Presenting NeIC projects: iOBS
NeIC’s iOBS project started in January 2019. The name iOBS stands for Improved Observation Usage in Numerical Weather Prediction. This article is written in collaboration with the project leader, Anette Lauen Borg from the Norwegian Meteorological Institute.
What is iOBS?
The purpose of the iOBS project is to provide research support for United Weather Centres (UWC), a cooperation on numerical weather prediction between 10 countries: Denmark, Finland, Iceland, Ireland, Netherlands, Norway, Sweden, Estonia, Lithuania and Latvia. Instead of each individual country buying and maintaining their own High-Performance Computing center to run their own weather model, the aim is to merge all the UWC partners’ computing resources.
The iOBS partners consist of CSC - IT Center for Science in Finland as well as the national meteorological institutes in Finland (FMI), Norway (MET Norway) and Sweden (SMHI). CSC contribute with their cloud based computing systems and IT staff, while the national meteorological institutes provide the meteorological resources for the project.
The UWC collaboration project is carried out in two steps. 1) MetCoOp, the ongoing cooperation between the national meteorological institutes of Finland, Norway, Sweden and Estonia that joined in January 2020 will be extended with Lithuania and Latvia. 2) A new operational collaboration between Denmark, Iceland, Netherlands and Ireland is being built parallel to MetCoOp.
Both of these collaborations will be effective from 2022, and by the year 2027 the merged UWC will include all 10 national meteorological institutes. The iOBS project is running until 1.1.2021, with a probable extension to June 2021, providing research results needed by the first step and later by the merged UWC.
Numerical Weather Prediction (NWP) is carried out by most of the individual national meteorological institutes in the UWC area. The numerical weather prediction model is the source of weather forecasts, and as input it needs weather observation data. In order to receive and process the weather observations provided by all the future UWC partners, as well as crowdsourced data from private citizens, it is important to plan and test the unified data flows. The iOBS project is both developing open software and testing available solutions to prepare for the expected, massive amounts of data.
Yr is a good example of a public service made possible by numerical weather prediction. Yr.no is a joint online weather service from MET Norway and the Norwegian Broadcasting Corporation NRK which offers weather forecasts in three languages for around 1 million places in Norway and 10 million places worldwide. It is used by up to 9 million people every week. The forecasts provided are a direct product of the NWP run by MetCoOp, which means that there doesn’t need to be any humans involved. Even a small increase in accuracy can affect millions of people in the Nordics, especially when UWC is in full operation.
Goals and expected results
During its lifetime, iOBS aims to develop data flow solutions that can serve as a sketch for future UWC systems. The main goals set for the project are listed below.
- Test new software to better exploit the weather observations data collected by the national meteorological institutes
- Make use of private weather observations in the weather prediction model
- Introduce machine learning in the quality control of weather observations
- Explore new dataflow solutions to handle the large amounts of incoming observations data in the near future
As a first step, existing software provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) is being tested as a tool to collect the public weather observations available to the meteorological institutes. The output of the software is sent to a customized numerical weather prediction model set up and customized by the project. Preliminary results from January 2020 show that the ECMWF software helps exploit the input data better.
At the same time, work to develop quality control software to process observations data from private weather stations is being carried on. Up to now, these data have been used to correct the weather forecast after the numerical weather prediction model has run. However, the iOBS project is working to make the model accept these data as input alongside the public weather observations and to find out if this will improve the weather forecasts provided to the public.