Use and evaluation of medium-range Use and evaluation of medium-range forecast at the forecast at the Hungarian Meteorological Service.
Ferenc WantuchHungarian Meteorological Service Kitaibel P. u. 1. H-1024 Budapest
E-mail:[email protected]
Talk outline
Introduction
Verification
Two interesting case studies
New products based on ECMWF.
Data flow and structure
NWP data source:
File format: GRIB BUFR
Visualisation
Netcdf file format
Code generalisation process
NWPNWPoutputoutput
Code generalisationCode generalisationby the help of
different method (PP, MOS)visibility forecast,
max, min temperaturescloudiness, ...
Forecast text Forecast text generator programgenerator program
End usersEnd users
Text and dataControl
VerificationVerification
Forecaster Forecaster outputoutput
Verification
A Complex Score was defined.
100% 0%
Min/Max temperature RMSE 0 °C 8 °C
Precip. Existence PC 100% 0%
Precip. Amount MAE 0 kateg. 1.6 kateg.
Average Wind Speed RMSE 0 m/s 6 m/s
Wind Gust RMSE 0 m/s 10 m/s
Cloud Cover RMSE 0 okta 6 okta
Complex Score - 100 % 0 %
Weight
Temperature 25+25%
Precipitation 22,5%
Wind 10%
Cloud Cover 17,5%
10
46
17,5
25 25
12,5
0
5
10
15
20
25
Tmin Tmax P.Exist P.Amounty Áv.Windl Wind gust Cloud Covert
%
1.night. 1.day 2.day 3.day 4.day 5.day 6.day
MAE (°C)
1.87
1.31
1.20
-
-
-
-
1.48
1.37
-
1.71
1.53
-
1.91
1.72
-
2.26
2.00
-
2.50
2.24
MAE (°C)
-
-
-
1.77
1.55
1.30
-
1.72
1.63
-
1.89
1.84
-
2.28
2.16
-
2.54
2.43
-
3.12
2.79
Prec.Exist.: PC (%)
79.95
82.73
83.11
76.76
80.36
81.74
-
80.92
78.60
-
73.54
76.28
-
74.23
75.46
-
73.21
74.00
-
71.03
70.77
Prec. amount: MAE
0.19
0.22
0.22
0.23
0.25
0.24
-
0.35
0.33
-
-
-
-
-
-
-
-
-
-
-
-
Verification 2004 (Hungary) ALADIN ECMWF Forecaster
Tmax:
Tmin:
1.night. 1.day 2.day 3.day 4.day 5.day 6.day
Average wind speed: MAE (m/s)
0.73
0.62
0.94
0.67
0.59
0.86
-
0.65
0.84
-
0.69
0.87
-
0.82
0.94
-
0.88
0.97
-
0.98
1.02
Gust: MAE
(m/s)
1.58
1.43
1.61
1.74
1.58
1.65
-
1.78
1.71
-
2.03
1.90
-
2.17
2.03
-
2.42
2.20
-
2.59
2.32
cloud cover : MAE
(okta)
1.08
0.86
0.65
1.08
0.87
0.68
-
0.98
0.79
-
1.10
0.86
-
1.31
1.01
-
1.46
1.12
-
1.65
1.19
Complex Score (%)
74.96
79.81
81.73
75.03
77.68
80.42
-
75.91
77.55
-
73.04
75.22
-
68.69
71.35
-
65.13
68.19
-
60.77
64.82
Verification results 2004 (Hungary) ALADIN ECMWF Forecaster
Felhőzet:
Verification results and trends
Verification results and trends
60,00
65,00
70,00
75,00
80,00
85,00
2000
. sze
pt.
2000
. dec
.
2001
. sze
pt.
2001
. dec
.
2002
. sze
pt.
2002
. dec
.
2003
. sze
pt.
2003
. dec
.
2004
. sze
pt.
2004
. dec
.
%
AL EC SzLineáris (AL) Lineáris (EC) Lineáris (Sz)
Complex Score for the first day 2000. júl – 2004. dec. (Country mean) ALADIN ECMWF Synoptic
2000. second period: changes in ECMWF algorithm + model upgrade: 60 => 40 km)
2003. Feb.: Anticyclon with snow surface, ALADIN is unsuccesfull
2004. December cold air pad
Minimum Temperature sesonal ME and RMSE
(2001 – 2004 Hungary mean)
ME: seasonal
RMSE: slowly decrease
EC + Forecaster:Minimum Temperature
-1
-0,5
0
0,5
1
1,5
2
2,5
3
3,5
4
2001win
2001sum
2002win
2002sum
2003win
2003sum
2004win
2004sum
°C
Tmax
-2
-1,5
-1-0,5
0
0,5
1
1,5
22,5
3
3,5
2001win
2001sum
2002win
2002sum
2003win
2003sum
2004win
2004sum
°C
Total Cloud Cover
-1
-0,5
0
0,5
1
1,5
2
2001win
2001sum
2002win
2002sum
2003win
2003sum
2004win
2004sum
octa
AL(ME) EC(ME) Fo(ME)AL(RMSE) EC(RMSE) Fo(RMSE)
50
60
70
80
90
100
jan febr márc ápr máj jún júl aug szept okt nov dec
%
AL EC Sz
Complex Score for the first day in 2004-ben (country mean)
ALADIN ECMWF Forecaster
~30%-correction in (cold air pad) inversion situation
Case study I. : Inversion and cold air pad.
(2004. December )
Stratus (cloud base 200-400 m, surface temperature –1, -3 C
Tops of mountains are clear
MSG comp321 sat. picture. 12.12. 12 2004 UTC
and observations ECMWF model forecast.
12 UTC-re dec.12 2004
The last ECMWF prediction.
100
20
SKC is predicted
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
okt. 1
.
okt. 8
.
okt. 1
5.
okt. 2
2.
okt. 2
9.
nov. 5
.
nov. 1
2.
nov. 1
9.
nov. 2
6.
dec. 3
.
dec. 1
0.
dec. 1
7.
dec. 2
4.
dec. 3
1.
(okt
a)
AL EC F
Cloud cover verification. ECMWF - ALADIN – Forecaster (1st day)
2004. October – November – December V
Case study II. : Flash flood in Matra Mountains.
(18.04.2005. )
The ECMWF model involved the possibility of
extreme precipitation
Succesful MM5 modell forecast based on ECMWF
initial condition.
Case study II.
New clusterisation method and early time torch construction depend on ECMWF at HMS.
Varying number of clusters for the Carpathian Basin
Torch construction before the dissemination time.
Clustering for Central European area
developed by Istvan Ihasz at HM S
Varying number of clusters are selected based on 120 hours of 500 hPa geopotential fieldCentral European Area : ( 55.00N 10.00E 40.00N 30.00E )
Clustering forEurope made byECM W F
Clustering forCentral Europemade by HM S
New torch diagrams and probability forecasts for energetic sector.
925 hPa,850 hPa,700 hPa, wind direction ,relativ hum.,temperture,… torches
Probablilty forecasts
Post processing application.
• Program demonstration
• Internet and business application
• End
Thank you for attention !