|
<br />384
<br /> 0.90
<br /> 0.80
<br />. 0.70
<br />,
<br /><;
<br />>
<br />c 0.60
<br />,9
<br />.~ 0.50
<br />u ,
<br /> .'
<br /> 0.40 ,
<br /> 0,30
<br /> 0
<br />(,)
<br /> 08
<br /> 0,6
<br />.
<br />,
<br />~
<br />c 04
<br />,9
<br />.~
<br />U
<br /> 0,2 --':':~
<br /> 0,0
<br /> 0
<br />(b)
<br />
<br />,
<br />
<br />P.J. BR~MAUD AND Y.B. POINTIN
<br />
<br />FORECASTING HEAVY RAINFALL FROM R.AIN CELL MOTION
<br />
<br />385
<br />
<br />dHS5- 88
<br />
<br />CC 88
<br />
<br />o
<br />
<br />,~..--
<br />
<br />'2
<br />"- -2
<br />E ,
<br />.s ---'
<br />. ~ ,,- -
<br />, -4
<br /><;
<br />>
<br />c
<br />0
<br />,
<br />-" -6
<br />"5
<br />
<br />
<br />--,--
<br />
<br />
<br />20
<br />STEP NUMBER
<br />
<br />40
<br />
<br />20
<br />STEP NUMBER
<br />
<br />40
<br />
<br />RI5 B8
<br />
<br />(e)
<br /> 8
<br /> ,
<br />'2 ,
<br />"- 8 ,
<br />E ,
<br />.s
<br />.
<br />, 4
<br /><;
<br />>
<br />c
<br />.g
<br />-" 2
<br />"
<br />u
<br />
<br />
<br />dHS5;- 88
<br />
<br />
<br />20
<br />STEP NUMBER
<br />
<br />40
<br />
<br />o
<br />o
<br />
<br />40
<br />
<br />20
<br />STEP NUMBER
<br />
<br />(d)
<br />
<br />Fig. 6. Efficiency curves of the methods PERS (dotted line), EXTRA (dashed line) and PARAPLUIE
<br />(solid line) according to tbecritcria (a) ce. (b) RIS, (e) dHSS_ and (d)dHSS+ fOf the Cevennes 1988 event.
<br />
<br />Fig. 6 (continued).
<br />
<br />others (the SCOUT 11,0 method requires pictures without ground echoes), As
<br />for Figs, 6(a)-6(d), the PERS, EXTRA, SCOUT 11,0 and PARAPLUIE
<br />methods are depicted with dotted line, dashed line, dash-dotted line and solid
<br />line, respectively, The Paris 1989 event is characterized by the development of
<br />isolated stonns, and the interval between radar pictures is 5 min.
<br />The results of Figs, 7(a)-7(d) show that the PERS method gives a signifi-
<br />cantly poorer performance during the Paris 1989 event than during the
<br />Cevennes 1988 one, This is due to the smaller number of rainy pixels in this
<br />event for which an efficient rainfall forecast is really needed, As for the results
<br />of Figs, 6(a)-6(d), the PARAPLUIE method tends to forecast the rain better
<br />
<br />than the other methods, but the differences between the PARAPLUIE and the
<br />EXTRA methods are smaller for the Paris 1989 event because the motions of
<br />the radar echoes are much more homogeneous during that event than during
<br />the Cevennes 1988 one,
<br />According to the efficiency curves of the CC, RI and, in particular, the dHS
<br />criteria, shown in Figs, 7(a), 7(b) and 7(d) respectively, the SCOUT II,O result;
<br />are characterized by slightly worse performances than that of PARAPLUIE,
<br />Moreover, the SCOUT II,O method tends to underestimate the observed rain
<br />more than the PARAPLUIE method does (Fig, 7(c)), During this event, the
<br />EXTRA method detects the global motion of all the isolated stonns, the
<br />
<br />
|