Lennart Bengtsons presentation.

The origin of numerical weather prediction in
Sweden and its position in the world
Looking back in time 50+ years
Lennart Bengtsson
2015-10-14
SMS MISU NWP Sweden
1
Content of talk
• Rossby, Charney och von Neumann
• IMI, BESK och barotropa modellet
• Varför satsade vi på filtrerade modeller?
• SMHI och 3-parameter modellen
• Det internationella arbetet och vägen till ECMWF
2015-10-14
SMS MISU NWP Sweden
2
Wilhelm Bjerknes, 1862-1951
the father of modern meteorology
• Das Problem der Wettervorhersage, betrachtet vom
Standpunkte der Mechanik und der Physik, Meteorologische
Zeitschrift, January 1904, pp 1-7.
• Bjerknes postulated that the necessary and sufficient
conditions for the rational solution of forecasting problems
are:
• 1. A sufficiently accurate knowledge of the state of the
atmosphere at the initial time.
• 2. A sufficiently accurate knowledge of the laws according to
which one state of the atmosphere develops from another.
2015-10-14
SMS MISU NWP Sweden
3
Lewis Fry Richardson
1881-1953
2015-10-14
• Richardson was an English
mathematical, physicist,
meteorologist, psychologist
and pacifist.
• He published in 1922
“Weather Prediction by
Numerical Process” that
was the first attempt to
numerically solve the
physical equations.
SMS MISU NWP Sweden
4
( Figure courtesy, Norman Phillips)
2015-10-14
SMS MISU NWP Sweden
5
First real-time operational NWP in Sweden
2015-10-14
SMS MISU NWP Sweden
6
A 72 hr forecast and its verification
The first ever operational forecast? Credit: B. Söderberg
2015-10-14
SMS MISU NWP Sweden
7
Prognosområden vid SMHI
2015-10-14
SMS MISU NWP Sweden
8
Varför 3-parametermodellen?
• D21s hastighet och minne var begränsat så det gällde att
optimera de resurser som fanns.
• Baserat på numeriska experiment med baroklina vågor på ett
beta-plan visade sig att en god approximation var att använda
1000, 500 och 300 hPa som nivåer.
• Detta visade sig vara fallet och 3-parametermodellen spelade
stor roll att övertyga meteorologerna att numeriska modeller
var användbara och rätt väg att gå.
2015-10-14
SMS MISU NWP Sweden
9
Presentation of weather maps in
the early 1960s
2015-10-14
The early computer at SMHI
SMS MISU NWP Sweden
10
Varför inte primitiva modeller?
• Dessa var inte bättre än de quasi-gestrofiska av samma skäl
som gjorde att Richardson misslyckades, nämligen
initialiseringen.
• Det var först Bennert Machenhauer som löste detta dilemma
vid ECMWF.
• Alternativet var att använda en dämpande tidsintegration men
detta tog musten ur de intensiva ovädren som man var mest
intresserad av i prognostjänsten
2015-10-14
SMS MISU NWP Sweden
11
Låt oss göra ett tankeexperiment!
• Antag att någon från dagens ECMWF hade stigit på H G
Wells tidsmaskin och besökt SMHI 1 mars 1965 med denna
dags prognosberäkning under armen
• Vad hade detta visat?
• Eftersom jag lyckades med detta magiska trick med hjälp av
Patrick Laloyaux vid ECMWF så kommer här resultatet:
2015-10-14
SMS MISU NWP Sweden
12
1 March 1965
2015-10-14
SMS MISU NWP Sweden
13
Dåtid: Prognos vid ECMWF hösten 1975
2015-10-14
SMS MISU NWP Sweden
14
Observationer 1 Mars 1965
2015-10-14
SMS MISU NWP Sweden
15
Nutid: ECMWF sommaren 2015
2015-10-14
SMS MISU NWP Sweden
16
Predictions 1965 and 2015
At 80% + 5days, at 60% + 6 days
2015-10-14
SMS MISU NWP Sweden
17
Results
Twenty-nine medium-range forecasts produced by the CERA system at 00 UTC
between 20 February and 20 March 1965 have been used to compute similar
forecast skill scores.
The anomaly correlation of the CERA 500-hPa geopotential height forecast has
been calculated as a function of lead time for the northern hemisphere
extratropics with respect to the CERA analysis.
The 80% anomaly correlation skill level is reached at 6 days and the 60% level at
8.5 days. This means that the use of a modern ECMWF assimilation and forecasting
systems extends those skill levels by 4.5 and 5 days, respectively.
2015-10-14
SMS MISU NWP Sweden
18
Vad har hänt på 50 år?
• Sedan 35 år har vi globala väderprognoser
• Prognoskvaliten har förbättrats med 4-6 dygn
• Extrema vädersystem som tropiska orkaner prognoseras
rutinmässigt
• Modellberäkningarna baseras på de primitiva ekvationerna eller
icke-hydrostatiska ekvationer.
• Vi kan kartlägga det globala klimatet baserat på dagliga
väderanalyser där analysen sker i 5- dimensioner!
2015-10-14
SMS MISU NWP Sweden
19
End
2015-10-14
SMS MISU NWP Sweden
20
The evolution of meteorology around 1850
The scientific reasons
•
•
•
Studies of synoptic networks of pressure, temperature and weather had
indicated typical structures in time and space suggesting that forecasts were
feasible.
A broad empirical understanding of the global climate had evolved over the
years through different studies starting with Hadley in the previous century.
The evolution of a theoretical framework in mechanics, thermo-dynamics and
hydrodynamics had evolved with contributions by the leading physicists of
the 19th century. A key role was played by Ecole Normal and Ecole
Polytechnique in Paris, e.g Laplace ( 1749-1827), Cauchy (1789-1857) and
Navier (1785 -1836).
2015-10-14
SMS MISU NWP Sweden
21
Richardson‘s dream
2015-10-14
SMS MISU NWP Sweden
22
Weather prediction according to Bjerknes
• 1. The three hydrodynamical equations of motion. These are
differential relations among the three velocity components, the
density, and pressure.
• 2. The continuity equation, which expresses the principle of the
conservation of mass during motion. This equation is again a
differential relation between the velocity components and the
density.
• 3. The equation of state of atmospheric air, which is a relation in
finite form among the density, pressure, temperature, and humidity
of a given mass of air.
• 4. The two laws of thermodynamics, which allow us to set up two
differential relations giving the rates of change of energy and
entropy during the changes of state that are taking place.
2015-10-14
SMS MISU NWP Sweden
23
The principle
of error
reduction in
data
assimilation
2015-10-14
SMS MISU NWP Sweden
24