Identification of Land Surface Variables Affecting Climate

Satellite remote sensing is capable of measuring land surface properties in the spectral domain allowing the identification and mapping of different land-use/land cover features. In this research, field measurements carried out along five different neighborhoods in the Manhattan borough of New York City, are compared with land cover classification products developed by supervised (Maximum likelihood) and unsupervised (K-Means) classifications of the Landsat imagery. The supervised classification maps are found to be very close to the actual features on the earth surface. However, the unsupervised classification maps showed various discrepancies when compared to the supervised classification maps, and the actual features on the ground, thus resulting in a poor method for land-cover characterization. According to the field measurements that were taken during the 2011 walks, the neighborhood with the highest temperature was 14th Street. On the other hand, land cover classification maps for both maximum likelihood and K-Means showed that the neighborhood with the greatest percentage of urban areas is 14th street with 93.6% and 20.6% respectively or urbanized areas. This result shows that urban areas like roads and buildings have the most influence to the increase of the local temperature.