The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
NDVI-NDVI mil 
NDVI veg -NDVI soil 
(8) 
NDVI 
Where, veg is the NDVI 0 f a p ure vegetation pixel 
while I so ‘ l is the NDVI of a pure soil pixel. 
NDVI 
In theory, sod should be zero for most soil types, but it 
changes from -0.1 to 0.2[11,12] because of the influences of 
many factors. veg should be the maximum of ND VI, 
but it will change with the spatial or temporal change because 
NDVI 
of the influences of vegetation types. Thus, veg and 
NDVI 
so ‘' can not be fixed values [13] even in the same image. 
Vegetation type changes with the change of the land use type. 
To the same land use type, vegetation types are same 
NDVI 
approximately [14], so the pixels’ veg are close to the 
NDVI 
same vegetation type; the pixels’ so ‘ l are close to the 
same soil type too. Thus, the land use map and the soil map 
NDVI NDVI 
may be used to compute veg and so,i . 
temperature can be seemed as island and be easily recognized. 
Meanwhile, there was not a clear center of UHI in the center of 
the urban. There structure of UHI was complex inlaying and 
showed a pattern of distribution centers. 
Table 1 represents the mean brightness of temperature of 
Caidian, Hanna, Jiangxia, the center urban and the whole 
region and the URI of the whole region for 14 years. It was 
found that the mean brightness of temperature of the whole 
region was increased from 13.37°C to 22.98°C. As the 
development of urbanization, the quality of hot environment of 
the center urban also degraded seriously. The mean brightness 
temperature was increased in the study districts, especially in 
the center urban. Comparison between the mean temperature of 
Oct 26, 1988 and Oct 13, 2002 in the center urban shows that 
the values increased 9.12 °C. 
To analysis the change of UHI, Urban Heat Index (URI) was 
introduced, which was based on the ratio of UHI area to urban 
area. The greater the index is, the more intense of the UHI is. 
The trend of UHI showed a obvious trend: as the urban area 
expanded, the URI of Wuhan consistent increased, especially in 
1991. The URI from 1988 to 2002 were 0.134, 0.153and 0.170, 
respectively. 
3.2 Characteristic of Vegetation Fraction and its variation 
3. RESULTS AND DISCUSSION 
3.1 Characteristic of LST and temperature variation 
The brightness temperature of Wuhan from 1988 to 2002 could 
be derived using formula (2) and (3). In order to minimize the 
impact of the time difference, the temperature was normalized 
to 0~1. Fig.2 shows the distribution of temperature 
the mean brightness of temperature(°C) 
1988 
1991 
1996 
2002 
Caidian 
13.69 
17.62 
21.23 
22.42 
Hannan 
13.19 
17.31 
21.33 
23.16 
Jiangxia 
13.75 
17.96 
21.96 
23.08 
center urban 
14.14 
18.06 
22.06 
23.26 
whole region 
13.37 
17.52 
21.64 
22.98 
URI of 
whole region 
0.134 
0.153 
0.158 
0.170 
Table 1 the mean brightness of temperature of different district 
and the URI of the whole region 
This paper estimates vegetation fraction of Wuhan based on the 
NDVI me thod for dimidiate pixel model. First, NDVI 0 f ¡ t 
is computed; then, the land use map and the soil map of the 
study area were used as reference to make a statistic of the 
whole NDVI va i ue According to the frequency statistical 
table, the NDVI frequency of 5% in the soil map was used 
NDVI 
110,1 , and that of 95% in the land use map was used as 
as 
NDVI NDVI NDVI 
veg ; at last, sod and veg which obtained 
above were taken into equation (9). So we acquired the 
vegetation fraction map of the study area. 
Figure.2 The normalized temperature distribution in 
the study area onOct 26, 1988(a), Oct 23, 1991(b), 
Oct 4, 1996(c) and Oct 13, 2002(d) 
maps, from which we can see that the UHI had an increase 
trend in the process of fluctuations for the past 14 years, 
especially in the center urban. From 1996, the land surface 
temperature of the urban, n which there is industrial areas, new 
urban and densely populated residential areas, as much higher 
than that of the suburb, and the area of high land surface 
From the Fig 3, we can see that vegetation fraction in the center 
urban, where performed as the center, was lower than that in 
suburb. During the 14years, the vegetation fraction was 
highest in 1988 and lowest in 2002. 
Table 2 shows the change of vegetation fraction in the 
multi-temporal. From 1988 to 1991, the average vegetation 
fraction of the whole region has dropped, but the overall 
downward trend is not very clear, only dropped 2.2%. The 
vegetation fraction in the center urban decreased significantly, 
1381