synoptic situation modern forecasts of 1953 storm (‘reforecasting’) expand in three ways:
DESCRIPTION
The 1953 North Sea Gale in Perspective of Historical NWP efforts . H. M. van den Dool, R.E. Kistler and S. Saha, at NCEP. Synoptic situation Modern Forecasts of 1953 storm (‘ReForecasting’) Expand in three ways: a) Barotropic Forecasts - PowerPoint PPT PresentationTRANSCRIPT
Synoptic situation Modern Forecasts of 1953 storm(‘ReForecasting’)
Expand in three ways:• a) Barotropic Forecasts• b) Initial error hand vs present analysis (’53 data)• c) (nearly) Contemporary UK NWP for ’53 case• Acknowledge Anders Persson about point c.
The 1953 North Sea Gale in Perspective of Historical NWP efforts.
H. M. van den Dool, R.E. Kistler and S. Saha, at NCEP
Mean Sea Level Pressure Maps for Jan 29 – Feb 1, 1953, every 3 hours from NCEP-NCAR Reanalysis (Kalnay et al
1996; Kistler 2001
29/0
30/0
31/0
01/0
From Hay, R. F. M., and J. Laing, 1954: The storm of 31st January - 1st February 1953. Mar. Obs., 24, 87-91.
Height fields Jan 28 - Feb 1, 1953 every 6 hours (Reanalysis)
• To be sure: This disaster was not just meteorology1) Oceanography, wind stress effect, Kelvin waves and astronomical tides, wind waves2) ‘Coastal’ defense, and its maintenance3) Preparedness, warning system, protocol, bureaucracy, communications, evacuation4) Assessing the risk of a given surge forecast
Forecasts valid for Feb, 1, 1953, 3Z
• Reanalysis model T62L28
The previous map is a contemporaneous hand analysis, digitized at 2.5 by 2.5 grid. The
original looks like this:
Source: KNMI, F. H. Schmidt- editor,1960: Meteorologische en Oceanografische aspecten van stormvloeden op de Nederlandse kust. Bijdrage tot het rapport van de Deltacommissie door het KNMI. Staatsdrukkerij- en Uitgeverijbedrijf - ‘s Gravenhage. pp200.
Synoptic situation Modern Forecasts of 1953 storm
(‘ReForecasting’) by NCEP(several resolutions; ETA), ECMWF(several resolutions), Hirlam
Expand in three ways for Historical NWP
aspects:• a) Barotropic Forecasts• b) Initial error: hand-analysis vs Re-analysis (’53
data)• c) (nearly) contemporary UK NWP of ’53 case
dAVA (global barotropic) forecasts
/t (ζ’-λ2ψ’) = - Vn .ζ’ - V’. (ζn + f + ζ’) - D4 ζ’- K ζ’,
• where Vn and ζn refer to climatological mean flow, and ζ’ is ‘anomaly’ vorticity.
• T30L1• Cressman-Bolin correction applied to anomaly.
Parameter λ2 found thru adjoint study (Rinne et al 1993)
• 1980’s type barotropic model
Verification by numbers (Anomaly Correlation 500 mb
height)
65
70
75
80
85
90
95
100
A
nom
aly
Corr
ela
tion
26/12Z
27/12Z 28/12Z 29/12Z 30/12Z 31/12Z 01/12Z 02/12Z 03/12Z
Verifying Time
Skill of 36 hour dAVA forecastsdomain: 50W-30E and 40N-67.5N
65707580859095
100 A
nom
aly
Corr
ela
tion
26/12Z
27/12Z 28/12Z 29/12Z 30/12Z 31/12Z 01/12Z 02/12Z 03/12Z
Verifying Time
dAVA 36 hoursMRF 84 hrs
Skill of NWP forecastsdomain: 50W-30E and 40N-67.5N
20
40
60
80
100
z500 a
nom
aly
correla
tion
0 1224364860728496108120132144lead (hours)
dAVA
MRF
Forecasts verifying Feb, 01, 1953 03Z
On the initial error
• ‘hand’analysis minus Re-analysis is ~2mb/25gpm for MSLP/Z500 respectively. Atl-Eu domain for 1953 data.
• CDAS/Reanalysis (T62L28) has currently (Jan): 1.5mb/12.0gpm (0hr, fit to radiosondes) 1.8mb/18.1gpm (6hr, guess to obs) (EuroAtl)
It is (?) amazing that the initial error for 1953 Reanl, with so much less data than in 2004, is not even a factor of 2 larger.
INITIAL
30 jan 15Z
31 jan 15Z (anl)
+24 hr barotropic forecasts
QJ 1958
Analysis 31 Jan 15Z 1953, 500mb(ft)
+24 hr by Sawyer_Bushby model
24 hr forecast of 500mb height by Sawyer_BushbyModel, verifying Jan 31 15Z, Met Magazine1954
“If the Sawyer-Bushby forecast had been right, the calamity would have been a lot more
calamitous than it already was”. Free after ‘Discussion’(1954) and
Hinds(1981).
Water Level Forecast 12 Hours
Conclusions:• With current NWP models and analyses, but 1953
data, forecasts out to 2 (4) days appear possible. KNMI(1960) conclusion about lack of data
(easy scapegoat) is clearly overstated (for this case). • Accurate barotropic forecast are possible out to 30
hrs (i.e. after the bomb had developed)• Differences between hand analysis and Reanalysis
are ~ 2mb and 25gpm (1953 data coverage). Jan2003 CDAS errors are ~1.6mb/15gpm.
• There are a few early NWP efforts with regard to the storm of 1953. Jan 30/31 15Z was one of the canned cases used at UKMO. Bushby-Sawyer +24hr forecast was remarkable!
A KNMI initiated warning system, implemented in 1916 (following a previous flood)
Three options:1) Say nothing (benign circumstances)2) Warn for significantly enhanced water levels for the next two astronomical high tides (by region)3) Warn for dangerously enhanced water levels for the next one astronomical high tide.
Saturday Jan 31 at 10am: option 2 warning was issued Saturday Jan 31 17:15 (they had to wait until after the afternoon high tide, in order to ‘avoid confusion’): option 3 (for the first time in history)
(Warnings were issued to other authorities (supposedly they knew what to do), and to the public via radio at 18:00 onward. Note there was no radio after midnight. Zero communication during the disaster.)
RMS-errors of MRF(GFS) forecasts, and initial errors.
Z500
Lead — > 0 24 48 72 96 hr climo
------------------------------------------------------------------- 30 Jan 3Z 0(28) 24 43 57 96 140 31 Jan 3Z 0(27) 34 54 68 95 129 01 Feb 3Z 0(22) 24 57 73 108 137 gpm
MSLP
Lead — > 0 24 48 72 96 hr climo ------------------------------------------------------------- 30 Jan 3Z 0(1.8) 3.2 5.2 7.4 9.6 10.731 Jan 3Z 0(2.0) 4.1 6.7 8.1 7.4 12.8
01 Feb 3Z 0(2.4) 2.9 6.3 6.7 9.8 15.1 mbar