268
C. Simmer
temperatures (determined from the IR-signal) below a certain threshold. The results for the
region under investigation can be summarized as follows (Arkin and Xie, 1994):
• The threshold temperature of -28 0 ( used in (11.3)) is applicable.
• Maximum correlations found were between 0.7 and 0.8, somewhat lower than obtained
from the GARP results.
• The optimal value of the constant of proportionality a and of the threshold temperature
Tc depend on area size and the time interval used.
• a and Tc depend on the region.
• For the same region a can vary from month to month by 100%.
• No improvement was obtained using additional predictors derived from IR, like spatial
inhomogenei ty.
PIP-l (WetNet)
The NASA-led WetNet Project is primarily focussed on algorithms based on data from the
U.S. military DMSP (Defence Meteorological Satellite Program) satellite SSM/I (Special Sensor
Microwave Imager) microwave radiometer. A central part of the Wet Net work program is a
series of Precipitation Intercomparison Projects (PIPs). The first of these projects, PIP-I, was
intended primarily to evaluate existing passive microwave algorithms both in relation to each
other and also against conventional (rain gauge) data sets. Within the PIP groups involved
in the development of rain retrieval algorithms agree on a common data base, which is first
analyzed independently by the groups. In PIP-I, intercomparisons of global rainfall estimates
for August, September, October and November of the first year of SSM/I data, 1987, have
been undertaken for 15 algorithms based on passive microwave DMSP-SSM/I image data,
one algorithm based on passive microwave NOAA (National Oceanographic and Atmospheric
Administration) MSU (Microwave Sounding Unit) sounder data, one infrared image data-based
algorithm, one combined passive microwave/infrared image data based algorithm, one numerical
weather prediction model and rainfall observations from rain-gauges.
The contributors to PIP-I agreed to use the following procedure to evaluate the results of the
different algorithms and to attempt a ranking, which could not be final having all the problems
mentioned above in mind.
1. Subjective ("eye-ball") inspections of the graphical displays was used to judge, whether
the obtained monthly distributions of rain fall were reasonable and in broad agreement
with what is known both on the global and regional scale.
2. Quantitative comparisons were made with the results of the Global Precipitation Climatology Project (GPCP) prepared by the Global Precipitation Climatology Center in
Offenbach, Germany (Rudolf et al., 1992), which is generally recognized as the best available rain-gauge based data set so far.
3. The authors of the rain-retrieval algorithms were asked to make an assessment of their
own algorithms based on the knowledge of their strengths and shortcomings in view of
the results obtained in 1. and 2.
C. Simmer
temperatures (determined from the IR-signal) below a certain threshold. The results for the
region under investigation can be summarized as follows (Arkin and Xie, 1994):
• The threshold temperature of -28 0 ( used in (11.3)) is applicable.
• Maximum correlations found were between 0.7 and 0.8, somewhat lower than obtained
from the GARP results.
• The optimal value of the constant of proportionality a and of the threshold temperature
Tc depend on area size and the time interval used.
• a and Tc depend on the region.
• For the same region a can vary from month to month by 100%.
• No improvement was obtained using additional predictors derived from IR, like spatial
inhomogenei ty.
PIP-l (WetNet)
The NASA-led WetNet Project is primarily focussed on algorithms based on data from the
U.S. military DMSP (Defence Meteorological Satellite Program) satellite SSM/I (Special Sensor
Microwave Imager) microwave radiometer. A central part of the Wet Net work program is a
series of Precipitation Intercomparison Projects (PIPs). The first of these projects, PIP-I, was
intended primarily to evaluate existing passive microwave algorithms both in relation to each
other and also against conventional (rain gauge) data sets. Within the PIP groups involved
in the development of rain retrieval algorithms agree on a common data base, which is first
analyzed independently by the groups. In PIP-I, intercomparisons of global rainfall estimates
for August, September, October and November of the first year of SSM/I data, 1987, have
been undertaken for 15 algorithms based on passive microwave DMSP-SSM/I image data,
one algorithm based on passive microwave NOAA (National Oceanographic and Atmospheric
Administration) MSU (Microwave Sounding Unit) sounder data, one infrared image data-based
algorithm, one combined passive microwave/infrared image data based algorithm, one numerical
weather prediction model and rainfall observations from rain-gauges.
The contributors to PIP-I agreed to use the following procedure to evaluate the results of the
different algorithms and to attempt a ranking, which could not be final having all the problems
mentioned above in mind.
1. Subjective ("eye-ball") inspections of the graphical displays was used to judge, whether
the obtained monthly distributions of rain fall were reasonable and in broad agreement
with what is known both on the global and regional scale.
2. Quantitative comparisons were made with the results of the Global Precipitation Climatology Project (GPCP) prepared by the Global Precipitation Climatology Center in
Offenbach, Germany (Rudolf et al., 1992), which is generally recognized as the best available rain-gauge based data set so far.
3. The authors of the rain-retrieval algorithms were asked to make an assessment of their
own algorithms based on the knowledge of their strengths and shortcomings in view of
the results obtained in 1. and 2.