摘要：Abstract
The purpose of this study is to spatiotemporally
explore the characteristics of urban temperatures
based on multi-temporal satellite data and historical in
situ measurements. As one of the most rapidly urbanized
cities in Canada, Saskatoon (SK) was selected as our study
area. Surface brightness retrieving, Pearson correlation,
linear regression modeling, and buffer analysis were applied
to different satellite datasets. The results indicate
that both Landsat and MODIS data can yield pronounced
estimations of daily air temperature with a significantly
adjusted R2 of 0.803 and 0.518 at the spatial scales of 120m
and 1000 m, respectively. MODIS monthly LST data is
highly suitable for monitoring the trend of monthly urban
air temperature throughout summer (June, July, and August)
due to a high average R2 of 0.8 (P<0.05), especially
for the warmest month (July). Our findings also reveal that
both the Saskatchewan River and urban green spaces have
statistically significant cooling effects on the surrounding
urban surface temperatures within 500 m and 200 m, respectively.
In addition, a multiple linear regression model
with four influential factors as independent variables can
be developed to estimate urban surface temperatures with
a highest adjusted R2 of 0.649 and a lowest standard error
of 0.076.
关键词：Keywords air temperature surface brightness temperature
Landsat MODIS regression modeling buffer
analysis
