The aim of weather forecasting is to forecast the state of the atmosphere at a specific time at a specific location or in a specific area. In fact, not only weather phenomena that affect the ground are meant, but the entire earth's atmosphere is considered.
As a physical event, the weather can be described by corresponding natural laws. The basic idea of a weather forecast is to derive a future state from a past and the current state of the atmosphere, using the known physical rules.
The mathematical constructs that describe these physical rules are, however, so-called non - linear equations. This means that even small changes in the initial state can lead to relatively large changes in the result of the calculation ( see also: butterfly effect ).
A distinction is essentially made between manual or synoptic weather forecast and numerical weather forecast , although a combination of both methods is still used today. This is related to the fact that current numerical forecast models also deliver inadequate results. In order to take into account the local climatology of weather stations, nowadays the numerical calculations are still followed by statistical methods, such as B. the MOS method Model Output Statistics .
The data on the current state of the atmosphere come from a network of ground measuring stations that measure wind speed , temperature , air pressure and humidity as well as the amount of precipitation. In addition, data from radiosondes , weather satellites , commercial aircraft and weather ships are used. The problem is the irregular distribution of these observations and measurements, as well as the fact that there are relatively few measuring stations in less developed countries and over the oceans.
The reference time for a day in the weather report (Mon, Tue, Wed, ...) is usually 23:51 UTC of the previous day to 23:50 UTC. Since the minimum temperatures of a day are to be measured around sunrise, the daily minimum in the weather report refers to this time and is therefore sometimes also called the early value . An example: The night from Saturday to Sunday should be particularly cold, with the lowest temperature just before sunrise. This temperature can be found in the weather report as the daily minimum for Sunday.
Prediction symbol: " covered "
Forecast Symbol: " rain "
Forecast symbol: " Rain "
Forecast symbol: " Thunderstorm "
Prediction symbol: " Snowfall "
Today a forecast for the coming week is about as reliable as it was thirty years ago for the next day. The 24-hour forecast achieves an accuracy of over 90%. The accuracy for the next 3 days is a little more than 75%.
However, the reliability varies greatly depending on the weather. With a stable winter high pressure situation it is sometimes possible to forecast a week with 90% certainty without any problems. In contrast, the forecast quality is often well below 70% for 24 hours with an unstable thunderstorm in summer. A distinction must also be made between temperature and precipitation in the forecast quality. Temperatures can be forecast much more precisely than precipitation.
The quality of the DWD model forecasts has risen steadily since 1968. New, more powerful computers and improved weather models as well as improved satellite data often led to a sharp increase in accuracy towards the optimum (1.0), as shown here using the example of ground pressure forecasts. From 1978 onwards, four forecast days could be calculated instead of two, and from 1991 even seven days. In 2008, a seven-day forecast was better than the two-day forecast at the beginning of the computer age in 1968. A very good overview of the historical and current accuracy of severe weather warnings and weather forecasts can be found in the brochure How good are weather forecasts? of the German Weather Service. With a hit rate of over 90 percent, the German Weather Service today predicts the temperature of the following day, 30 years ago this value was only a good 70 percent; In the case of wind speed, the information is correct in more than 95 percent, in the case of the amount of precipitation in over 80 percent of cases.
Reasons for the lack of reliability
Sometimes the weather forecast is not as reliable as the general public and various disciplines would like. This is mainly due to two causes:
- the incomplete knowledge of what is actually happening in the earth's atmosphere .
- What exactly is happening in the atmosphere? → Basic research. Weather phenomena that are not yet understood must be researched. Example: How do soil properties (heat, moisture content, albedo ) interact with the deeper layers of air?
- Where and when does something happen in the atmosphere? → Not all required data is collected, and where it is collected, there are inevitably gaps. Examples: Weather stations are not located everywhere, mountain valleys are not always covered by weather radars, the number of weather balloons is limited due to the costs in terms of both spatial and temporal coverage.
- How does something happen in the atmosphere? → The actual meteorological relationships must be translated into a sufficiently precise computational model. These reflect the respective advances in basic research as well as the computing capacity of the computer (the computing grid currently typically has a mesh size of 1 km - 20 km). The mean effects of the processes that take place on scales smaller than the calculation grid width (e.g. melting of a hailstone) must be described by means of simplifying approximations (so-called parameterizations).
