406
I. Papadakis and G.A. Schultz
conversion of the digital counts of the thennal infrared and water vapour images into temperature and mean relative humidity in the upper troposphere respectively;
atmospheric corrections are not necessary since it has been shown (Rott et. al.
1986) by means of a atmospheric simulation model "LOWTRAN - 5" (Kneizys et. al. 1980) that atmospheric correction is important only for low levels in
the atmosphere. Levels below 700 hPa are of main influence for atmospheric
transmittance. However, since for rainfall estimation purposes only higher
cloud levels are of interest, the radiance of the thermal infrared channel can be
converted into temperature without considering atmospheric effects. Uncertainties about cloud emissivity cause larger errors than the effects of atmospheric transmittance.
For image processing techniques applicable here see Chap. 3.
18.2.2 MODUL II: Assessment of the monthly area precipitation on the basis
of multi-temporal satellite imagery
A number of scientists have already used data from geostationary satellites to
compute precipitation from convective Clouds (e.g. Arkin 1979; Griffith et aI.,
1978; Hardy et ai. 1989; Sorooshian 1997 and others). Some of their schemes are,
however, based on the individual tracking of cloud entities throughout their lifetimes. These methods require substantial computing capacity and are applicable
only over limited areas in space and time periods. For the reconstruction of long
hydrometeorological time series the method for deriving rainfall estimates has to
be relatively simple. Arkin (1979) has shown that the fractional cloud cover, which
is colder than a certain cloud top threshold temperature, is proportional to the
accumulated rainfall amount below (Fig. 18.3).
Hardy et ai. (1989) used the duration of cold cloud (CCD) as an indication of
rain and the rainfall total within a time period.
There are, however, some limitations in the Arkin and Hardy approach. The
threshold cloud top temperature is the only one parameter taken into account,
although two cloud cells with the same cloud top temperature and with comparable
sizes produce different rainfall amounts depending on the state of their environmental humidity (KrUger and Schultz, 1982). Clouds in a moist environment produce considerably more rain than those in a relatively dry environment. The influence of the humidity on the rainfall intensity is also emphasized by Adler and
Mack (1984). The mean relative humidity in the upper troposphere obtained by the
water vapour channel of the satellite system Meteosat can be used as an indicator
for the humidity state of the environment of clouds or cloud systems.
The principles of the techniques mentioned above for estimating rainfall from
temperature of cloud tops are to associate rainfall with cold (high) clouds assuming that these are the tops of active convective cells or convective systems.
Figure 18.4 shows examples of the correlation between rainfall rate and cloud
top height (i.e. cloud top temperature) for various regions in North America.
I. Papadakis and G.A. Schultz
conversion of the digital counts of the thennal infrared and water vapour images into temperature and mean relative humidity in the upper troposphere respectively;
atmospheric corrections are not necessary since it has been shown (Rott et. al.
1986) by means of a atmospheric simulation model "LOWTRAN - 5" (Kneizys et. al. 1980) that atmospheric correction is important only for low levels in
the atmosphere. Levels below 700 hPa are of main influence for atmospheric
transmittance. However, since for rainfall estimation purposes only higher
cloud levels are of interest, the radiance of the thermal infrared channel can be
converted into temperature without considering atmospheric effects. Uncertainties about cloud emissivity cause larger errors than the effects of atmospheric transmittance.
For image processing techniques applicable here see Chap. 3.
18.2.2 MODUL II: Assessment of the monthly area precipitation on the basis
of multi-temporal satellite imagery
A number of scientists have already used data from geostationary satellites to
compute precipitation from convective Clouds (e.g. Arkin 1979; Griffith et aI.,
1978; Hardy et ai. 1989; Sorooshian 1997 and others). Some of their schemes are,
however, based on the individual tracking of cloud entities throughout their lifetimes. These methods require substantial computing capacity and are applicable
only over limited areas in space and time periods. For the reconstruction of long
hydrometeorological time series the method for deriving rainfall estimates has to
be relatively simple. Arkin (1979) has shown that the fractional cloud cover, which
is colder than a certain cloud top threshold temperature, is proportional to the
accumulated rainfall amount below (Fig. 18.3).
Hardy et ai. (1989) used the duration of cold cloud (CCD) as an indication of
rain and the rainfall total within a time period.
There are, however, some limitations in the Arkin and Hardy approach. The
threshold cloud top temperature is the only one parameter taken into account,
although two cloud cells with the same cloud top temperature and with comparable
sizes produce different rainfall amounts depending on the state of their environmental humidity (KrUger and Schultz, 1982). Clouds in a moist environment produce considerably more rain than those in a relatively dry environment. The influence of the humidity on the rainfall intensity is also emphasized by Adler and
Mack (1984). The mean relative humidity in the upper troposphere obtained by the
water vapour channel of the satellite system Meteosat can be used as an indicator
for the humidity state of the environment of clouds or cloud systems.
The principles of the techniques mentioned above for estimating rainfall from
temperature of cloud tops are to associate rainfall with cold (high) clouds assuming that these are the tops of active convective cells or convective systems.
Figure 18.4 shows examples of the correlation between rainfall rate and cloud
top height (i.e. cloud top temperature) for various regions in North America.