Abstract
The relationship between urban land surface material fractions (ULSMFs) and brightness temperature has long attracted attention in research on urban environments. In this paper, a multiple endmember spectral mixture analysis (MESMA) method was applied to extract vegetation-impervious surface-soil (V-I-S) fractions in each pixel, and the surface brightness temperature was derived by using the radiation in the upper atmosphere, on the basis of Landsat 8 images. Then, a clustering analysis, ternary triangular chart (TTC), and a multivariate statistical analysis were applied to ascertain the relationship between the fractions in each pixel and the land surface brightness temperature (LSBT). The hypsometric TTC, as well as the geographical distribution features of the LSBT, revealed that the changes in LSBT were associated with the high fractions of impervious surfaces (or vegetation), in addition to the temperature distribution differences across locations with varying land-cover types. The data fitting results showed that the comprehensive endmember fractions of V-I-S explained 98.6% of fluctuating LSBT, and the impervious surface fraction had a positive impact on the LSBT, whereas the fraction of vegetation had a negative impact on the LSBT.
Highlights
The hybrid spatial distribution pattern of urban land-cover types causes heat islands to be distributed more widely
Because water has a single spectral characteristic and it has a high impact on land surface brightness temperature (LSBT), it was primarily extracted by using the Modified characteristic and it has a high impact on LSBT, it was primarily extracted by using the Modified
We focus on the relationship between urban land surface material fractions (ULSMFs) fractions and LSBT, which means that we classified one pixel into three components and analysed the relationship between the fractions and LSBT, achieving an accuracy of 98.6% in multivariate regression analysis
Summary
The hybrid spatial distribution pattern of urban land-cover types causes heat islands to be distributed more widely. The land-cover types associated with urban areas, such as bare soil, semi-bare soil, and impervious surfaces, have high land surface brightness temperature (LSBT) values [1]. Several researches have reported the effects of urban surficial characteristics on the distribution of heat islands, by using vegetation indexes to indicate the correlation between land-cover types and LSBT [2,3,4,5]. The gap in scales between the OBIA data (regional) and the derived temperature (pixel-wise) can cause unwanted computational errors when obtaining correlation coefficients. By using time-series land-use/cover data, Zhu et al have applied a continuous classification and change detection (CCCD) algorithm to explore the impacts of land-use/cover
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
