310
P. Joe
velocity and then retrieve the air velocity by the difference from the mean Doppler velocity
(Rogers, 1964); using a minimum detectable size assumption which is assumed to move with
the air (Battan, 1981); or exploiting the clear air velocity component to the spectrum for large
wavelength systems (Wakasugi et al., 1987). However, errors of only ±0.25 m S-1 in vertical
velocity can cause errors of 100% in drop number concentrations (Atlas et. al., 1973).
12.6.13 Vertical profile effects
Highly variable vertical reflectance profiles are observed, both within a given storm and from
storm to storm because of growth or evaporation of precipitation, air motion and change of
phase (ice and water in the melting layer, or bright band). This will lead to a non-uniform
distribution of reflectance with height. As the beamwidth increases with range, the profile will
be averaged resulting in a smoother vertical distribution of reflectance (see Fig. 12.22).
In convective rainfall, the advantages of radar are most pronounced and experience shows less
difficulty with the vertical profile problem. But in stratiform rain or snow, the vertical profile
becomes more important (see Fig. 12.23). As the beam becomes wider with increasing distance,
the vertical distance between the sample volume of the radar and the ground usually increases
too. Therefore, the differences between estimates of rainfall by radar and the rain measured
at the ground also increases. Reflectance generally decreases with height. Therefore, radar
estimations underestimate the rain measured at the surface.
No one method of compensating for the effects of the vertical reflectance profile in real time
is widely accepted. A climatological mean range-dependent factor can be applied to obtain
a first-order correction. Different factors may be appropriate for different storm types (for
example, convective versus stratiform).
The vertical profiles in storms vary with location and time, and the lowest level visible to the
radar usually varies because of irregularities in the radar horizon. Consequently a point-bypoint correction process using a representative vertical profile for each zone of concern may be
needed to obtain the best results.
Representative profiles can be obtained from the radar volume scan data themselves, climatological summaries, or storm models (Joss and Lee, 1993). At long ranges, for low level storms,
and especially when low antenna elevations are blocked by obstacles such as mountains, the
underestimate may be severe (the spreading of the beam also becomes a problem in convective
rain).
The basic procedure of deducing rainfall rates from measured radar reflectivities for hydrological
applications requires the following steps:
• Making sure that the hardware is stable by calibration and maintenance.
• Correcting for errors using the vertical reflectance profile.
• Taking into account all the information about the Ze - R relationship and deducing the
rainfall.
• Adjustment with raingauges.
The first three parts are based on known physical laws and only the last one uses a statistical
approach to compensate for residual errors. This allows the statistical methods to work most
efficiently. After making the profile corrections, one should use a reflectance - rainrate relationship appropriate to the situation, geography as well as season, to deduce the value of R (see
previous discussion).
P. Joe
velocity and then retrieve the air velocity by the difference from the mean Doppler velocity
(Rogers, 1964); using a minimum detectable size assumption which is assumed to move with
the air (Battan, 1981); or exploiting the clear air velocity component to the spectrum for large
wavelength systems (Wakasugi et al., 1987). However, errors of only ±0.25 m S-1 in vertical
velocity can cause errors of 100% in drop number concentrations (Atlas et. al., 1973).
12.6.13 Vertical profile effects
Highly variable vertical reflectance profiles are observed, both within a given storm and from
storm to storm because of growth or evaporation of precipitation, air motion and change of
phase (ice and water in the melting layer, or bright band). This will lead to a non-uniform
distribution of reflectance with height. As the beamwidth increases with range, the profile will
be averaged resulting in a smoother vertical distribution of reflectance (see Fig. 12.22).
In convective rainfall, the advantages of radar are most pronounced and experience shows less
difficulty with the vertical profile problem. But in stratiform rain or snow, the vertical profile
becomes more important (see Fig. 12.23). As the beam becomes wider with increasing distance,
the vertical distance between the sample volume of the radar and the ground usually increases
too. Therefore, the differences between estimates of rainfall by radar and the rain measured
at the ground also increases. Reflectance generally decreases with height. Therefore, radar
estimations underestimate the rain measured at the surface.
No one method of compensating for the effects of the vertical reflectance profile in real time
is widely accepted. A climatological mean range-dependent factor can be applied to obtain
a first-order correction. Different factors may be appropriate for different storm types (for
example, convective versus stratiform).
The vertical profiles in storms vary with location and time, and the lowest level visible to the
radar usually varies because of irregularities in the radar horizon. Consequently a point-bypoint correction process using a representative vertical profile for each zone of concern may be
needed to obtain the best results.
Representative profiles can be obtained from the radar volume scan data themselves, climatological summaries, or storm models (Joss and Lee, 1993). At long ranges, for low level storms,
and especially when low antenna elevations are blocked by obstacles such as mountains, the
underestimate may be severe (the spreading of the beam also becomes a problem in convective
rain).
The basic procedure of deducing rainfall rates from measured radar reflectivities for hydrological
applications requires the following steps:
• Making sure that the hardware is stable by calibration and maintenance.
• Correcting for errors using the vertical reflectance profile.
• Taking into account all the information about the Ze - R relationship and deducing the
rainfall.
• Adjustment with raingauges.
The first three parts are based on known physical laws and only the last one uses a statistical
approach to compensate for residual errors. This allows the statistical methods to work most
efficiently. After making the profile corrections, one should use a reflectance - rainrate relationship appropriate to the situation, geography as well as season, to deduce the value of R (see
previous discussion).