Math IA – Mathematics of predicting the weather?
Updated: Mar 20
I always feel that the Singapore Metrological agency’s notification on Channel News Asia is always somewhat of an “anti-indicator”.
As recent as the Chinese New Year period, there didn’t seem to be any warnings of the wet and cold weather that lasted for quite a few days and when a notice that the weather was expected to carry on was broadcasted…viola! Bright sunshine again!
It got me thinking how those guys analyze the weather so I just want to share two simple methods that even a secondary school student can understand.
Method 1 – Extrapolate based on weather pattern from last few days.
For example, if last few days have been hot and dry – claim that the heatwave is expected to carry on. If past few days have been wet and windy, proclaim that the wet weather is expected to carry on for couple weeks.
Clearly, extrapolation is not reliable (past is not indicative of the future) and weather is easily affected by extraneous factors.
Method 2 – Statistical pattern
Simply provide the prediction based on past years data. It is a form of extrapolation too, but the time horizons are different (days in method 1, years here). Of course an inherent problem is that the Gregorian calendar is not based on weather patterns and hence is not a very good baseline tool. But quite simply if Singapore has always experienced persistent rainfall from Oct-Dec, make that prediction year on year then.
So of course, if the observations from both methods are in sync, then it increases the confidence level.
Both methods I described above are somewhat primitive and most students will have used before, it’s important to know the pros and cons of any methods one is using so you can employ a combination of techniques.
Disclaimer: I do not actually know the methods employed by the Singapore metrological agency, but it may be hard to predict the weather here due to our small land size, which makes conditions change quickly.
Forecasting is both an Art and a Science, unfortunately I don’t know enough Geography here to say more though I’m familiar with most forecasting method out there.
I read here that the Met Service's accuracy of weather forecasts range between 75 per cent for 24-hour forecasts and 90 per cent for short-term two-hour "nowcasts".
My goodness, it’s like I have a 90% forecast accuracy of football matches, just ask me for my forecast 85 minutes into a football game though.