Model Output Statistics (MOS)

• 

• Model Output Statistics, MOS, is a statistical method which improves model forecasts. Current global numerical models have a limited horizontal and vertical resolution. The model grid box (typically 25x25km) represents the average and does not represent the weather at a specific location. Within a grid box you could have open farm land, woods, a city or a coast. All the local effects are averaged in the model which results in errors when compared with the actual weather recorded in observations.

  With the MOS method a large data set of observations is compared with the historical model forecast. The MOS “learns” the differences and adjusts these errors in future forecasts.

Multi Model MOS

• One of MeteoGroup’s recent developments is a Multi Model MOS which combines several models into one forecast. Using this technique the strengths of every model are taken into account to produce a better forecast overall.

  The current MOS produces hourly forecasts for 15 days in the future.

  The map below shows all MOS stations. 

•  

  Below the actual temperature MOS forecast for Wageningen (The Netherlands) is shown. Forecast period is 15 days.

•  

  MeteoGroup has also developed MeteoBase, a tool enabling the meteorologist to make adjustments to the MOS forecasts. With this tool the meteorologist can amend any weather element using specialist knowledge of the location and weather situation.  Adjustments made on MeteoBase will automatically flow through other forecasted weather elements, as shown in the diagram below.