FLOOD05893

• County: Statewide
• Basin: Statewide
• Title: Forecasting Heavy Rainfall from Rain Cell Motion using Radar Data
• Date: 6/2/1992
• Prepared For: Journal of Hydrology
• Prepared By: Elsevier Science Publishers
• Floodplain - Doc Type: Educational/Technical/Reference Information

386

g
~ 0.60
c
.'2
.~ C 5C
c

(al
08
0,6
.
0
"
>
c 04
0
'0
2<
0
u
02
0,0
0
(b)

P.J. BREMAUD AND Y.B. POINTIN

387

FORECASTING HEAVY RAINFALL FROM RAIN CELL MOTION

.::.'.5-::

CC 15

dHSJ- 15

o

:),70

~
E- -2
.s

,-~~

,-
__,_J

i

- ~ -, -

.
0 -4
"
>
c
,2 -- --
.~ -6 .
U J
0

20
STEP NUMBER

40

~:_ - I

[
8.4Cf
!
~.3C :
o

20
SEP NUMBER

40

(el

RlJ 15

dHS3+ \5

8

/-_:-
,-:-,'
.--

?
'E6
.s

.
o
"
>
c
2
.~ 2
U

o
o

20
STEP NUMBER

40

20
5 rr P NUMrJF:R

'0

(d)

Fig. 7. Efficiency curves or the methods PERS (dotted line), EXTRA (dashed line), SCOUT 11.0 (dash-
dolled line) and PARAPlUIE (solid line) according to the criteria (a) ce, (b) RD, (e) dHSL and
(d) dHS3+ for the Paris 1989 event.

Fig. 7 (continued).

PARAPLUIE method detects the motion of the intense rain cells (CEL
echoes) inside each storm and the SCOUT 11,0 method detects the individual
motion of each storm (f echoes), As the rain cells are small and have
somewhat homogeneous motions, the relative perfonnances of the EXTRA
and PARAPLUIE methods are closely related for that event. The slightly
worse performance of the SCOUT 11,0 method relative to the PARAPLUIE
method shows that the motion of intense rain cells is different from the motion
of the storms in which \they are embedded.
Therefore, the PARAPLUlE method appears in this case to provide

against false alarms, to forecast the location of important hydrological events
and to evaluate the quantity of the observed rain at the forecast time better
in general than the SCOUT 11.0 method,

CONCLUSIONS

The best way to forecast rainfall at very short range is to extrapolate the
radar echoes in real time. This forecasting procedure is particularly well suited
for urban hydrological purposes, which require spatially and temporally
detailed forecasts (with less than I km spatial resolution and a forecasting time
interval of a few tens of minutes), For this last application, we have developed

,