WEPROG - Weather & Energy PROGnoses - was established in early 2003 in Germany and in 2005 in Denmark with the aim of developing a real-time ensemble prediction system for the electricity market and those companies that deal with weather sensitive goods or services.
WEPROG's founders and directors Jess U. Jørgensen and Dr. Corinna Möhrlen have developed an Ensemble Prediction System based on a multi-scheme approach (MSEPS) in order to provide a solution to a problem that has been researched extensively over the past two decades, both in meteorology, wind engineering and Renewables in general.
The multi-scheme ensemble approach provides information about a crucial parameter: physical uncertainty of the weather development.
In WEPROG we value security and the continued development of our system with the help of newest technology advances. For this reason, our operational processes are built upon highest IT standards with ISO IEC27001 certified process chains, whether it be at our own hardware clusters, leased hardware or cloud based resources for highest scalability. Maximum redundancy and security is key for us in our operations, where we continuously produce forecasts on 365 days a year. Real-time weather products are generated four times per day at 00,06,12,18 UTC and up to 7 days ahead for all continents.
WEPROG operates on a 24/7 basis, where forecasting of weather related derivates such as wind power and solar power take place in the short-term with schedules down to 10min. WEPROG's inverted Ensemble Kalman Filter technique provides the basis of such forecasting products that require data assimilation with customer specific measurements.
In the following we provide an overview of our developments from when we started in 2003 and show how we continuously strive to improve our forecasting system, develop new products that help enhancing the use and deployment of Renewables as well as make use of newest technology in the area of high-performance computing.
A BRIEF HISTORY OF WEPROG
WEPROG was established in Germany in the beginning of 2003 with it's main objectives being to transfer an approach that was until then only used in experimental research into a fully operational prediction system. Funded by the Danish electrical system operator ELTRA (now Energinet.dk), the operational Multi-Scheme Ensemble Prediction System (MSEPS) has been developed over 2 years by a team of scientists with experience in meteorology, engineering and super-computing from the past 20 years.
The first demonstration system covering Europe and the Atlantic started in operation in summer 2003 after an extensive programming period. One year later WEPROG started to produce forecasts focused over Denmark 4 times per day in three different horizontal resolutions (45km, 15km and 5km), as 75 forecasts of the expected wind power load for the next 72 hours.
In 2005 WEPROG began generating weather forecasts for all continents in limited (60km) resolution. This has ensured that the model system is capable of coping with all types of weather and that the service is available internationally. In spring 2005 WEPROG ApS (Ltd) was established in Denmark in order to coordinate real-time services and to expand the services internationally by establishing extended computing resources in Denmark.
In 2006 the success of the global forecasts lead to the introduction of hemispheric boundary generating models for the northern hemisphere as well as the southern hemisphere, which enabled WEPROG to producing 6-day forecasts every 6 hours (since July 2006).
Additionally, the increased international activities suggested that it was time to add redundancy to the operational hardware and management of the operational services. Mid 2006 the establishment of a a third operational system in a professionally operated hosting centre in Berlin, Germany took place.
In 2007 short-term forecasting and ramp rate predictions came into operation as well as preparations for a hardware upgrade at the end of that year. The new hardware, inclusive storage was placed in a second professionally operated hosting centre at a strategic network point in Frankfurt, Germany. This strategic move delivered a higher bandwidth to international customers and also enabled the setting up of leased lines and VPN connections to WEPROG's customers. Technically this new system was accompanied by a new web-based monitoring system and an enhanced web-based graphical interface.
With it's 5 year anniversary, the WEPROG team channeled later in 2008 and 2009 many of the "lessons learned" into a recoding of the wind power prediction tool and the introduction of an Ensemble Kalman-Filter approach for upscaling, data assimilation of more than only meteorological measurements and ultra-short-term forecasting.
The development of our own database software and hardware was another milestone of 2008. The new database is now fully scalable, freely expandable and is steadily growing. In spring 2009 it was hosting 250mio data files of forecasts for online usage. The database is connected to WEPROG's web-based graphical software package , which has been expanded with advanced features to take advantage of the historic online-database. The advanced features can be used to carry out case-studies and study past events many years backwards in time.
In the beginning of 2009, a semi-operational hemispheric ensemble system with all 75 members has been started in order to improve forecast quality and to make the forecasting processes more efficient. Also because by January 2009 120GW of wind power was located on the Northern hemisphere. The hemispheric ensemble system is expected to significantly improve the forecast quality on all time scales mainly because of the disappearance of boundary problems in the forecasting process.
In 2010 the hemispheric ensemble system was brought into operation and enabled WEPROG to enhance our standard forecast horizons to 6 days (144h). With introduction of the operational hemispheric model, it was now also possible to go up to 10 days ahead in time if required. The year 2010 was otherwise marked by a number of research topics and further optimisations on data handling, the IT infrastructure and communication.
Optimisation work in the IT operations lead to the introduction of WebServices in the beginning of 2011, which enabled faster and more precise timing of data exchange between WEPROG's forecast data generation and our customer's download processes. This was a necessary step to 2 more milestone developments in 2011: the inverted Ensemble Kalman Filter (iEnKF) that now can use any type of measurement in the data assimilation for short-term forecasting and the introduction of our "Early-Bird"-Forecast, which is optimised to meet the European Power Exchange gate closure both in winter and summer time. We achieve an execution time of 11 minutes with the Early-Bird forecast from arrival of the 06UTC analysis to the delivery of 75 weather forecasts inclusive all postprocessing for the energy market products.
The year 2012 has been a continuation of developments in IT Service solutions and Web-based applications to ensure the best possible exchange of data between WEPROG and our valued customers. On the meteorological and engineering side, WEPROG progressed in adding a number of new products that make use of the uncertainty information of the ensemble weather data for customers that are in the trading of energy and push forward the integration of wind power and solar power into the system operations. Two of our major achievements were the publication of strategic trading practices (see WEPROG Information -> Publications ) and the roll-out of reserve forecasts with the use of Ensemble data.
In 2013 the major achievement was to bring reserve forecasts into operation and deliver them in real-time to a transmission system operator. At the end of 2013, efficiency work on the IT side began with a look-ahead to cloud computing in 2014.
In summer 2015 we launched of our first project for the public: our Weather WebApp "find the weather you like" at weather.weprog.com. It is a weather App that is browser based with the purpose of being the same application people work with whether they are on the go with their mobile phones, at home with their tablets or at work with large screens and lots of overview. It's a work-in-progress project that shall bring the great advantages of ensemble forecasting to the public and enhance people's ability to understand weather developments, enhance their ability to plan according to the weather and reduce their environmental footprint.
The years 2016-2018 were in the spirit of research and translating WEPROG's experience into industry gudelines in the IEA Wind Task 36, where our director, Corinna Möhrlen, took a leading role in the development of recommended practices guidelines for the selection of forecasting solutions in the power industry. Additionally, she chaired the IEA Wind Task 36 workpackage on use of uncertainty forecasting in the power industry. All generated publicatios for the IEA Wind Task 36 are open acccess and can be found on the ieawindforecasting.dk webpae and the associated YouTube channel.
In the new phase of the IEA Wind Task 36 work on recommended practices for uncertainty and probabilistic forecasting has been started in 2019 together with industry guidelines for the application of meteorological measurements for renewable energy forecasting in system operations. First drafts are expected in 2020. At two workshops in Copenhagen under the IEA Wind Pathway at the ICEM in June 2019 and at the International Wind Integration Workshop in Dublin in October 2019, first drafts will be presented and discussed within the community.