Let's start this article assuming we're planning on going for a walk this weekend. Today, Monday, we will see the weather forecast for the weekend and this could not be better! A radiant sun and warmth in perspective, an ideal setting for a beautiful walk! After 24 hours, don't let the devil weave them, let's check the weather forecast again. Now, today is already Tuesday and what is not our astonishment to see that after all the forecast points to cool weather and rain on the weekend! How does the forecast change so drastically from one day to the next?
WHERE DO THE METEOROLOGICAL FORECASTS COME?
Have you ever experienced a similar situation described above? I'm sure he will. And what's the reason for that? First of all, it is important to understand how weather forecasts are made. These days, forecasts are made using weather models. Meteorological models are computer programs composed of numerous equations that make calculations in order to predict the weather for the next hours, days, even months. The meteorological data collected by thousands of meteorological stations, airplanes, ships, soundings, satellites, among others, are prepared and placed in the meteorological models that, from these data, will solve the multiple equations, simulating the evolution of the atmosphere in the near future. An efficient meteorological model will be one that has the most meteorological data collected, the one that has the best pre-processing of this data and a better set of functions (equations) that, from the initial data, will simulate the state of the entire atmosphere (or only a part) in the future. Due to the high complexity of this whole process, weather model programs run on supercomputers. For example, the European Centre of Medium-Range Weather Forecasts (ECMWF) model has 2 supercomputers, each with a capacity of 10 petabytes and a bandwidth of 350 gigabytes per second.
THE TECHNOLOGICAL PROBLEM
But if technology today is so evolved, why do predictions differ so much from one day to the next? The answer is simple: models do not have, and will never have, the ability to simulate the atmosphere as a whole. There is no capacity to collect the initial meteorological data from all over the world, from all over the atmosphere, and even the data that is collected can have imperfections, can be wrong. Do you know Chaos Theory? She says that a tiny change at the start of an event could have huge and absolutely unknown consequences in the future. So, by making the analogy with the weather forecast, just a thermometer has a deviation of 0.1oC that this difference can result in a totally different forecast than you would expect.
A PRACTICAL EXAMPLE
In the following images, we have the forecast for March 5th at 7pm. In the first image, generated on February 26, we can verify that it will rain practically throughout the territory of mainland Portugal, but this rain will not be very strong. In the second image, we have the forecast generated the next day, that is, on February 27. Here we can see that the forecast has changed: it no longer rains in the Algarve and the rain forecast for the North area will be much more intense than it was the day before (in the previous image). We can also verify that in the first image snowfall was predicted (marked by diagonal strokes) for many locations in Spain, but in the second image the predicted snow fell drastically. This is a good example of how from one day to the next the predictions can change.
PURSUANT TO CONCLUSION
In short, weather forecasting is very complicated to calculate, it's a very volatile process. First, because it is impossible to obtain data on the entire atmosphere, either at surface level or at altitude. Second, because the collected data with which the model will be processed, may be wrong. Third, because the equations that make up the model may not be the most correct, because it is extremely complicated to translate into equations all the processes that influence meteorology.
English