2004 Volume 113 Issue 4 Pages 482-494
Urban Heat Island, a meteorological phenomenon by which the air temperature in an urban area increases beyond that in the suburbs, grows with the progress of urbanization. The difference of air temperatures between city center and suburbs is called Urban Heat Island Intensity (UHII). UHII is usually calculated as a fluctuation of the highest and lowest values of ground-observed air temperatures. However, the magnitude of UHII may vary with the locations of observations. Satellite remote sensing has been expected to be effective for obtaining thermal information of the earth's surface with a high resolution. However, the possibility and the technique for evaluating UHII with brightness temperature (BTUHII) from satellite images have not been verified yet. This paper, taking Tokyo as the study area, aims to clarify a method for calculating BTUHII with Landsat/TM thermal images and verifying its usability with in situ ATUHII. Based on the principle of ATUHII, we consideredthat BTUHII must be estimated with the same spatial resolution as the spatial scale at which air temperatures are formed. This is called the resolution condition. Moreover, the highest and lowest temperatures for BTUHII must come from similar land uses as those for ATUHII. This is called the land use condition. Our previous research has shown that the brightnesstemperatures reach a maximal correlation with air temperatures at a distance of about 600 m in Tokyo. Invoking this result, we improved the spatial resolution of brightness temperature by applying a low path filter to the Landsat/TM night thermal image dated March 1, 1999. To match the land use condition, we regrouped the 10 m high-resolution land use data into 8 land use categories according to the brightness temperature and NDVI. In addition, we divided the study area from the city center to the suburbs into 6 zones : CBD, SubCBD, within 10 km, 10-20 km, 20-30 km, and beyond 30 km to take account of the impact of land uses in differentdistance spheres. The following facts have been successfully confirmed. (1) In an urban context like Tokyo, resampling brightness temperature image into a resolution of 500 m can effectively remove the influence of cooling or heating facilities, and make the spatial structure of UHII much clearer. (2) BTUHII is a little smaller than ATUHII, but they are still comparable. (3) Although BTUHII can not substitute for ATUHII completely, the trend of BTUHII decreasing from the city center to the suburbs is coincident with ATUHII so BTUHII would be an effective indicator for a comparison of thermal attributes between different urban districts or land use categories. We believe that these conclusions have great significance for using satellite sensed thermal images in research on urban climatology and the practice of urban planning.