Airhead Weather-girlies and blood-sucking leeches

For some time past, I have been taking data from a Huger weather station, and importing it into SQL Server. I then use the data to run up graphs, calculate trends and to attempt to predict the weather. It is a wonderful source of real data for trying out techniques.

The curious thing is that the result is not always right. I thought there was something wrong with my logic or the equipment. I checked against the pronouncements of the Meteorological Office. They got it wrong just as badly as I did, if not worse. In fact, their predictions were only marginally better than predicting that today’s weather will be the same as yesterdays.

I became most curious about the failure of the Met Office to predict the weather. It is not because the task is too complex to be amenable to the application of science. For some time, there was an Icelandic website that ran a model on a cluster of PCs, and achieved an uncanny reliability for their predictions of the UK weather, far more accurate than the Met Office with their vast taxpayer-funded computer. Sadly, it closed down. (Sorry, it is now back in operation as a subscription-only service at -pronounced ‘There’ ) Nowadays, I use Spanish and German University websites which are almost as good. The USA runs weather models for Europe which are getting to be reasonably accurate, but which need to be interpreted for the layman.

In the nineteen thirties, there was a spectacularly successful system for predicting the UK weather based on the reports of ‘observers’ based across the country, but the system ceased due to the War. It was able to give farmers an accuracy of the time of rainfall within half an hour throughout the country.

I was left pondering the failings of our weather forecasters I was therefore intrigued when I came across the following article, whilst poring through a Victorian book

‘A leech kept in a phial or bottle, partly filled with water, will indicate approaching change in the weather. Place on a window-ledge an eight-ounce phial containing a leech and about six ounces of water, and watch it daily. When the weather continues serene and beautiful, the leech will lie motionless at the bottom of the phial, rolled in a spiral form. When it begins to rain at noon, or a little before or after, the leech will be found at the top of its lodging, where it remains until the weather becomes settled. When wind approaches, the leech gallops about its limpid habitation with great liveliness, seldom resting until the wind becomes violent. When a thunder-storm is about to appear, the animal seeks a lodgment above the level of the water, displays great uneasiness, and moves about in convulsive-like threads. In clear frost, as in fine summer weather, it will lie constantly at the bottom; whereas in snowy weather, as in rain, it dwells at the very mouth of the phial. The observer should cover the mouth of the phial with a piece of linen, and change the water every week or two.’

My own subsequent experiments have proved that leeches are, in fact very adept at weather forecasting, with accuracy higher than the airhead weather-girlie on the telly. This seems to be because, in streams and rivers, its prey, the bugs and worms, get washed down past their habitat only during rain, and particularly after storms. (in deference to politically-correct kindness to all living creatures, I must point out that my leeches stay in the bottle only for a short while before being returned to the river and being replaced by a blood-sucking colleague.) . In all fairness, I have to say that the leech is mute on the subject of fronts, high-pressure ridges and so on, and the Met office is more accurate in predicting the weather more than twelve hours ahead, but for predicting when you need your brolly or raincoat, the leech wins hands-down.

It therefore occurred to me that a national system of leeches in bottles, with web cams trained on them might provide just the sort of accuracy we all need. Even better, is there any way of providing a digital feed to a SQL Server database? One that can tell the difference between a gallop ‘about its limpid habitation with great liveliness’, and ‘a lodgment above the level of the water, display(ing) great uneasiness’. If so, then onward to a truly twenty-first century way of forecasting the weather, and a new and impressive level of accuracy.