Doboku Gakkai Ronbunshuu G
Online ISSN : 1880-6082
ISSN-L : 1880-6082
Paper (In English)
PRACTICAL IDENTIFICATION METHOD FOR VEGETATION IN URBAN REGION USING TWO SPECTRAL EDGE IN HIGH RESOLUTION SATELLITE IMAGERY
Jinyoung KIMHiroaki FURUMAIToshiya ARAMAKI
Author information
JOURNAL FREE ACCESS

2010 Volume 66 Issue 1 Pages 46-55

Details
Abstract
  The accurate information on vegetation-covered area as a pervious surface is necessary to improve the accuracy of runoff analysis for non-point pollutant loading and groundwater recharge in urban region. In this study, practical bisection method for distinguishing vegetation from non-vegetation in urban regions was proposed. The Green Bias Index (GBI) which is derived from the strong relationship between the red edge and green edge in high resolution satellite image stands for the pigment property of vegetation as the bias terms of the green reflectance between near-infrared and red reflectance. Through the application of the GBI in study area, the vegetation higher than 99.7% to ground truth data could be identified, while the misidentification rate from non-vegetation is lower than 0.2% and 0.1% in roof and road, respectively. The width error between identified and measured vegetations was zero to one pixel for eleven sites selected in the study area. Finally, the spatial and temporal variance of GBI was investigated with five satellite images obtained on different date and location.
Content from these authors
© 2010 by Japan Society of Civil Engineers
Previous article
feedback
Top