Retrieval of Precipitation from Satellites
261
11.6 Remote sensing methods
11.6.1 Visible and infrared
Several methods have been developed to use visible and infrared radiation measurements to
retrieve rainrate estimates:
• GPI-Type Methods
• Outgoing Longwave Radiation (OLR)
• Cloud droplet size in the upper cloud
• Lightning
Since lightning is produced by charge separation processes which involve precipitation-size
hydrometeors lightning has some relation to the convective activity within a cloud. Thus a
connection between the frequency of lightning detectable from satellites and rainrate at the
surface can be expected. But not every raining cloud produces lightning, which limits the
applicability of these methods to only one class of precipitation.
Raining clouds must have a fraction of large droplets in the upper part of the cloud. Since
Mie-scattering is important in the visible and infrared spectral range raining clouds can in
principal be distinguished from non-raining clouds by these effects. This methods are still in
the development stage.
The outgoing longwave radiation is dominated by the cloud-top temperature. The colder the
cloud the lower the OLR. Methods based on OLR measurements are in principle very similar
to the GPI-type methods, which are the most frequently used techniques. They are based on
the assumption that the rain produced in convective clouds is proportional to the area Ao with
cloud-top temperatures below a predefined threshold To:
RR ex Ao with Ao area with T < To
(11.3)
By using the infrared observations from geostationary satellites and radar-observed rainfall data
for the GARP (Global Atmospheric Research Program) Atlantic Tropical Experiment (GATE),
Arkin (1979) found that the correlation between 6-hour rainfall and the fractional coverage of
cold clouds is above 0.8 when the fractional coverage is defined by To between 230 and 240 K.
These findings and further analysis (Arkin and Meisner, 1987) gave rise to the foundation of
the GOES (Geostationary Operational Environmental Satellite) Precipitation Index (GPI)
RR mm/h = 3.0 mm/h x
fractional area of 2.5 x 2.5 with T < 235K (11.4)
The method works in the tropics over land and ocean, but becomes increasingly inaccurate at
higher latitudes. The main advantage of the method is the high temporal resolution because
visible/infrared sensors can be used on satellites in geostationary orbits. Comparison with radar
and raingage estimates have shown that (11.3) is indeed a valid assumption, but To can differ
from case to case. The constant of proportionality has a regional dependence and may vary by
more than 100% from month to month (Arkin and Xie, 1994). Thus large efforts are made to
determine To and the constant of proportionality a from additional information. Candidates
are the time evolution, the spatial inhomogeneity of the cloud, and information from visible
sensors or microwave measurements, if available.
261
11.6 Remote sensing methods
11.6.1 Visible and infrared
Several methods have been developed to use visible and infrared radiation measurements to
retrieve rainrate estimates:
• GPI-Type Methods
• Outgoing Longwave Radiation (OLR)
• Cloud droplet size in the upper cloud
• Lightning
Since lightning is produced by charge separation processes which involve precipitation-size
hydrometeors lightning has some relation to the convective activity within a cloud. Thus a
connection between the frequency of lightning detectable from satellites and rainrate at the
surface can be expected. But not every raining cloud produces lightning, which limits the
applicability of these methods to only one class of precipitation.
Raining clouds must have a fraction of large droplets in the upper part of the cloud. Since
Mie-scattering is important in the visible and infrared spectral range raining clouds can in
principal be distinguished from non-raining clouds by these effects. This methods are still in
the development stage.
The outgoing longwave radiation is dominated by the cloud-top temperature. The colder the
cloud the lower the OLR. Methods based on OLR measurements are in principle very similar
to the GPI-type methods, which are the most frequently used techniques. They are based on
the assumption that the rain produced in convective clouds is proportional to the area Ao with
cloud-top temperatures below a predefined threshold To:
RR ex Ao with Ao area with T < To
(11.3)
By using the infrared observations from geostationary satellites and radar-observed rainfall data
for the GARP (Global Atmospheric Research Program) Atlantic Tropical Experiment (GATE),
Arkin (1979) found that the correlation between 6-hour rainfall and the fractional coverage of
cold clouds is above 0.8 when the fractional coverage is defined by To between 230 and 240 K.
These findings and further analysis (Arkin and Meisner, 1987) gave rise to the foundation of
the GOES (Geostationary Operational Environmental Satellite) Precipitation Index (GPI)
RR mm/h = 3.0 mm/h x
fractional area of 2.5 x 2.5 with T < 235K (11.4)
The method works in the tropics over land and ocean, but becomes increasingly inaccurate at
higher latitudes. The main advantage of the method is the high temporal resolution because
visible/infrared sensors can be used on satellites in geostationary orbits. Comparison with radar
and raingage estimates have shown that (11.3) is indeed a valid assumption, but To can differ
from case to case. The constant of proportionality has a regional dependence and may vary by
more than 100% from month to month (Arkin and Xie, 1994). Thus large efforts are made to
determine To and the constant of proportionality a from additional information. Candidates
are the time evolution, the spatial inhomogeneity of the cloud, and information from visible
sensors or microwave measurements, if available.