Multi-Sensor Satellite Data And GIS For Land-use/Land-Cover Mapping In The Rural-Urban Fringe Of The Greater Toronto Area

Mosharef Ali Mir
Student
York University, Canada
mmali001@yahoo.com


Abstract :
This research investigates the capability of the multisensor satellite data (i.e. Landsat TM, ETM+ and IRS-1D LISS III) for extracting landuse/land-cover information in the rural-urban fringe of the GTA using various image fusion methods.

For image fusion, seven fusion techniques, specifically Intensity-Hue-Saturation, Principal Component Analysis, Combined Principal Component Analysis and Intensity-Hue-Saturation, Image Lock Data Fusion, Multiplicative Model, Brovey Transform, and High-Pass Filtering Model were tested. . Landuse/land-cover information was extracted from original and fused images using maximum likelihood classification algorithm with a sixteen-class landuse/land-cover classes adapted from USGS landuse/land-cover classification scheme. Kappa coefficients, user’s, producer’s, and overall accuracies were calculated and compared for image classifications.

The results demonstrate that, for identifying landuse/land-cover classes, the PCA technique for combining ETM+ and IRS-1D MS & PAN data yielded the best results with the Kappa of 0.90 and the overall accuracy of 91.81% compared to other techniques and images, while the BT and High Pass Filtering method yielded poorest results.

Ongoing monitoring of the study area using satellite remote sensing and GIS would be of real value for decision-making aimed at conserving the ecological integrity and sustainable management of the GTA.