560
A.D. Del Genio
Surprisingly, little effort has been made to utilize visible channel information to discriminate
precipitating from non- precipitating cold clouds. Fu et al. (1990) have shown that a combined visible-infrared threshold method, applied either to ISCCP B3 radiances or C1 optical
thicknesses and cloud top temperatures, improves upon standard outgoing longwave radiation
identifications of locations of deep convection by isolating the convecting region from other,
less reflective high clouds. The visible channel appears to be particularly useful in identifying
convection in different regions with different temperatures (e.g., east vs. west tropical Pacific):
IR-only techniques validated in warm, heavily precipitating regions sometimes miss convection
in regions with cooler temperatures and hence shallower convective cloud tops. Indications that
their deep convective cloud (DCC) index might bear some relationship to convective precipitation come from the demonstration that the diurnal cycle of DCC over the tropical Pacific
agrees in phase with that from rain gage data, and from point correlations of enhanced DCC
cover with radiosonde-derived positive values of convective available potential energy (Fu et
al., 1994). But no quantitative relationship of DCC to rainfall has been demonstrated. TRMM
data will provide a conclusive test of whether visible reflectances can enhance the quantitative
determination of rain.
20.4.4 Combined algorithms
The real advantage of TRMM lies in its potential to use information from several types of
instruments to derive the best absolute monthly rainfall product. At this time, it has not yet
been determined what form such an algorithm will take in practice once TRMM starts returning
data. Since each instrument, different channels in a given instrument, and different retrieval
philosophies all have strengths and weaknesses that are understood to some extent, an initial
TRMM "combined" algorithm might simply be a "decision tree", in which different individual
algorithms are applied in specific situations for which they are judged most reliable. Of more
interest, though, is the simultaneous application of data from two or more instruments to exploit
the advantages of each (rain sensitivity, vertical profile information vs. spatial coverage).
One class of algorithms combines single-frequency radar data with passive microwave data to
derive rain rate. For example, the vertical profile of hydrometeors can be estimated from the
radar and used as an input to a passive microwave cloud radiative transfer model to refine
the passive microwave rain estimate (Fujita et al., 1985). Alternatively, the path-integrated
extinction derived from passive microwave measurements can be used as a constraint on the
radar equation to derive rainfall ate (Weinman et al., 1990). Another class of approaches is
using passive microwave data to refine infrared estimates of rainfall. Adler et al. (1993), for
example, compute separate rainfall totals from passive microwave and infrared algorithms, the
latter sampled to coincide with the times and locations of the microwave data. The ratio of the
two estimates is then used to calibrate a fully-sampled infrared estimate of the rainfall. Finally,
Atlas and Bell (1991) relate radar-derived area-time integrals to analogous infrared integrals
based on the GPI to improve the accuracy of large-scale rain estimates.
20.5 Complementary TRMM Science
20.5.1 Lightning Imaging Sensor (LIS)
Lightning occurrence is not by itself an indicator of rainfall, but recent evidence suggests that its
presence or absence is indicative of the dynamical setting in which convective rain is generated.
Optical detectors on the DMSP satellite reveal an instantaneous global distribution of lightning
A.D. Del Genio
Surprisingly, little effort has been made to utilize visible channel information to discriminate
precipitating from non- precipitating cold clouds. Fu et al. (1990) have shown that a combined visible-infrared threshold method, applied either to ISCCP B3 radiances or C1 optical
thicknesses and cloud top temperatures, improves upon standard outgoing longwave radiation
identifications of locations of deep convection by isolating the convecting region from other,
less reflective high clouds. The visible channel appears to be particularly useful in identifying
convection in different regions with different temperatures (e.g., east vs. west tropical Pacific):
IR-only techniques validated in warm, heavily precipitating regions sometimes miss convection
in regions with cooler temperatures and hence shallower convective cloud tops. Indications that
their deep convective cloud (DCC) index might bear some relationship to convective precipitation come from the demonstration that the diurnal cycle of DCC over the tropical Pacific
agrees in phase with that from rain gage data, and from point correlations of enhanced DCC
cover with radiosonde-derived positive values of convective available potential energy (Fu et
al., 1994). But no quantitative relationship of DCC to rainfall has been demonstrated. TRMM
data will provide a conclusive test of whether visible reflectances can enhance the quantitative
determination of rain.
20.4.4 Combined algorithms
The real advantage of TRMM lies in its potential to use information from several types of
instruments to derive the best absolute monthly rainfall product. At this time, it has not yet
been determined what form such an algorithm will take in practice once TRMM starts returning
data. Since each instrument, different channels in a given instrument, and different retrieval
philosophies all have strengths and weaknesses that are understood to some extent, an initial
TRMM "combined" algorithm might simply be a "decision tree", in which different individual
algorithms are applied in specific situations for which they are judged most reliable. Of more
interest, though, is the simultaneous application of data from two or more instruments to exploit
the advantages of each (rain sensitivity, vertical profile information vs. spatial coverage).
One class of algorithms combines single-frequency radar data with passive microwave data to
derive rain rate. For example, the vertical profile of hydrometeors can be estimated from the
radar and used as an input to a passive microwave cloud radiative transfer model to refine
the passive microwave rain estimate (Fujita et al., 1985). Alternatively, the path-integrated
extinction derived from passive microwave measurements can be used as a constraint on the
radar equation to derive rainfall ate (Weinman et al., 1990). Another class of approaches is
using passive microwave data to refine infrared estimates of rainfall. Adler et al. (1993), for
example, compute separate rainfall totals from passive microwave and infrared algorithms, the
latter sampled to coincide with the times and locations of the microwave data. The ratio of the
two estimates is then used to calibrate a fully-sampled infrared estimate of the rainfall. Finally,
Atlas and Bell (1991) relate radar-derived area-time integrals to analogous infrared integrals
based on the GPI to improve the accuracy of large-scale rain estimates.
20.5 Complementary TRMM Science
20.5.1 Lightning Imaging Sensor (LIS)
Lightning occurrence is not by itself an indicator of rainfall, but recent evidence suggests that its
presence or absence is indicative of the dynamical setting in which convective rain is generated.
Optical detectors on the DMSP satellite reveal an instantaneous global distribution of lightning