Weather Forecasting

WEATHER FORECASTING

How can simulation modeling be used in weather forecasting?

Simulation modelling is very important in weather forecasting and very large and expensive computer systems are used in forecasting the weather.

Weather forecasting requires a huge amount of (worldwide) data to be acquired, communicated to the computer then analysed using very complex calculations. This has to be done very quickly as weather forecasts are no use if out-of-date. This is clearly very expensive both in term of hardware and software. The computer itself will be very expensive, as will thousands of weather stations, many high in the atmosphere, etc or in remote inhospitable places like deserts or the Arctic. The software will involve large programs, which will be expensive to write and to test, and which may need regular expensive updating as weather patterns change and knowledge increases. Very high graphics capabilities will be needed for e.g. TV weather forecasting programmes where viewers will expect to see changing weather patterns displayed graphically.

How are models created?

Collect Data

Ø Large number of measurements such as pressure, temperature, humidity, etc.

Ø Readings from all around the world from satellites, balloons, etc.

Derive Model

Ø From laws of physics.

Ø Models provide a set of equations to solve / Produce a set of equations (mathematical model) which are solved to predict weather twice a day

Test / Compare

Ø Actual with observed.

Ø Create model / test the model.

Ø What if investigations.

Computers in weather systems

Information comes from all over the world. There are inputs from thousands of weather stations e.g. satellites, balloons, radar, ships, from huge geographical areas/whole world. Weather forecasts are made by collecting quantitative - numerical - data about the current state of the atmosphere Data is captured by using a variety of sensors In the case of weather models, data such as rain fall, temperature and wind speed are fed into a computer Data is transmitted and collected centrally from thousands of sensors This data is then put into a mathematical model
Requires the processing of a huge amount of data and the comparison with huge amounts of historical data. Predictions are made based on current conditions
Requires very complex calculations which will require large, complex programs. A series of calculations is performed on the raw data on it to determine how it will change over time Normally, mathematical modelling is done by powerful computers, which can carry out many calculations per second The computer uses equations produced from the scientific understanding of atmospheric processes such as fluid dynamics and thermodynamic equations.
Processing has to be done very quickly as weather forecasts are no use if out-of-date
Weather is often extremely unstable / chaotic / hard to predict
May require very good graphics for visual representation
Models are created by producing a set of equations (mathematical model) which are solved to predict weather twice a day. The more sophisticated and up-to-date your model is, the more accurate your forecast should be. Parallel processing is generally used for complex calculations in mathematical weather models
Collect data, compare actual with observed, create model, test the model.
Large number of measurements such as pressure, temperature, humidity.
Models are created by producing a set of equations (mathematical model) which are solved to predict weather twice a day. All of the above information is put into the model to ensure it is realistic and as reliable as possible.
Distributed processing enables many computers to share the load and allows collaboration across countries.

Advantages Simulation Modelling has brought to weather forecasting

Ø Only takes 1 hour to produce a 6 day forecast.

Ø Can predict path of hurricanes, etc.

Ø Can help farmers plan work / Local councils plan / etc.

Disadvantages Simulation Modelling has brought to weather forecasting

Ø 160 million equations to solve – cost of buying a supercomputer. Equipment is extremely expensive.

Ø Long range forecasts cannot be 100% accurate in predictions.

Ø Freak storm / unusual patterns difficult to predict.

CASE STUDY - Algorithm for a Weather Station

A weather station records monthly rainfall figures in millimetres (mm) for a year.

You may be asked to write an algorithm, using pseudo-code or a high level programming language, which will use these twelve monthly rainfall figures as input. The program should output:

• the total rainfall for the year

• the mean monthly rainfall for the year

• the month numbers (1 for January, etc) where the rainfall was above the mean.

Your algorithm should contain meaningful identifiers.

Example solution

Examiner Tip

What the examiners are looking for in your answer: -

Page updated

Report abuse