389
Instantaneous Precipitation and Latent Heating Estimation over Land
combined retrievals, had been carried out over oceans (Grecu et al. 2004; Masunaga
and Kummerow 2005), but to a significantly smaller extent over land.
The challenges in deriving satellite combined radar–radiometer retrievals over
land stem from the fact that land is usually characterized by large emissivity, which
makes it hard to distinguish the precipitation signal from variations in surface emissivity. This is the reason why only high-frequency radiometer observations (37 GHz
and higher) are directly used in radiometer retrievals over land. Lower frequency
observations are used to quantify the impact of other factors (e.g., surface emissivity
and liquid water path) on the brightness temperature observations that are directly
used in retrieving precipitation. High-frequency radiometer observations over land
(37 GHz to a certain extent but especially 85 GHz and higher) are highly sensitive to
the electromagnetic properties of solid precipitation particles (e.g., ice crystal, snow
flakes, graupel, and hail particles) above the freezing level and significantly less sensitive to properties of liquid-phase precipitation. Nevertheless, because the amount
of solid precipitation at a given location is correlated to the amount of precipitation
below the freezing level, it is possible to infer that the surface precipitation from
observations is mostly sensitive to precipitation particles above the freezing level.
Brightness temperatures characterized by a surface temperature below the freezing
temperature are not necessarily easier to interpret, because in such situations, snow
may accumulate on the ground, which can significantly affect the surface emissivity. Despite being more ambiguous than over ocean retrievals, combined TRMM
radar–radiometer precipitation over land are extremely useful, because they can be
used to construct databases of precipitation and associated brightness temperatures
for sensors with different geometries, frequencies, and resolutions than those of
TMI’s. Thus, the utility of coincident PR TMI observations demonstrated by Grecu
and Anagnostou (2001) and McCollum and Ferraro (2003) can be extended to other
sensors. An example of such a sensor is AMSU-B (Goodrum et al. 1999) onboard
the U.S. National Oceanic and Atmospheric Administration satellites. AMSU-B is a
sensor that features five channels between 89 and 183.30 GHz and a spatial resolution near a nadir of 15°km. Given the significant different viewing geometry as well
as the existence of sounding channels, the observed TMI brightness temperatures
can be used in the development of a precipitation AMSU-B brightness temperature
database only through a combined PR TMI retrieval.
In this study, a methodology for estimating precipitation and latent heating over
land from combined radar–radiometer retrieval is presented. A brightness temperature sensitivity analysis that justifies the combined methodology is discussed in
Section 16.2. Details concerning the mathematical formulation and its solution are
given in Section 16.3. Implications on the incoming GPM Mission and issues that
need to be addressed by future research are emphasized in Section 16.4.
16.2 BRIGHTNESS TEMPERATURE SENSITIVITY ANALYSIS
Although significantly better determined than radiometer-only observations, singlefrequency radar-only observations cannot be uniquely associated with precipitation. This is because the particle size distribution (PSD) variability within radar
observing volumes exhibits at least two degrees of freedom. That is, at least two
Instantaneous Precipitation and Latent Heating Estimation over Land
combined retrievals, had been carried out over oceans (Grecu et al. 2004; Masunaga
and Kummerow 2005), but to a significantly smaller extent over land.
The challenges in deriving satellite combined radar–radiometer retrievals over
land stem from the fact that land is usually characterized by large emissivity, which
makes it hard to distinguish the precipitation signal from variations in surface emissivity. This is the reason why only high-frequency radiometer observations (37 GHz
and higher) are directly used in radiometer retrievals over land. Lower frequency
observations are used to quantify the impact of other factors (e.g., surface emissivity
and liquid water path) on the brightness temperature observations that are directly
used in retrieving precipitation. High-frequency radiometer observations over land
(37 GHz to a certain extent but especially 85 GHz and higher) are highly sensitive to
the electromagnetic properties of solid precipitation particles (e.g., ice crystal, snow
flakes, graupel, and hail particles) above the freezing level and significantly less sensitive to properties of liquid-phase precipitation. Nevertheless, because the amount
of solid precipitation at a given location is correlated to the amount of precipitation
below the freezing level, it is possible to infer that the surface precipitation from
observations is mostly sensitive to precipitation particles above the freezing level.
Brightness temperatures characterized by a surface temperature below the freezing
temperature are not necessarily easier to interpret, because in such situations, snow
may accumulate on the ground, which can significantly affect the surface emissivity. Despite being more ambiguous than over ocean retrievals, combined TRMM
radar–radiometer precipitation over land are extremely useful, because they can be
used to construct databases of precipitation and associated brightness temperatures
for sensors with different geometries, frequencies, and resolutions than those of
TMI’s. Thus, the utility of coincident PR TMI observations demonstrated by Grecu
and Anagnostou (2001) and McCollum and Ferraro (2003) can be extended to other
sensors. An example of such a sensor is AMSU-B (Goodrum et al. 1999) onboard
the U.S. National Oceanic and Atmospheric Administration satellites. AMSU-B is a
sensor that features five channels between 89 and 183.30 GHz and a spatial resolution near a nadir of 15°km. Given the significant different viewing geometry as well
as the existence of sounding channels, the observed TMI brightness temperatures
can be used in the development of a precipitation AMSU-B brightness temperature
database only through a combined PR TMI retrieval.
In this study, a methodology for estimating precipitation and latent heating over
land from combined radar–radiometer retrieval is presented. A brightness temperature sensitivity analysis that justifies the combined methodology is discussed in
Section 16.2. Details concerning the mathematical formulation and its solution are
given in Section 16.3. Implications on the incoming GPM Mission and issues that
need to be addressed by future research are emphasized in Section 16.4.
16.2 BRIGHTNESS TEMPERATURE SENSITIVITY ANALYSIS
Although significantly better determined than radiometer-only observations, singlefrequency radar-only observations cannot be uniquely associated with precipitation. This is because the particle size distribution (PSD) variability within radar
observing volumes exhibits at least two degrees of freedom. That is, at least two