OPTIMAL SPATIAL RESOLUTION FOR APPALACHIAN HARDWOOD FOREST IMAGES

摘要

The aim of the study was to determine the optimal spatial resolution through analysis of the spatial autocorrelation and the semivariance of optical remote sensing images of Appalachian Hardwoods (Eastern USA) acquired with Airborne Multispectral Digital Camera Systems (ADAR 1000, produced by Positive Systems, Montana, USA). The images acquired have a very high spatial resolution, with 0.18 m pixel size. The methods applied consist of measurements of spatial autocorrelation made on spatial profiles of the imagery and of measurements of semivariance made on matrix files (images). As a control method, measurements were made in 181 plots, each 400 square meters, where were measured diameter at breast height and crown diameter.The optimum value of spatial resolution based on autocorrelation measurements was 6.4 meters. The average value of the range of the semivariograms was 6.7 meters for Rook case type of measurement (along pixels' edges) and 7 meters for Bishop case type of measurement (along pixels' diagonal). From the relationship between diameter at breast height and crown diameter of the field plots suggest an average crown diameter of 6.9 meters. From this relationship, can be determined the average diameter at breast height as a function of the spatial resolution derived from autocorrelation and semivariance measurements.Consequently, the optimal spatial resolution for these Appalachian Hardwoods plots is between 6 and 7 meters, and the results could be used successfully for producing images with optimal resolution that would give high accuracy without using images of overwhelming sizes.