Urban Heat Island in Summer of Nanjing Based on TM Data

Yili Lu,Xuezhi Feng,Pengfeng Xiao,Chenglei Shen,Jia Sun

DOI: https://doi.org/10.1109/urs.2009.5137628

2009-01-01

Abstract:When the area covered by buildings increases a lot, the heat island effect appears, that is the phenomenon that temperature of urban area is higher than suburb, and central area has higher temperature than surrounding areas. Heat island effect is the direct representation of worse urban environment. It is not a single and independent temperature index since it relates to a variety of factors. Since the 1970s, remote sensing technology has been applied to measure land surface temperature in the research on heat island effect. This paper takes Nanjing main city as the research area and chooses Landsat TM image in 2003 as the research data. The paper makes using of NDVI to extract the green land from Nanjing urban area. NDBI is also introduced in to extract the built-up land from the urban area. This paper then utilizes mono-window algorithm (created by Z. H Qin) to retrieve land surface temperature and brightness temperature. In order to understand the relationship among these special indices and learn their influence on heat island effect, .correlation analysis is done among land surface temperature, NDBI and NDVI. These correlation coefficients indicate that the influence of buildings on the heat island effect is positive; in contrast vegetation plays an important role in weakening the heat island effect. In order to further learn the influence of different types of underlying surface on the surface land temperature of Nanjing urban area, this paper applies supervised classification to the research area. According to the results, this paper classifies Nanjing urban area as four types of land: high density built-up land, low density built-up land, green land and water body. By analyzing the correlation between these two indices and different underlying surface temperature, the paper gets the correlation matrix and reaches an expected conclusion.