Early days

The art of weather forecasting began with early civilizations using reoccurring astronomical and meteorological events to help monitor seasonal changes in the weather. Around 650 B.C., the Babylonians tried to predict short-term weather changes based on the appearance of clouds and optical phenomena such as haloes. By 300 B.C., Chinese astronomers had developed a calendar that divided the year into 24 festivals, each festival associated with a different type of weather. Around 340 B.C., the Greek philosopher Aristotle wrote Meteorologica, a philosophical treatise that included theories about the formation of rain, clouds, hail, wind, thunder, lightning, and hurricanes. Aristotle made some remark-ably acute observations concerning the weather, along with some significant errors, and his four-volume text was considered by many to be the authority on weather theory for almost 2000 years.

Even today, in many rural communities, people can assess the weather forecasting by simply looking out their window, sensing the movement of the clouds, paying attention to the changes in wind blow, and based on their own experience predict how it is going to change.

 

Throughout the centuries, attempts have been made to produce forecasts based on weather lore and personal observations. However, by the end of the Renaissance, it had become increasingly evident that the speculations of the natural philosophers were inadequate and that greater knowledge was necessary for further understanding of the atmosphere.

 

Statistic forecasting

Since the 17th century, with the development of transportation and telecommunication, people started developing statistical assumptions towards forecasting, assuming for example that if a weather system is in Italy and on the next day it appears in Greece, and it reoccur, it is high likely that it would repeat itself; and so local weather assumptions started to evolve.

 

Subjective forecasting

But identifying general weather behavior in the large scale couldn’t help the individual farmer, who would like to know whether it is going to rain in his field or not.

 

So, in the early 20th, people become aware of the term forecaster, and the responsibility of the forecaster became to be to try and translate the general weather behavior to a local weather forecast projection.

 

Numerical forecasting

It was not until the early 20th century, with the work of Lewis Fry Richardson that people started developing Numerical models to predict the weather.

Numerical forecasting uses mathematical models of the atmosphere and oceans to predict the weather based on current weather conditions. But it was not until the development of computer modeling that numerical forecasting produced realistic and usable predictions.

Actually, numerical models were one of the most significant drivers for the development of super computers in the world.

The problem with numerical models is that they assume that the physics are the same everywhere, which is not necessarily true. Though the rules of physics are the same everywhere, the relative weight of the various physical factors differ according to the local geography. In addition, the chaotic nature of the atmosphere, and the limited accuracy of computational calculation, limit the range and the accuracy of the predictions for the local scale.

 

The future

So how would an ultra light pilot know what wind to expect on his flight?

 

Today in a time where we have a huge computational power and a lot of experience in predicting the weather, combining between the three forecasting methodologies would yield the best forecast.

Taking the numerical models as a base, down-scaling it using statistical methods and then eliminating abnormal behaviors using subjective method.

 

The advantage of combining the three methodologies is that it allows us to localize the weather forecast based on the numerical models to the specific point, using the statistics that reflects local factors.

 

Meteo-Logic is doing exactly that, getting familiar with a local geographical location using the measured history of the atmospheric variables, and using this familiarity to adjust the numerical model and achieve the best forecasting result for that specific location.

 

This article was written together with Dr. Baruch Ziv, a senior forecaster and lecturer at the Open University and Tel Aviv University for three decades, teaching synoptic meteorology and applied meteorology, including air pollution & agro-meteorology.