6 Precipitation
117
Recent use of radar-gauge comparisons has attempted to match the probability
density functions of both radar estimates and raingauge measurements. The assumption made is that made by Calheiros and Zawadzki (1987) namely that the unconditional probability for any given rainfall intensity is constant everywhere in the radar
field, and that this probability was sampled in a representative way by raingauges. The
need for this assumption is removed by Rosenfeld et al. (1993) who assumed that the
conditional probability density function of rainfall intensity, regardless of the integrated rain depth for any given rainfall regime, is the same everywhere. The Ze:R
relationship so derived varies greatly with the rainfall regime and with the radial
distance from the radar
This work was developed further by Rosenfeld et al. (1994) and Rosenfeld et al.
(1995a, b), and has become known as the Window Probability Matching Method
(WPMM). Rainfall types are classified using the horizontal radial reflectivity gradients, the cloud depth as scaled by the effective efficiency, the bright band fraction
within the radar field window and the height of the freezing level. The results of tests
in Australia and Israel are very impressive, and comparisons of these results with
radar data for convective cases in North West England (Fig. 6.2) suggest a more
general applicability. It should be noted that the WPMM does not require that range
effects be removed or that the bright-band be corrected in the radar imagery, but
Convective Maritime
0.8
0.6
r::
0
0.4
' 13
c:
:l
0.2
u.
r::
.Q
0
'5
.0 .;::
·0.2
iii
is ·0.4
CI)
>
.' til ·0.6
'5
E -0.6
:>
U
-,
-'.2
, ",......
~ .. !I'" V; VI
/'
V
~...-,'" V V
J
I
, /
' :/ / /
, V
I
J
/
I,
, J
i'
. :
I
, ...-: J
/ '
.
/ /
I
Gau e CDF 1."'./
,.
1 I
:~
I
.. I
I
.... ,' ,
...--...r 11/
I - - Ra~ar ~DFs ..--l
I
- - Raingauges
........ ..... , Radar 20 - 49km I'
I
- -- Radar 50 - 79km
- - Radar 80 - 109km V
- - Radar 110 · 139km
I
-, .4
-20 -15 ·10 -5
0
5
10 15 20 25
30 35 40 45 50 55 60
dBR; dBZ
Fig. 6.2. General probability-matched relations between radar reflectivity and rainfall rate. The heavy
dashed lines are for the Harrop Edge radar (NW England) data, and the other lines are as published
by Rosenfeld et aI. (1993)
117
Recent use of radar-gauge comparisons has attempted to match the probability
density functions of both radar estimates and raingauge measurements. The assumption made is that made by Calheiros and Zawadzki (1987) namely that the unconditional probability for any given rainfall intensity is constant everywhere in the radar
field, and that this probability was sampled in a representative way by raingauges. The
need for this assumption is removed by Rosenfeld et al. (1993) who assumed that the
conditional probability density function of rainfall intensity, regardless of the integrated rain depth for any given rainfall regime, is the same everywhere. The Ze:R
relationship so derived varies greatly with the rainfall regime and with the radial
distance from the radar
This work was developed further by Rosenfeld et al. (1994) and Rosenfeld et al.
(1995a, b), and has become known as the Window Probability Matching Method
(WPMM). Rainfall types are classified using the horizontal radial reflectivity gradients, the cloud depth as scaled by the effective efficiency, the bright band fraction
within the radar field window and the height of the freezing level. The results of tests
in Australia and Israel are very impressive, and comparisons of these results with
radar data for convective cases in North West England (Fig. 6.2) suggest a more
general applicability. It should be noted that the WPMM does not require that range
effects be removed or that the bright-band be corrected in the radar imagery, but
Convective Maritime
0.8
0.6
r::
0
0.4
' 13
c:
:l
0.2
u.
r::
.Q
0
'5
.0 .;::
·0.2
iii
is ·0.4
CI)
>
.' til ·0.6
'5
E -0.6
:>
U
-,
-'.2
, ",......
~ .. !I'" V; VI
/'
V
~...-,'" V V
J
I
, /
' :/ / /
, V
I
J
/
I,
, J
i'
. :
I
, ...-: J
/ '
.
/ /
I
Gau e CDF 1."'./
,.
1 I
:~
I
.. I
I
.... ,' ,
...--...r 11/
I - - Ra~ar ~DFs ..--l
I
- - Raingauges
........ ..... , Radar 20 - 49km I'
I
- -- Radar 50 - 79km
- - Radar 80 - 109km V
- - Radar 110 · 139km
I
-, .4
-20 -15 ·10 -5
0
5
10 15 20 25
30 35 40 45 50 55 60
dBR; dBZ
Fig. 6.2. General probability-matched relations between radar reflectivity and rainfall rate. The heavy
dashed lines are for the Harrop Edge radar (NW England) data, and the other lines are as published
by Rosenfeld et aI. (1993)