554
A.D. Del Genio
o
10
20
30
RAIN RATE (mmlhr)
Figure 20.3: An example of a log-normal probability density function, which is characteristic
of rainfall and other cloud statistics in a variety of tropical convective regions.
A geostationary satellite, which flies at high altitude, views essentially all of one face of the
Earth almost continuously, and so can sample a given location adequately. But TRMM must
fly at very low altitude to have acceptable footprint sizes for the PR, whose antenna size is
limited by mass and power constraints to be significantly smaller than a typical surface-based
rain radar. Thus, TRMM views only a small fraction of the Earth at a given instant and
returns to the same location only every several days. Within a particular (500 km)2 region,
some satellite passes may cover all of the area, while others will only partly sample the raining
region. The sampling error that results is a function of the spacecraft altitude and orbit
inclination. Thus, the realized sampling error involves several tradeoffs: (1) Increasing orbit
altitude provides better spatial coverage, but at the price of reduced resolution and accuracy for
individual rain retrievals; (2) Decreasing orbit inclination provides better coverage of the diurnal
cycle and thus reduces sampling error, but also reduces the fraction of the globe over which rain
information will be acquired. The diurnal cycle issue is an especially important one for TRMM:
Precipitation has an especially strong diurnal variation over land (cf. McGarry and Reed, 1978),
with maxima occurring in afternoon over many locations but in the evening at others. Over
oceans, the diurnal cycle is much weaker but noticeable, with early morning maxima typical
over many locations (cf. Gray and Jacobson, 1977). The geographic variability of diurnal cycle
strength and phase mandates that TRMM fly at a fairly low inclination to adequately sample
these variations, not only so that its monthly means are sufficiently accurate, but because
the diurnal cycle itself contains information about the physical processes that determine rain
occurrence. A variety of sampling studies have been conducted in support of TRMM (cf. Shin
and North, 1988); these suggest that a 10% monthly mean accuracy is achieveable at 500 km
scales for a 350 km altitude orbit and 35° inclination. It is largely these sampling considerations
that have determined the orbit configuration for TRMM.
A.D. Del Genio
o
10
20
30
RAIN RATE (mmlhr)
Figure 20.3: An example of a log-normal probability density function, which is characteristic
of rainfall and other cloud statistics in a variety of tropical convective regions.
A geostationary satellite, which flies at high altitude, views essentially all of one face of the
Earth almost continuously, and so can sample a given location adequately. But TRMM must
fly at very low altitude to have acceptable footprint sizes for the PR, whose antenna size is
limited by mass and power constraints to be significantly smaller than a typical surface-based
rain radar. Thus, TRMM views only a small fraction of the Earth at a given instant and
returns to the same location only every several days. Within a particular (500 km)2 region,
some satellite passes may cover all of the area, while others will only partly sample the raining
region. The sampling error that results is a function of the spacecraft altitude and orbit
inclination. Thus, the realized sampling error involves several tradeoffs: (1) Increasing orbit
altitude provides better spatial coverage, but at the price of reduced resolution and accuracy for
individual rain retrievals; (2) Decreasing orbit inclination provides better coverage of the diurnal
cycle and thus reduces sampling error, but also reduces the fraction of the globe over which rain
information will be acquired. The diurnal cycle issue is an especially important one for TRMM:
Precipitation has an especially strong diurnal variation over land (cf. McGarry and Reed, 1978),
with maxima occurring in afternoon over many locations but in the evening at others. Over
oceans, the diurnal cycle is much weaker but noticeable, with early morning maxima typical
over many locations (cf. Gray and Jacobson, 1977). The geographic variability of diurnal cycle
strength and phase mandates that TRMM fly at a fairly low inclination to adequately sample
these variations, not only so that its monthly means are sufficiently accurate, but because
the diurnal cycle itself contains information about the physical processes that determine rain
occurrence. A variety of sampling studies have been conducted in support of TRMM (cf. Shin
and North, 1988); these suggest that a 10% monthly mean accuracy is achieveable at 500 km
scales for a 350 km altitude orbit and 35° inclination. It is largely these sampling considerations
that have determined the orbit configuration for TRMM.