MOS-Forecast interview with Eric Terpstra, Head of the R&D team
What is the MOS method?
MOS stands for Model Output Statistics. It is a statistical post-processing method that has been used in modern weather forecasting since the end of the 1990s. To put it simply: in order to obtain the most accurate forecast possible for the respective location, forecast results of numerical weather models (DMO = Direct Model Output) are combined with current and historical measured values on site as well as satellite and radar images.
The MOS technique starts where the large model calculations stop. With the help of statistics, possibilities are sought to compensate for the known weaknesses of the models in the best possible way in the respective weather situation. A colleague in the field has compared it very vividly with the horse and rider principle: The model calculation gallops off, the statistics sit firmly in the saddle, drive and rein in where necessary.
What added value does the MOS method offer?
With the help of the MOS method, weather elements and derived variables can be predicted with pinpoint accuracy. Depending on the application and business area, very specific weather parameters are important in each case. In addition, the demand for granular weather data is constantly increasing. Currently, Weather Solutions has developed and deployed proprietary forecast models for road surfaces, for the energy industry and for agriculture.
For the operational planning of winter road maintenance, these include the condition and temperature of various road surfaces such as roads, cycle paths and footpaths, bridges, but also aircraft wings, rails and overhead lines of railways as a function of air temperature and humidity. It is not only possible to predict if but also exactly where and when it will become slippery.
For operators of fully automated greenhouses, information on wind speed, air temperature, cloud cover and radiation is enormously important in order to be able to optimally control the climate within the respective production facility, its energy consumption and the irrigation of the plants.
Knowledge of hyperlocal weather effects is also a decisive competitive advantage when integrating renewable energies into electricity grids and markets. For operators of onshore and offshore wind farms, forecasts of wind speed and direction for different hub heights at 10, 20, 30, ... 120, 140 and 200 m are therefore crucial. Solar parks, on the other hand, require specific information on direct and indirect radiation as well as on radiation during the day and at night.
How do you develop a MOS?
At Weather Solutions, an interdisciplinary team is active in the area of research and development. Meteorologists, mathematicians and software developers are part of the core team that works on the various MOS projects. For the meteorological team, which is active in weather forecasting, MOS is one of the main working tools. They put it to the test, so to speak, on a daily basis and give the development team regular feedback on what is already working very well and what should be improved or adapted if necessary.
Basically, we are able to make precise forecasts in hourly resolution up to 15 days in advance for Europe. In other words: If it is necessary for customers to know what wind speed to expect in 10 days at 3 pm at location X, then we provide this information.
Depending on the area of application, we offer our customers short update cycles (10-minutely, 1-hourly and 3-hourly) and rapid model updates several times a day when new data arrive. If required, we provide the weather parameters in an even higher granularity.