- the unpredictable part of the weather → chaos research , butterfly effect . Under what circumstances does a small inaccuracy in the weather model or in the measured values result in a very inaccurate forecast?
For reasons of computing time and the large amounts of data that arise, the air and water masses involved cannot yet be taken into account with satisfactory accuracy. Too many individual factors play a role, the interplay of which cannot be fully analyzed to date or in the near future. Therefore, local influences such as mountains and their irregularly shaped slopes, the effects of different irradiation due to "incorrectly" calculated cloud cover , the vegetation (forest to field!) Or the rock make so much that the accuracy for the next 4 to 7 days is significantly reduced. None of the methods used today extends beyond twenty days into the future, so longer predictions e.g. For example, seasonal weather forecasts are dubious and are avoided by most weather reports on television and radio.
From the existing data models, very reliable forecasts for weather-dependent industries can be created for different forecast areas. The German Weather Service has created a forecast module for the grain moisture in fully ripe grain. This model predicts the grain moisture of grain with high accuracy based on soil moisture, air humidity and solar radiation. The forecast time extends 5 days into the future. The forecasts are only made during the summer grain harvest.
The theory of meteorology has been largely clarified by the gas laws , thermodynamics and fluid mechanics , but due to small-scale effects of up to kilometers in size, it cannot calculate all air movements with sufficient accuracy. For example, on a sunny day, the temperature over dark and light surfaces can differ by several degrees. Something similar occurs between the sunny side and the shadow side of a mountain ridge or between bodies of water and solid ground.
Weather forecasting apps are very popular and they are preinstalled on many smartphones (e.g. "Weather" on iPhones ). The apps often get their forecasts from the publicly accessible and free GFS model of the US American NOAA . This has the advantage for the user that forecasts are available for the entire world, and for the provider that no integration with regional weather forecasts is necessary in terms of software. However, GFS works with a 28 km × 28 km grid, which makes it impossible, for example, to depict mountain ranges precisely - they appear as plateaus in the meteorological model. The weather forecast for Sion, for example - 515 m above sea level, located in the middle of the Alps and one of the warmest places in Switzerland - is then based on an adaptation to previous experience and not directly from the weather model.
It is also criticized that weather apps rarely provide information about their limits. For example, long-term forecasts are often presented in the same way as the much more reliable short-term forecasts. There are also often no indications of how often and at what times the forecasts are updated.
Attempts to predict the weather have been handed down since ancient times and should go back even longer if you consider how much people - especially in agriculture - were dependent on precipitation and temperature.
So-called lost days , known as the farmer's rule, are to be seen as an attempt to further subdivide the times in between into weather-relevant sections analogous to the seasons that seem to run the same way over and over again. It was assumed that on lost days - similar to a node in a decision tree - the weather and the weather, depending on the state on that day, would take a certain further course, which could be determined from tradition and later records. The scientifically based weather forecast began with similar methods: in the time before the telegraphy one tried to recognize short-term patterns in the weather and, for example, to predict tomorrow's weather from the precipitation, the temperature and the air pressure of the last three days.
Otto von Guericke first recognized the connection between a drop in air pressure and the lift of a storm in 1660 . Simple hygrometers were also known earlier - such as the " weather thistle ", whose rolled up petals indicate increased humidity and thus warned of rain.
A European network of stations with simultaneous observations for the synoptic method emerged soon after 1800. A lack of communication options prevented their triumphant advance: riders and stagecoaches were namely too slow at 60 to 120 kilometers per day to transport the measurement data to a distant city in a useful time - the speed of weather fronts is about 30 to 60 km / h.
The first decisive change in weather forecasting came about with the widespread expansion of the telegraph network in the 19th century. Commercial telegraphy came into being in 1835, nationwide networks in European countries existed around 1855, and the first east-west telegraph connection in the USA came about in 1861. Thanks to telegrams, the measurement data could be “sent ahead” of the weather, depending on the wind direction, which made the first useful weather forecasts possible: The London Times published the first weather forecasts in 1861, and the first weather maps were published in the same year.
