Ensemble forecasts

Despite numerous weather stations worldwide and modern satellite and radar technology, we know little about the current status of the atmosphere beyond Europe and North America. This is especially disadvantageous for mid-term and long-term forecasts as "our" weather develops within the framework of a global weather cycle and the interchange between atmosphere and the oceans.
We often perceive the weather as a chaotic, dynamic system. A lack of knowledge regarding current atmospheric conditions, but also the underlying mathematical equation contributes to the fact that even minor input failures in the numerical model can lead to increasing forecast inaccuracies. Even this problem, however, can be solved using statistics: by determining the variability/imprecision of the initial situation, extremes can be filtered out and calculated more precisely. As a result, it is possible to make statements regarding weather situation stability and their probabilities.
mminternational can draw on ensemble simulations from the US weather service (GFS) and the European weather service (ECMWF).

Model Output Statistics (MOS)

Weather forecast models are normally divided into two basic systems: there are numerical models that calculate the atmosphere’s dynamics taking into account the laws of physics (partial differential equations), and there are systems that calculate forecasts based on statistical analyses. Numerical forecast models are quickly overburdened due to insufficient computer resources when calculating local weather situations. Statistical MOS methods help make forecasts considerably more precise, e.g. temperature forecasts in small depressions, local wind systems, orographic precipitation or hours of sunshine on slopes.
mminternational has therefore developed its own MOS system, which allows “point forecasting”. It also determines the statistical correlation between the output of a “rough” weather forecast model and the weather station data. This allows precise forecasts for those stations with observation data. These forecasts are calculated every hour for around 17,000 weather stations worldwide. mminternational is therefore able to provide its customers with forecasts of 10 days in advance for almost any location at hourly resolutions.
The unique feature is an expert-level system that analyses the results of different forecasting systems and defines an optimum forecast based on up-to-date station data.

Literature

N. Dorband, M. J. Fengler, A. Gumann, S. Laps:
Modern Techniques for Numerical Weather Prediction – A Picture Drawn from Kyrill.
Handbook of Geomathematics, Springer 2010 (in print)

Nowcasting

In order to detect the entire dynamics of short-term trends, mminternational operates its own nowcasting system, which is incorporates up-to-date measurement data from the station network as well as recent radar and lightning data. Radar images are analysed in real time, thunderstorm cells are extracted and their movements, drift speed and hazard potential are determined. Furthermore, satellite images from different spectral channels can be used to identify the type and degree of cloud cover. This way, thunderstorm cells can be detected at an early stage. The data is automatically and constantly evaluated and made available to meteorologists in the Severe Weather Centers.

Numerical models

Meteorologists draw on all weather data available to do their work. This data includes weather data provided by the high-density weather station network, radar and lightning information and satellite images. It describes the current state of our atmosphere, e.g. temperature, wind speed and the humidity of different air layers.

The air mass itself is subject to the physical laws of thermodynamics and flow dynamics. This is why the development of atmospheric conditions can be described using mathematical equations. Once you have these results, you also have the meteorological trends.
In spite of the fact that the flow equations have been around for almost 200 years now, solving them is complex and time-consuming. Calculating global forecasts several times a day takes several hours even for the fastest computers.

mminternational has access to the results of more than 12 models of different weather services, including the European (ECMWF), British (UKMO/UKNA/UKNX), American (GFS/ETA/NOGAPS), and German (DWD) weather services.
All models differ in the resolution and properties provided. Our meteorologists analyse the relevant models and interpret the weather trends to be expected, drawing on their many years of experience.

These models are also the basis of our MOS systems, which add statistical refinement to global models, which generally have less sophisticated resolution. This refinement allows us to forecast local weather phenomena with point accuracy.

Weather forecasts for the offshore market

Numerical forecast models are quickly overburdened due to insufficient computer resources when calculating local weather situations, especially in the offshore area. Statistical forecast (Model Output Statistics – MOS) models help make forecasts, e.g. wind speed forecasts, considerably more precise.
mminternational has therefore developed its own MOS system, which allows "point forecasting". It also determines the statistical correlation between the output of a "rough" weather forecast model and local weather station data.
mminternational provides these highly accurate forecasts for the FINO 1 research platform near the German island of Borkum in hourly resolution for 10 days in advance.
As part of the RAVE project the forecasts for FINO 1 are available to all lead partners. This proposal is designed to exploit the short time windows for installation and future maintenance of the first German offshore wind farm alpha ventus as effectively as possible. Moreover, it is possible to forecast the generated wind power for 10 days in advance, as it can calculate wind forecast data for up to 100 m above sea level.

Sustainable Energy Management project (SEMS)

The Sustainable Energy Management System (SEMS) project is part of the 6th European Union (EU) Framework Programme. The project, which was launched by 24 partners, aims to provide an active role model by realising practical measures in established model communities in Luxembourg, Germany, Austria and Poland in the fields of energy efficiency, energy conservation and in particular the integration of renewable energies into the regional (electrical) energy supply. The aim is to increase the proportion of renewable energy and reduce emissions of climate-damaging gases significantly in these regions in the medium term.

Cooperation between IZES and mminternational

• mminternational is the weather data provider and meteorological contact partner for IZES GmbH.
• Within the framework of the SEMS project they are responsible for the field of networking, modelling/simulation and control of regional energy producers via SIEMENS' Decentralized Energy Management System (DEMS) tool.

The Decentralised Energy Management System (DEMS) for renewable energies: forecasting ability and deployment planning

• The basis for the Decentralised Energy Management System (DEMS) is forecasting ability and deployment planning.
• These criteria are determined by the weather as a major factor influencing the supply of electricity from renewable energy sources.
• Total energy generation and load is forecast on this basis.

You can find further information at: www.sems-project.eu