Model output statistics
Model Output Statistics ( MOS ) is a statistical method in modern weather forecasting that was developed in the USA in the 1960s / 1970s. Often these are multilinear regression equations that are applied to numerical weather models . Nowadays, MOS processes are used worldwide and are used as an aid primarily for local weather forecasting.
background
Numerical weather models such as the American Global Forecast System (GFS) or the forecast model of the European Center for Medium-Range Weather Forecasts (EZMW / ECMWF) in Reading calculate the future state of the atmosphere at certain times by determining the actual state on the basis of measurements and probe ascents - Predict the state by solving the equations for later times. This is done for so-called grid points with which the earth is covered like a network. Depending on how far apart these grid points are, the quality of the prognosis also varies, especially for soil parameters: a large distance causes local inaccuracies, a narrow grid offers greater regionality.
The GFS z. B. works with a mesh size of 35 km, the EZMW / ECMWF sets grid points at a distance of 25 km. This is usually sufficient to predict the state of the higher atmosphere very well. The forecast of soil parameters such as B. Maximum temperature, minimum temperature or ground wind for a certain location, however, is too imprecise for everyday use. This is especially true if one considers a spatially strongly varying meteorological variable (lowest temperature in highly structured terrain on a cloudless night).
On the other hand, there are purely statistical weather models , which, however, hardly provide any useful results for a period longer than 6 hours. In order to close this gap, MOS processes were developed.
functionality
MOS methods are multilinear regression equations that are used to verify or change the direct forecast results of numerical weather models ( Direct Model Output, DMO ) for a specific location on the basis of station measurements that must be available for the location. In this way, the advantages of a numerical model are combined with the measured values on site in order to obtain the most accurate possible forecast for the respective location.
By means of regression, those parameters, known as predictors , are filtered out of the DMO that have a significant influence on other meteorological variables, e.g. B. rewrite the ground weather. These parameters are called predictands . In principle, every numerical weather model can be coupled with a MOS method; several, slightly different MOS processes can also be applied to the same model. The prerequisite, however, is that the station for which a local forecast is to be created has a series of measurements lasting at least 18 months.
presentation
The results of a MOS run are usually shown either in diagram form (MOS diagram) , which z. B. is quite common in the USA, or used to generate weather symbols, as is often found in Germany, for example. A program uses the MOS parameters to generate the required and required symbols such as B. cloud / sun / precipitation, wind arrow, precipitation probability etc.
Examples in use
Forecasting of water levels and storm surges
The Federal Maritime and Hydrographic Agency publishes current probable water levels and storm surge warnings in accordance with the Maritime Tasks Act § 1 Paragraph 9 (includes: maritime survey service, tide, water level and storm surge warning service, ice news service, geomagnetic service) certified according to DIN EN ISO 9001 based on calculations according to " MOS process ". The data measured so far, the astronomical prerequisites for the tides and a curve prediction automatically calculated using the MOS method are displayed. In addition, a manually calculated peak value is published in the 24-hour forecast , which is more relevant in case of doubt. The measured values of the unchecked raw data / level values are made available to the Federal Maritime and Hydrographic Agency by the General Directorate for Waterways and Shipping .
Web links
- European Center for Medium-Range Weather Forecasts (ECMWF)
- Model description of the Global Forecast System (GFS)
- MOSMIX system of the German Weather Service
- Example of a typical MOS diagram
- Statistical forecast systems at Meteo Service weather research
- The Front, NOAA's National Weather Service, Volume 6, Number 2, 06/2006 (PDF; 1.7 MB)
- American Meteorological Society: Glossary of meteorology