296
P. Joe
12.6 Precipitation Measurements
12.6.1 Precipitation estimation
Radar has a long history of use in estimating the intensity and thereby the amount and distribution of precipitation falling within the radar viewing area with a good resolution in time
and space. Most studies have been associated with rainfall but snow measurements can also be
made with appropriate allowances for target composition. Readers should consult reviews by
Joss and Waldvogel (1990) and Smith (1990) for a comprehensive discussion on the state of the
art, the techniques, the problems and pitfalls, and the effectiveness and accuracy. Ground level
precipitation estimates from typical radar systems are made for areas of 2 km 2 , successively for
5-10 minute periods using low elevation PPI scans with beamwidths of 10.
When the radar beam or its sidelobes encounter ground targets, persistent echoes occur which
can add up in time to appear as large rain amounts if no precautions are taken in the data
analysis. A clutter map in the computer memory may be used to eliminate clutter. But as
clutter, and especially that caused by anomalous propagation, is variable in time and space,
clutter maps are difficult to use in a general context. Doppler techniques hold promise for
reducing these difficulties but some problems are likely to remain (Passarelli et al., 1982).
These and other questions, such as the choice of the wavelength, errors caused by attenuation,
considerations when choosing a radar site for hydrological applications, hardware calibration
of radar systems, sampling and averaging and the meteorological adjustment of radar data, are
discussed in Joss and Waldvogel (1990) and Smith (1990). The present brieftreatment considers
only rainfall measurements; little operational experience is available about radar measurements
of snow and even less about measurements of hail.
12.6.2 Precipitation characteristics which affect radar measurements
Drop size distributions
Returned power or radar reflectance factor is the basic precipitation parameter because of its
commonality. Basic to the interpretation of Z is its relationship to the drop size distribution
(DSD) since Z is equal to the sum of the diameter of drops to the sixth power per unit volume.
The reader is referred to textbooks on cloud physics (Pruppacher and Klett, 1978; Rogers and
Yau, 1989; Mason, 1971) which discuss the formation of clouds and precipitation.
Marshall-Palmer (1948) (hereafter referred to as M-P) observed that rain DSD's are exponentially distributed (see Fig. 12.8) and can be empirically expressed as:
N(D) = Noexp(-AD)
A = 4.1R- o . 21 mm- 1
No = 8 X 10 3 m- 3 mm- 1
(12.11 )
(12.12)
(12.13)
Higher order gamma formulations have been proposed (Ulbrich, 1983; Cataneo and Stout, 1968;
Austin and Geotis, 1979) but the formulation above remains the most popular.
For hail, exponential distributions have also been observed (Douglas, 1964; Hitschfeld and
Stauder, 1965; Federer and Waldvogel, 1975; Spahn and Smith, 1976; Cheng and English,
1983). Various results have been expressed as
No = 115A 3 . 63
(12.14)
P. Joe
12.6 Precipitation Measurements
12.6.1 Precipitation estimation
Radar has a long history of use in estimating the intensity and thereby the amount and distribution of precipitation falling within the radar viewing area with a good resolution in time
and space. Most studies have been associated with rainfall but snow measurements can also be
made with appropriate allowances for target composition. Readers should consult reviews by
Joss and Waldvogel (1990) and Smith (1990) for a comprehensive discussion on the state of the
art, the techniques, the problems and pitfalls, and the effectiveness and accuracy. Ground level
precipitation estimates from typical radar systems are made for areas of 2 km 2 , successively for
5-10 minute periods using low elevation PPI scans with beamwidths of 10.
When the radar beam or its sidelobes encounter ground targets, persistent echoes occur which
can add up in time to appear as large rain amounts if no precautions are taken in the data
analysis. A clutter map in the computer memory may be used to eliminate clutter. But as
clutter, and especially that caused by anomalous propagation, is variable in time and space,
clutter maps are difficult to use in a general context. Doppler techniques hold promise for
reducing these difficulties but some problems are likely to remain (Passarelli et al., 1982).
These and other questions, such as the choice of the wavelength, errors caused by attenuation,
considerations when choosing a radar site for hydrological applications, hardware calibration
of radar systems, sampling and averaging and the meteorological adjustment of radar data, are
discussed in Joss and Waldvogel (1990) and Smith (1990). The present brieftreatment considers
only rainfall measurements; little operational experience is available about radar measurements
of snow and even less about measurements of hail.
12.6.2 Precipitation characteristics which affect radar measurements
Drop size distributions
Returned power or radar reflectance factor is the basic precipitation parameter because of its
commonality. Basic to the interpretation of Z is its relationship to the drop size distribution
(DSD) since Z is equal to the sum of the diameter of drops to the sixth power per unit volume.
The reader is referred to textbooks on cloud physics (Pruppacher and Klett, 1978; Rogers and
Yau, 1989; Mason, 1971) which discuss the formation of clouds and precipitation.
Marshall-Palmer (1948) (hereafter referred to as M-P) observed that rain DSD's are exponentially distributed (see Fig. 12.8) and can be empirically expressed as:
N(D) = Noexp(-AD)
A = 4.1R- o . 21 mm- 1
No = 8 X 10 3 m- 3 mm- 1
(12.11 )
(12.12)
(12.13)
Higher order gamma formulations have been proposed (Ulbrich, 1983; Cataneo and Stout, 1968;
Austin and Geotis, 1979) but the formulation above remains the most popular.
For hail, exponential distributions have also been observed (Douglas, 1964; Hitschfeld and
Stauder, 1965; Federer and Waldvogel, 1975; Spahn and Smith, 1976; Cheng and English,
1983). Various results have been expressed as
No = 115A 3 . 63
(12.14)