You are on page 1of 17

Prof. S. K.

Dash, IIT Delhi

Basics of Numerical Weather Prediction


The horizontal momentum equation,
r
dV r σ ∂φ r
+ fkˆ ×V = − ∇φ + ∇p s + F
dt p s ∂σ

Continuity equation,
r ∂ ∂p s
∇. ( ps V ) + ( p s σ& ) + =0
∂σ ∂t
Thermodynamic energy equation,
1 d ⎡ ∂φ ⎤ RT
⎢σ ⎥ + [ p s σ& + σ p& s ] = − Q
R dt ⎣ ∂σ ⎦ C p p
Prof. S. K. Dash, IIT Delhi
Hydrostatic equation,

∂φ RT
=−
∂σ σ
Surface pressure tendency equation,

∂p s
[ r
]
1
= − ∫ ∇. p s V d σ
∂t 0

and the moisture equation,



∂t
[ r
]
[ p s q ] + ∇. p s q V + ∂ [ p s q σ& ] = p s S
∂σ

The above set of six equations can be solved in principle to get


r
the values of six unknowns viz., horizontal wind velocity V ,
surface pressure p s , temperature T , moisture q , geopotential
r
φ and sigma velocity σ& . The meteorological parameters V , p s ,
T and q are time dependent while σ& and φ are called the
diagnostic fields.
Prof. S. K. Dash, IIT Delhi
These equations constitute a closed system, which can be
solved at all future times
™ From a given initial state
™ With the prescribed boundary conditions

The governing equations can be written as


∂X
= D( X ) + P( X )
∂t
Here, X (x, y, z, t) : any model variable e.g. wind, Temperature,
Humidity, Surface pressure etc.

D : Dynamics e.g. advection, pressure forces etc.

P : Physics e.g. evaporation and Condensation of


water, solar heating, infra-red cooling, Frictional
drag at the surface of the earth etc.
Prof. S. K. Dash, IIT Delhi

Richardson Factory (1922)


Picture Credit : A. Lannerback
Prof. S. K. Dash, IIT Delhi

Classification of Forecasts
Usually a weather forecast is for a given region
and has a definite period of validity. According to
the period of validity one can classify as:
– Nowcasting : 0-6hrs
– Short range : up to 3 days
– Medium range : 3-10 days
– Long/Extended range : more than 10 days
Monthly and seasonal means
-- Climate : 30yrs or more
Prof. S. K. Dash, IIT Delhi

Important Surface Boundary Conditions

¾ Sea Surface Temperature


¾ Snow Cover and Snow Depth
¾ Soil Moisture
¾ Vegetation Cover
Influence of initial and surface boundary conditions at different time scales of model
integrations:
Now-casting and
Short-range (1-3 days): Primarily initial values of atmospheric parameters
Medium-range (3-10 days): Both the initial values and surface conditions
Monthly and more : Mostly surface boundary conditions
Climate: Chemical composition of atmosphere
Prof. S. K. Dash, IIT Delhi
Weather observations and
Forecasting
Values of meteorological parameters and
weather conditions at a future time
constitute weather forecasting:
• Values of Temperature, Wind, Humidity,
amounts of Rainfall and Snowfall
• Cyclones, Clouds, Sunshine hours, Fog
• Heat & cold wave conditions, Extremes
Prof. S. K. Dash, IIT Delhi

Important sources of weather/climate data


a) Surface
Surface observatories
Automatic Weather Stations
Ships
b) Upper air
Balloons
Aircrafts
Radars
c) Satellite derived fields
Prof. S. K. Dash, IIT Delhi

Some landmarks in Weather Forecasting:


–Invention of barometer (1643)
Torricelli
–Invention of telegraph (1845)
Synoptic map displayed in London
in 1851
–1850-1920 National Weather
Organisations grew. Primary tool of
forecast was the surface pressure
map and the forecasts were done
by empirical means.
Prof. S. K. Dash, IIT Delhi

- V. Bjerknes (1904) recognised forecasting as an


initial value problem.
He realised that the basic set of equations are very complicated with
highly nonlinear terms and these partial differential equations do not
possess closed solutions. However, there were inadequate data to
define initial conditions at that time.
- L.F. Richardson (1922) sought to solve the system of
equations by desk calculator
A failure – forecasted pressure changes were found one
order of magnitude greater than the actual.
- 1930s invention of radiosonde for upper air data and
further empirical improvements in forecasting.
Prof. S. K. Dash, IIT Delhi

Synoptic conditions
(mean sea level )
Prof. S. K. Dash, IIT Delhi

Synoptic conditions
(850 hPa analysis)
Prof. S. K. Dash, IIT Delhi
Different Methods of Forecasting
Synoptic, Numerical Weather Prediction (NWP) and
Statistical Methods
Synoptic method is based on examination of synoptic
charts and close watch of weather systems.
NWP is based on basic equations and physical principles:
Initial conditions and boundary conditions are important.

Forecasting techniques for medium range are basically different from short
range, because the extrapolation technique decreases in accuracy as the
period of extrapolation increases. For 10 days forecast one needs data from all
over the globe.
For longer days of forecasts, slowly varying boundary conditions and their
variations become more important – Sea Surface Temperature (SST)
anomalies, variations in surface albedo, snow cover and depth, soil moisture
etc.
In the climate time scale, atmospheric chemistry plays a dominant role.
Prof. S. K. Dash, IIT Delhi

• Statistical : Regression equation


Simple y = a + bx
y : predictant
x : predictor
Multiple linear regression equation
y = a + b1x1+ b2x2 +….+ bnxn
Prof. S. K. Dash, IIT Delhi

Growth of NWP due to the factors:

1. Theoretical research by Rossby, Petterssen,


Bjerkness, Charney, Eady, Eliassen etc. led to
some direct application to practical problems:
Radical departure from pure empiricism.

2. Large data collected by satellites, balloons and


rockets.

3. High Performance Computers (HPC)


Prof. S. K. Dash, IIT Delhi

Accuracy of weather prediction by NWP


limited by errors due to:

1) Truncations associated with the replacement of the


fundamental atmospheric (partial differential) equations
by numerical schemes.

2) Errors in the initial conditions of the atmospheric


variables in terms of their measurements.

3) Approximations in physical processes in the


parameterization schemes in numerical models..
Prof. S. K. Dash, IIT Delhi

Necessity of Global Models


• For short range forecasting Limited Area Models (LAM)
can be used. It usually takes a couple of days for
influences from the distant parts to reach the area.

• For medium and long-range forecasts:


Impact of snow, sea surface temperature (SST),
deserts and other surface conditions are important.
– The reflection of sun’s energy back to space from
clouds.
– Influence of large mountain ranges must also be
included.
These forcings are of global scale and their effects are
cumulative. Hence, for medium and long/extended
range forecasts global models are necessary.

You might also like