User experience with extended range forecasts -- climatic aspects of ECMWF products Christof Appenzeller Wolfgang Müller Heike Kunz Mark Liniger ERA-40.
Post on 27-Mar-2015
214 Views
Preview:
Transcript
User experience with extended range forecasts-- climatic aspects of ECMWF products
Christof AppenzellerWolfgang MüllerHeike KunzMark Liniger
ERA-40 compared to Swiss observations
Verification of seas. forecasts monthly forecasts in electricity
market seas. forecasts in media
2
ERA-40 T2 Upscaling
ZERA-40 - ZSchw.Stat. ~ 40m
3
ERA-40 T2 seasonal CycleT
emp
erat
ure
[°C
]
Temperature:
ERA-40, Swiss Stations
16
14
12
10
8
6
4
2
0
- 2
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
T[°
C]
Temperature difference:
ERA-40 – Swiss Stations
1.2
0.8
0.4
0.0
-0.4
-0.8
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
ERA-40 in Summer 0.8°C too high
ERA-40 in Winter 0.4°C too low
4
ERA-40 & GPS IWV: Correlation
Morland et al., submitted
5
IWV: Height dependent bias
Morland et al., submitted
6
System 2: Climate variability
ERA-401959 - 2001
System 2Nov. Fcsts.1987 - 2001
Müller et al. (2005), Clim. Dyn.
DJF North Atlantic Oscillation (NAO)EOF Z500
7
Predictability of NAO
Müller et al. (2005), Clim. Dyn.
regression of Z500 fields
8
Impact of climate variability
ERA-401959 - 2001
System 2Nov. Fcsts.1987 - 2001
Müller et al. (2005), Clim. Dyn.
DJF North Atlantic Oscillation (NAO)SVD Z500 & T2m
9
Predictability of NAO Impact
Müller et al. (2005), Clim. Dyn.
regression of T2m fields
10
Verification grid point
RPSSdT2m, 1987 – 2002, fc234, 3 cat.
Schwierz et al., subm.
against ERA-40
Perfect model approach
11
extended range forecast products at MeteoSwiss
seasonal forecasts T2m, Precip. Klimagramms probability maps
monthly forecasts T2m, Precip, Geopot. tercile maps
Klimagramm
12
Weather risk application
months
tem
pera
ture
ano
mal
y
Seasonal Forecast
Probabilistic Downscaling
local weather risk
13
Electricity Market
Electricity company in Southern Switzerland Monthly forecasts
14
Using monthly forecasts Contracts at EEX (European Energy Exchange) Lowermost tercile for S-Germany (spatial avg).
Forward Prices vs. Temperature
0
10
20
30
40
50
60
02.0
2.2
005
04.0
2.2
005
06.0
2.2
005
08.0
2.2
005
10.0
2.2
005
12.0
2.2
005
14.0
2.2
005
16.0
2.2
005
18.0
2.2
005
20.0
2.2
005
22.0
2.2
005
24.0
2.2
005
26.0
2.2
005
28.0
2.2
005
02.0
3.2
005
04.0
3.2
005
06.0
3.2
005
08.0
3.2
005
10.0
3.2
005
12.0
3.2
005
14.0
3.2
005
16.0
3.2
005
18.0
3.2
005
[€/M
Wh
]
0
10
20
30
40
50
60
70
80
90
100
[%]
Germany Peaks Wk12- 05
Temperature low er tercile
EE
X E
u/M
Wh
low
erm
ost t
erci
le %
Lead time
Contracts for peak load in week 12 (21.-27.3.2005)
T2 Zurich
• persistence?• other forecasts?• ?
15
Summer 2005 in Switzerland
ProbabilityT2m > norm
Klimagramm for Swiss-Temperature
16
MediaBasis MeteoSwiss does not intend to publish seasonal forecasts
for Switzerland. Public presentation of Swiss climate research programme.Interview Probability of Temperature in JJA > norm. Signal: 40-50%. Norm value: 17 oC. (JJA daily avg). Hit rate around 55 to 60%. Ocean information is important.Published Headline:
“MeteoSwiss: Summer will be cool” 40% used as signal. Hit rate compared to skill of 84% for next day forecast. Map of current buoys as basis for forecasts.
17
same and next days:
prime time news on national television (French and German).
several radio interviews. copied by several news paper (without feedback). scientific facts and context somewhat distorted and
wrongly quoted. main message: cool summer
18
19
Competing Sunday Press Journal Front page
“Summer 05: Weather-forecast destroys tourism!Tourism directors criticize meteorologists”
Headline“Weather forecasters drive away tourists”
Short street interviews → people would use seasonal forecasts for holiday planning.
Tourism industry: “Other countries would never have the idea to talk badly about their own products”
Forecasts are reason for low earnings in tourism industry ?! “Would the forecasts predict a nice summer, we would be happy”
20
Conclusions for handling press
JJA average of 17 oC is cool. Give clear numbers, no ranges/choices → extremes will be chosen Available actual forecasts will be published and used. A “scientific”
context or “experimental” label can get lost. Observations are preferred to models as technological basis for
forecasts. Should seasonal forecasts be published for the public? if yes, how?
Bulletin, press release, selected interviews, individually on request Tourism industry: stronger fluctuations due to weather and climate
forecasts? Tourists (would) use forecasts for booking, industry seems not to.
top related