Two Britons, Francis Beaufort and Robert FitzRoy , are closely linked to improving the weather forecast. Beaufort developed the wind scale named after him , and FitzRoy headed a department on the Board of Trade that collected weather reports from ship captains. In 1859 the accident happened to the ship Royal Charter , which capsized in a heavy storm. This prompted FitzRoy to draw weather maps, and fifteen weather stations on land provided data for storm warnings, which were then telegraphed to the ports. In the industrialization, during the 1870s, the telegraph network was finally expanded to such an extent that synoptic weather forecasts were made possible. In addition to telegraphy, standardized terms for weather conditions and cloud types had to be created (first classification by Luke Howard in 1802 , in 1896 the World Meteorological Organization published the International Cloud Atlas as the first standard work). Around 1900, many national weather services came into being, which in cooperation developed a large-scale synoptic weather forecast. In 1911, the British “Met Office” broadcast the first storm warnings to seafarers over the radio, and the North Atlantic Ice Warning Service came into being as a result of the Titanic disaster in 1912.
Since current measured values are essential for weather forecasts, automatic weather stations were developed early on in order to receive weather data from remote areas. In late 1943, for example, the German Navy built a battery-operated weather station in Canada that was supposed to send readings every three hours for six months.
The future weather is calculated
The English scientist Lewis Fry Richardson came up with the first idea for weather forecasting using “computers” in 1922. (In the past, " computer " was the name for a person who did calculations, for example in land surveying .) He imagined that the entire atmosphere of the earth could be divided into departments; the boundaries between these departments are defined by the height above sea level as well as by the latitude and longitude. In each of these departments there is a person who applies certain linear equations. He calculates forecast values for his department from the physical condition of his own and neighboring departments (wind speed, wind direction, air humidity, temperature ...) and forwards the results to his neighboring departments so that they can carry out the next calculation step. This basic principle - the subdivision of the atmosphere into compartments, which are initialized at the beginning of the calculation with actual, measured starting values - is still used today.
Today weather forecasting is unthinkable without powerful computers, and in the 1950s and 1960s meteorology was the driving force behind the construction of the first supercomputers . From the point of view of computer technology, it is interesting that the identical calculation steps have to be carried out countless times for the weather forecast - namely once for each of the compartments. Parallel computers that perform operations simultaneously (in parallel) instead of one after the other (in series) are therefore particularly suitable for weather forecasting. But since the speed of the individual computer processors has hardly been increased since the 2010s, the most powerful computers in the world have thousands of processors or processor cores that compute the same tasks at the same time.
The importance of supercomputers and an adequate database was underlined by Hurricane Sandy (2012). While the American National Weather Service ran various forecast models on one computer, the European ECMWF had a more powerful computer center that only worked with one model. While the ECMWF was able to predict seven days in advance - and with a low probability of error - that the hurricane would hit the American mainland, the NWS model considered this to be unlikely. Thanks to the better ECMWF model, the disaster warning could be issued much earlier - but later calculations have shown that the ECMWF model would have come to the same result as the NWS without the data from the polar weather satellites.
More and better data
Since the 1950s, the data basis for computational weather forecasts has also been improved, for example by means of weather satellites , a network of radio weather probes and weather radars. The relatively reliable forecast period in mid- latitudes rose from around 3 days to 4–5 days, which meant a noticeable improvement for many sectors of the economy , in transport or in construction, as well as for planning in agriculture .
The automation and networking of the weather stations also brought great progress. While weather stations previously had to be operated and maintained by specialists, modern weather stations automatically send the data to the weather services. Weather stations with their own power supply ( solar cells and batteries ) and a satellite modem can be set up and operated for years even in the most remote areas.
Recent developments in data science make it possible, for example, to recognize patterns in the measured values themselves and, for example, to find out which measuring stations deliver the most important or most reliable values under which circumstances. In this way, useful knowledge can also be gained with private, hardly standardized weather stations.
