background
Before the introduction of the first computer-based weather models, it was hardly possible for meteorologists to make satisfactory forecasts for a longer period of time.
At that time, meteorologists only had access to weather observations from the current and previous days and weather maps created on their basis. The first regularly produced and published weather maps date from around 1860. This meant that at best the current state of the atmosphere in a limited area (such as parts of Europe) was reasonably known. In the beginning, data collection was mostly limited to the atmosphere near the ground. Only over the years did more and more information become available from greater heights and from larger areas. Data from higher altitudes became available through regular ascents of weather balloons. This knowledge proved to be very important for creating weather forecasts. Using sophisticated methods (based partly on physical laws, partly on empirical values), it was actually possible to create weather maps for the future by hand. However, these methods were rarely able to provide forecasts for more than 1 to 2 days.
This changed with the development of so-called supercomputers, i.e. high-performance mainframe computers, in the 1970s. With their help, the equations of fluid physics and thermodynamics, which describe weather processes precisely, could be approximately solved for future points in time within a reasonable time frame (numerical weather forecast). In order to be able to apply the equations in this way, the current state of the atmosphere must be fed into the weather model used for this purpose as accurately as possible and, ideally, with only small errors. Global data collection is essential for a multi-day forecast. In addition to measurements from ground weather stations or weather balloons, data from commercial aircraft and weather satellites also supplement the information available today. At the same time, the results of previous model calculations are incorporated into the initial state to be determined. This is known as data assimilation.
Today, weather models provide meteorologists with a huge range of weather maps. This gives them a practically unlimited selection of different parameters that describe the current and future state of the atmosphere. At the same time, their temporal and spatial detail leaves little to be desired. The question that always arises is: is all of this correct? Because of course the model forecasts are not free of errors. The extent of the deviation from reality varies from weather situation to weather situation and generally increases from forecast day to forecast day.
Nevertheless, with the increase in computing power of mainframe computers and the resulting ability to depict physics ever better, remarkable progress has actually been made in recent decades. Accurate model forecasts for ever longer periods of time have become possible. As a study by the British Meteorological Office into the forecasting of air pressure has shown, the reliability of the forecast for the fourth to fifth day of the forecast is now as high as it was 40 years ago for the first day after the start of the calculation. However, the chaotic nature of the atmosphere always puts a natural end to visions of a longer-term weather forecast – beyond 1 to 2 weeks.
Wednesday, June 26, 2024 (Release date)