KNMI DataLab

The KNMI DataLab is a new entity inside KNMI to deal with the new and dynamic challenges of the Big Data age.

The DataLab aims to be the central point at KNMI for Big Data-related topics. Since decades, KNMI handles huge amounts of data for weather forecasting and climate modeling that increase every year. In addition to this big volume of data, the DataLab explores other aspects of Big Data such as velocity (real-time remote sensing), variety (combination of different data from various sources), and veracity (assess the level of confidence of gathered data). These Big Data aspects and the combination of meteorological data with data from other domains provide more insights and additional value.

 

KNMI DataLab data-driven innovation lifecycle
KNMI DataLab data-driven innovation lifecycle
WOW-NL
WOW-NL

Activities:

To support KNMI in dealing with the challenges of Big Data, the DataLab performs the following activities:

  • support of experiments and projects with data science skills and knowledge
  • guidance to set up the best environment for projects and to select the most appropriate technology for the challenge
  • internal and external collaboration to create new knowledge by the addition of the meteorological dimension to the domain of the experiment under investigation
  • provision of an effective platform for sharing data and reproducible experiment results

Projects:

Currently, the DataLab is engaged with two main thematic areas around which a series of projects revolve.

The first theme is “KNMI in the Street”: the goal is to achieve high-resolution weather and air quality observations and forecasts. Citizens in a modern society are more and more demanding accurate weather conditions where it matters them most, thus in the streets where people live. There are two experiments running at the moment. The first aims to observe and forecast  meteorological conditions at fine grained level using crowd sourced weather observations (https://wow.knmi.nl/).  The crowd sourced data come from hundreds of passionate users that want to share their weather station data.  The second experiment aims to combine air quality observations  from a multitude of  heterogeneous sensors. These sensors vary from the national high accuracy air monitoring network, to relatively low-cost sensors used in city networks, and satellite atmospheric observations. The goal is to have a real-time high resolution air quality maps.

The second theme is “Weather and Traffic”: the goal is to better integrate and understand the relationship between weather and traffic. This enables more accurate and localized preventive warnings about road conditions and traffic re-routing to avoid critical situations. One experiment deals with automatic recognition of fog conditions from camera pictures. An artificial intelligence approach is applied to process images to detect fog situations using traffic cameras. The second experiment consists of the integration of high-resolution weather in traffic models and to reconstruct accidents conditions.