Verification Precipitation verification (overestimation): a common view of the behaviour of the LM, aLMo and LAMI Francis Schubiger and Pirmin Kaufmann, MeteoSwiss Uli Damrath, DWD Elena Oberto, Marco Turco, Paolo Bertolotto, ARPA Piedmont
Dec 27, 2015
Verification
Precipitation verification (overestimation):a common view
of the behaviour of the LM, aLMo and LAMI
Francis Schubiger and Pirmin Kaufmann, MeteoSwissUli Damrath, DWD
Elena Oberto, Marco Turco, Paolo Bertolotto, ARPA Piedmont
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 2
Verificationdifference of mean monthly rain gauges in D and CH
C. Schraff, DWD
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 3
VerificationVerification of aLMo Winter 03/04 over Switzerland
full line: obs (ANETZ); dashed: black: aLMo, red: LM
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 4
VerificationVerification of aLMo Winter 02/03 over Switzerland
full line: obs (ANETZ); dashed: black: aLMo, red: LM
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 5
VerificationVerification of the cloud ice testsuite in aLMofrom 04.02-19.03.2004
full line: obs (ANETZ); dashed: black: aLMo
blue: aLMo-cloud ice, red: LM
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 6
VerificationVerification of aLMo Winter 04/05 over Switzerland
full line: obs (ANETZ); dashed: black: aLMo, red: LM
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 7
Verification
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 8
VerificationVerification with raingauges in D and CH
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 9
VerificationVerification with raingauges in CH
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 10
VerificationVerification of aLMo Winter 04/05 SYNOP
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 11
VerificationPrecipitation: scores of aLMo Winter 00/01
verification with 12h-sums from SYNOP
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 12
VerificationPrecipitation: scores of aLMo Winter 01/02
verification with 12h-sums from SYNOP
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 13
VerificationPrecipitation: scores of aLMo Winter 02/03
verification with 12h-sums from SYNOP
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 14
VerificationPrecipitation: scores of aLMo Winter 03/04
verification with 12h-sums from SYNOP
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 15
VerificationPrecipitation: scores of aLMo Winter 04/05
verification with 12h-sums from SYNOP
VerificationLatest results in the precipitation verification over Piedmont
Elena Oberto, Marco Turco, Paolo Bertolotto(*) ARPA Piemonte, Torino, Italy
• Eyeball verification: seasonal cum maps of D+1 forecast
• Seasonal statistical indices
• Special focus over last winter (DJF’05): Statistical indices for increasing thresholds
Scatter plot of daily precipitation over each basins: seasonal and monthly
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 17
Verification
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 18
Verification
Prec (mm)
ALMO
LAMI
LOKAL
DJF 2004
OBSERVED
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 20
Verification
Scale (mm)
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 21
VerificationOBSERVED
Prec (mm)
aLMo 00
LM-DWD 00
LAMI 00
Verification over Piedmont ; winter 04/05
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 22
VerificationSeasonal trend of BIAS and ETS Threshold=10mm/24h Forecast time: +00/+24.
The error bars indicates 2.5th and 97.5th percentiles of resampled distribution, applied to the "reference" model.Djf’05 too much dryno statistic
in the next slides there is a special focus
LoKalaLMoLAMI
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 23
VerificationSeasonal trend of BIAS and ETS Threshold=10mm/24h Forecast time: +24/+48.
The error bars indicates 2.5th and 97.5th percentiles of resampled distribution, applied to the "reference" model.
aLMo
LAMI
LoKal
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 24
VerificationConclusions
a possible cause for the precipitation overestimation is the cloud ice scheme
but another causes (masked by the precipitation variability) are also possible, because
already before the introduction of cloud ice scheme we had seasons with precipitation overestimationwe have regions (especially in southern Europe) with no changes in the yearly (seasonal) biases in the last years
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 25
VerificationOutlook (1)
more work necessary to fix the causes for precipitation overestimation
scatter plots of daily precipitation sums 'LM vs raingauges, SYNOP-stations, RADAR areal mean’ : this helps to see if this overestimation is due to (some few) cases where the model gives (much) precipitation that is not observed, or whether overestimation of precipitation is a problem in all cases
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 26
Verification
winter 2004(djf)
aLMo D+1
Scatterplot of mean daily precipitation over the Piedmont basins
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 27
Verification
winter 2004
(djf)
Lokal D+1
Scatterplot of mean daily precipitation over the Piedmont basins
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 28
Verification
winter 2004
(djf)
LAMI D+1
Scatterplot of mean daily precipitation over the Piedmont basins
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 29
Verification
winter 2005(djf)
aLMo D+1
Scatterplot of mean daily precipitation over the Piedmont basins
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 30
Verification
winter 2005
(djf)
Lokal D+1
Scatterplot of mean daily precipitation over the Piedmont basins
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 31
Verification
winter 2005
(djf)
LAMI D+1
Scatterplot of mean daily precipitation over the Piedmont basins
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 32
Verificationscatterplot daily precipitation Germany: August 2002
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 33
VerificationScatterplot daily precipitation Germany: December 2002
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 34
VerificationScatterplot daily precipitation Germany: December 2004
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 35
VerificationOutlook (2)
more work would be necessary to fix the causes for precipitation overestimation
scatter plots of daily precipitation sums 'LM vs raingauges, SYNOP-stations, RADAR areal mean’ : this helps to see if this overestimation is due to (some few) cases where the model gives (much) precipitation that is not observed, or whether overestimation of precipitation is a problem in all cases
set up a simple conditional verification by discriminating:
between events of different (observed) vertical stability i.e., unstable = convective (?), stable = stratiform (?)convective/stratiform precip in the model. This would help to attribute the problem to the grid-scale precipitation scheme or the convection scheme, or something else.
franci
s.sc
hubig
er@
Mete
osw
iss.
chC
OSM
O G
enera
l M
eeti
ng,
Zuri
ch,
20
-23
.09
.05 36
VerificationOutlook (3)
strong link to WG1-WG3 for sensitivity studies on a set of selected cases / episodes:
analysis / data assimilation
dynamics and numerics
physics