performed with variables controlling for AREA. A full model multiple regression was performed with all 15 predictor variables. A new approach was undertaken where a new full model regression analysis with AREA and all pixel variables together accounted for 94.035 percent of the variability of NSPECIES, with a confidence of approximately 94 percent. Use of a Venn diagram demonstrated that although 88.18 percent of the variability of NSPECIES was accounted for by the pixel brightness data, as a result of the multiple regressions, the influence or overlap interaction of AREA could be subtracted. Equation B, determined as the best equation since the mathematical relationship accounted for 94 percent of the variability of NSPECIES. In conclusion, prediction Equation B demonstrated that a good relation exists between pixel brightness variation from aerial photographs and corresponding numbers of plant species for tallgrass prairie preserves, and that such pixel data can be used to predict numbers of plant species