Overview of the developments
Advances in weather forecasting have been made since the 1950s - without setting a chronological order - by:
- Measuring stations
- Stations have become more numerous, a denser network
- Measurement of additional parameters such as smog particles (which serve as condensation nuclei for precipitation)
- automatic transmission of the measured values
- today also with GPS
- Weather radars
- for the short term forecast
- to supplement the rainfall measurements of the weather stations
- Ability to distinguish between different types and intensities of precipitation
- improved simulation models such as B. the non-hydrostatic mesoscale model as part of the Weather Research and Forecasting Model
- specialized forecasts and weather models
- Remote sensing by satellite
- not only the visible light, but also the infrared spectrum is now evaluated
- Vegetation analysis - the different vegetation influences the warming of the soil and the evaporation of water, both factors which in turn influence the weather
- improved topographic models ( SRTM ) to better model wind and incline rain
- more powerful computers
- for more model runs within the same time: Do the forecasts match, even if the starting values are changed slightly? (→ Ensemble models)
- enables a reduction of the grid in the calculation models in order to increase the reliability of the models
- Examples: Global System Forecasting the NOAA : km 28 km world × 28 or 70 km x 70 km, 3 km x 3 km Europe on certain models of the company meteoblue, COSMO-DE of the German Weather Service 2.8 km x 2.8 km with 50 layers of altitude, MeteoSwiss COSMO-1 with 1.1 km × 1.1 km, AROME of ZAMG in Austria 2.5 × 2.5 km
- (Online) weather services
- Local weather forecasts, e.g. B. by adapting a large-scale forecast to local conditions using empirical values
- Emergence of private weather services; Competitive struggles
- Changed leisure behavior of the population (outdoor sports, for example) requires more precise forecasts: People want to know on Monday whether the weekend weather is good
Outlook into the future:
- The reduction in size of the grid, due to more powerful computers, allows even more precise and even more long-term forecasts.
- Today's grid sizes (4 km × 4 km) are often larger than typical thunderstorm cells. With smaller grids (e.g. 1 km × 1 km), small-scale phenomena such as the convection of air masses can be modeled at all, and thus the location and time of thunderstorms, the same can be said for tornadoes , for which no useful forecast models have yet been found exist.
- Even better understanding of meteorological effects, especially the interaction of the ground with the lowest layers of the atmosphere.
Weather forecast in the media
On German television:
- With weather maps on television and in newspapers, it is common to graphically highlight the transmission or distribution area. For example, the weather map in First Germany is set off in color from the neighboring countries; The regional border of North Rhine-Westphalia can be seen in the maps of the regional broadcasts of the WDR .
- With the two major television broadcasters (Das Erste and ZDF ), the graphic highlighting of Germany has only been customary since reunification in 1990. Before that, a Central Europe map without borders was used in both the Tagesschau and today .
- Despite the division of Germany, the weather reports from Das Erste and ZDF predicted the weather for both German states, which was most conspicuously expressed by an all-German use of the weather symbols. While some major West German cities and Berlin were shown on the weather map of the first , Leipzig was also to be found on the ZDF map . The term GDR was not used, instead it was called "in the east", analogous to "in the north" or "in the south-west" for regions in what was then the Federal Republic.
- Until the 1960s, the weather map of the Tagesschau Germany was shown within the boundaries of 1937 .
- After the parliamentary decision of the reunification, the Tagesschau first introduced a weather map that showed the clearly defined area of the reunified Germany.
- On March 30, 2008, the Germany weather map for the 20 o'clock news coverage contained the following cities: Kiel, Hamburg, Hanover, Cologne, Frankfurt, Stuttgart, Munich, Rostock, Berlin and Dresden. Instead of Hanover, Cologne and Frankfurt on the 19 o'clock today map , however, it was Kassel, Bonn and Saarbrücken (in other today editions a more detailed map with further cities is shown).
Austria is traversed lengthways by the main Alpine ridge, which forms a clear weather divide. Weather phenomena often only occur regionally, which was or is often illustrated on an outline map of Austria including the federal state borders by a presenter with hand gestures sweeping over the areas or by pointing a finger, for example by Bernhard Kletter . The data is collected by the state weather services ZAMG or Austrocontrol (aviation weather) and forecast calculations (AROME model chain and INCA nowcasting), weather advice and warnings are carried out. There are also some private weather services such as B. UBIMET with its own measuring points as well as other authorities (environmental authorities, hydrological services, road maintenance service, avalanche warning services that carry out meteorological measurements). There are 5 weather radars (Vienna Airport, Salzburg, Innsbruck, Zirbitzkogel, Valluga) and 4 weather balloon launch sites (Vienna, Graz, Linz, Innsbruck) as well as several hundred largely automated ground stations. The state broadcaster ORF has its own weather department for forecasting (similar to ZDF in Germany), which also includes experienced meteorologists.
