Analysis of surface temperature in an urban area using supervised spatial autocorrelation and Moran's I

Landsat ETM + imagery is useful for analyzing urban thermal environments because it provides visible and near-infrared data. However, this approach is limited by the low resolution of Landsat images compared to the unit size of buildings in urban areas, which complicates any attempt to apply these images to the urban scale in terms of district unit plans or architectural plans. This paper describes a methodology that improves the accuracy of surface temperature analysis in urban areas by using satellite imagery and associated land cover data. Land cover types in the city of Goyang, South Korea, were reclassified into biotope types and climatope types. Moran's I was then calculated for the thermal infrared images according to the land cover types. Resampling the Landsat ETM + thermal infrared images according to the pixel content ratio and performing supervised classifications based on biotope and climatope maps improved the accuracy of the surface temperature evaluation in an urban area. Moran's I values of 0.9066 and 0.9279 were found according to the classification system of biotope and climatope maps, respectively. In addition, reclassification using supervised spatial autocorrelation and Moran's I may provide better results than other methods applied to surface temperature evaluations of complex urban areas. These results verified that reclassifying the land cover type based on biotope and climatope maps and determining the spatial autocorrelation within the classified data leads to a more accurate evaluation of the surface temperature distribution by land cover type than observed for any other classification method.

Automated summary

This study improves the accuracy of surface temperature analysis in urban areas by using satellite imagery and land cover data. Reclassifying land cover types and performing supervised classifications based on biotope and climatope maps leads to more accurate surface temperature evaluation.