5 Remote Sensing in Hydrological Modeling
95
IX)
I
C\J
/-1
'/
'I
10
'I
"""'
C\J
'I
~
'I
'-'
/
~ ~
/
.2
E!
C\J
I
8.
I
I
,
E II) ~
I
I-<
,
.~
{
(
0
I
C\J
/
AVHRR
/
IX)
/
- - - - - - - -
Actual
....
,-, I
, I
5
10
15
20
Local Time (hours)
Fig. 5.3. Diurnal air temperature estimation. Air temperature is derived at the A VHRR afternoon
overpass time. This mid-afternoon temperature is then used to scale an average diurnal curve for
the location in question using climatology. Although useful for days that are typical climatologically, the method will result in errors when actual conditions depart from these, e.g. when cloud
conditions change within a day. This plot is for I June 87 for a station in Missouri, within the
Red-Arkansas basin
between clear areas. However, this will then cause the retrieved temperatures to be
biased towards clear conditions.
5.4.5 Surface Air Humidity
The differential absorption of water vapor by channels 4 and 5 of A VHRR used to
derive surface temperature may also be exploited to fmd total column precipitable
water. Others have noted the relationship between precipitable water and the channe14-5 temperature difference (e.g. Eck and Holben 1994), however, this relationship is incomplete and rarely exploited. A confounding factor is that the temperature difference is also a function of surface temperature, and not just precipitable
water (Prabakara et al. 1979). As shown by Dubayah (see Fig. 5 in Prince and
Goward, 1995), the slope of the line relating the temperature difference as a function of surface temperature may be used to predict precipitable water. Using
simulations with an atmospheric radiative transfer code (LOWTRAN7, Kneizys et
al. 1988) the following relationship was created relating precipitable water to the
temperature difference (Prince and Goward, 1995):