In the early years, around 1970, still in black and white, a screen less than 1 m wide was set into the rear wall of the studio, onto which weather symbols and temperature values were projected brightly from behind. The emerging color television cameras demanded greater illuminance, which could only be achieved by front projection onto a retroreflective screen, which, however, creates shadows on the contours of the moderator standing in the projection beam. With digital image processing it was later possible to combine two moving images live. The presenter seems to be standing in front of a wall with a "weather map" that fills the width of the television screen and is looking, while he stands left and then right in the picture to illustrate areas in the east or west of the country with gestures or to point to symbols that have been discussed. In fact, the studio back wall is one color green or blue to the weather map, including moving färbigen elements is there to be transmitted overall image into attribute where the presenter does not cover the single-colored background. The moderator stands to the side (once to the left and once to the right) from the center of the map (of the overall picture) and appears to be looking at the map on the nearby studio wall while explaining the weather with hand gestures. In fact, however, he is looking at a control monitor (of two) with the overall image slightly outside the camera view. Training of the moderator is necessary so that he can manage the gestures quickly and purposefully and also the sufficient turning of the face from the camera point of view so that the glancing view of the eye does not reveal where the actor is looking really focused and the magic succeeds.
After the main news on SRF , the broadcast SRF Meteo will be presented live and in almost any weather from the roof of the Leutschenbach TV studio. If diagrams are shown, the moderator is not shown; For the sake of simplicity, the recorded and computer-generated images are separated. The studio has a shallow tub filled with water, by means of which the current wind and precipitation can be recognized.
- Precipitations probability
- Weather forecast - weather forecast over a longer period of time
- Meteorological expressions
- NAVTEX - global system for disseminating safety and weather information for seafaring
- Temperature extremes
- Hans von Storch: The climate system and its modeling: an introduction Springer Verlag, 1999, ISBN 978-3-540-65830-6 . (Chapter 6.1 Weather forecast models)
- Utecht Burkhard: Weather and Climate Teubner BG, 2005, ISBN 978-3-519-10208-3
- European Center for Medium-Term Weather Forecasts (ECMWF), cooperation with weather services from 26 European countries
- Forecast quality Tabular view of the forecast quality of weather forecasts
Weather forecasts from German-speaking national weather services:
- German Weather Service : Explanation of the forecast
- MeteoSwiss (SMA): www.meteoschweiz.ch
- Central Institute for Meteorology and Geodynamics (ZAMG), Austria: www.zamg.ac.at
- Information on the daily and monthly values on the website of the German Weather Service
- Weather report for Berlin , wetteronline.de
- dwd.de (PDF), accessed on May 4, 2016.
- Weather forecasts . In: Der Spiegel . No. 38 , 2016 ( online ).
- Why weather apps are often wrong. January 17, 2017, accessed May 6, 2018 .
- arstechnica.com. Retrieved March 6, 2017
- washingtonpost.com, accessed March 6, 2017
- noaa.gov , accessed July 8, 2014.
- meteoblue.com , accessed on July 8, 2014.
- dwd.de , accessed on July 8, 2014.
- meteoschweiz.admin.ch ( Memento of the original from July 14, 2014 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice. Retrieved July 8, 2014.
- COSMO-1 - high-resolution forecasts for the Alpine region - MeteoSwiss. Retrieved February 2, 2019 .
- AROME - ZAMG. Retrieved February 2, 2019 .
- INCA - ZAMG. Retrieved February 2, 2019 .
- Austro Control GmbH - Weather radar. Retrieved February 2, 2019 .
- probe Hunting Weather probes in Western Europe: locations for launching weather balloons in Austria. Retrieved February 2, 2019 .
- Weather stations - ZAMG. Retrieved February 2, 2019 .