Our department develops, improves, runs and analyses weather and climate models for weather forecasts and climate projections, based on knowledge of the important physical processes, sources of variability, advanced statistical and machine learning methods, and techniques to measure model quality.
Improvement of, and research into, weather and climate models supports one of KNMI's most important core tasks: warning and advising on weather and climate conditions that have a major impact on society. The guiding principle here is demonstrable and continuous improvement of both forecasts and projections, and the interpretation of high-impact events.
We use the following weather and climate models:
Harmonie-Arome. This is the high-resolution weather and climate model used by KNMI for short-term weather forecasts. The model is used and further developed at KNMI and in European collaborations in the ACCORD consortium (A Consortium for Convection-scale modeling Research and Development) and in UWC-West (United Weather Center West - more information in English or Dutch).
IFS (Integrated Forecasting System) and OpenIFS (the open source version of IFS). This is the global weather model used for medium-term forecasting. The model is being further developed under the direction of the European center ECMWF.
EC-Earth. This is the Earth system model used for global climate simulations, including future projections. EC-Earth is developed by a European consortium consisting of national meteorological services and research institutes, and KNMI is one of the core partners of the consortium. The model is used, among other things, for contributing to the Coupled Model Intercomparison Project (CMIP) and, hence, to the assessment reports of the IPCC.
RACMO and HCLIM. These are the regional climate models that are used, among other things, to create the climate scenarios for the Netherlands.
Lotos-Euros. This is the regional air quality model used for short-term air quality forecasts in the Copernicus Atmosphere Monitoring Service. The model is being further developed in collaboration with TNO.
You can read more about weather and climate models here: Infographic KNMI weather and climate models
The department is divided into 8 clusters:
The 'Data Science' cluster performs research and development (R&D) using data science methods, mostly statistical or machine learning (ML) techniques, applied to numerical weather prediction (NWP) model output. The group consists of about 10 people.
The largest part of the cluster is involved in statistical post-processing of NWP model output, with the goal of improving NWP model forecasts by correcting systematic model errors. Lead times vary from hours (nowcasting) to months (seasonal forecasting). The use of probabilistic weather forecasts is central because of the inherent uncertainty in weather prediction.
Extreme weather: focus is on improving probabilistic forecasts of extreme weather events, as those are important to improve the weather warnings of KNMI
Verification: to know how skillful probabilistic forecasts are or how much they have improved, verification of these forecasts is important
Physics-based: Improving weather forecasts using physics-based ML techniques.
Contact Maurice Schmeits for more information about this cluster.
On a daily basis there is an abundance of observations available that can be used to improve the start of Harmonie-Arome model runs (Figure 1). A complicating factor, however, is the various differences between these observations, for example, in quality, coverage or time frequency. It is the model component called data-assimilation, being developed in the ACCORD consortium (http://www.umr-cnrm.fr/accord), that tries to find a suitable model start by combining available observations and model data.
The 'Data Assimilation' cluster contributes to this effort by preparing new observation sets for assimilation and to improve the actual numerical algorithms that perform the assimilation. The current focus is on preparing for the future satellite MTG-IRS, which will produce profiles of temperature and humidity every 30 minutes. This is unprecedented and the data is ideally suited for a recently developed data-assimilation algorithm.
Improve the use of observation by developing pre-processing techniques and design of flow dependent algorithms like 4DVAR (four-dimensional variational data-assimilation)
Explore new observations and prepare for future (satellite) observations like MTG-IRS (Meteosat Third Generation InfraRed Sounder)
Infer the consequences of sub-kilometer resolution for data assimilation, with respect to observations use and type of algorithms
Diagnose the use of observations in models and explore their use for verification
Ensemble forecasting offers the opportunity to receive early warnings of severe weather that could have been missed otherwise by deterministic forecasting. Especially for high resolution models like Harmonie-Arome probabilistic forecasting is unavoidable.
In the cluster ‘Ensembles’, research is focused on improving the Harmonie-Arome ensemble performance, in particular for low visibility events. To that end sensitive model parameters have been selected in relevant physics parameterizations and made part of the stochastically perturbed parametrization (SPP) scheme. This scheme is central in the common framework of the Harmonie-Arome ensemble community (www.hirlam.org) to generate ensembles. As such, the team contributes to the definition of the UWC-W ensemble.
Future work will involve the selection of other sensitive model parameters and devising techniques to optimally combine perturbed model parameters.
Extend the stochastically perturbed parametrization scheme with other parameters.
Explore ways to verify the ensemble.
Contact Jan Barkmeijer for more information about this cluster.
HARMONIE-AROME is the high-resolution weather and climate model used, and partly developed, at KNMI. Harmonie has proven its value in operational practice, especially in cases of extreme weather conditions. Nevertheless, model deficiencies remain and especially improvements in the model physics have a large impact on the model quality. Therefore, in our cluster a continuous effort is made to improve the model by means of improved physics. This is done in close collaboration with our partners in the ACCORD (A Consortium for Convection-scale modelling Research and Development) consortium. A main focal point in our research concerns very high-resolution runs (between 100 and 500m, Figure 2) which are especially relevant for forecasts of e.g., high precipitation, wind (gust) and urban climate.
Improved representation of the physical processes with a focus on: turbulence, convection, urban surface, clouds and microphysics.
Adaptation of the model physics to increased resolutions.
Continuous model evaluation to determine the strengths and weaknesses and to provide inspiration for future model developments.
Contact Wim de Rooy for more information about this cluster.
The 'Climate Attribution' cluster answers the question whether and to what extent an extreme weather event is related to climate change. In attribution studies we use observations to detect trends (if present) and climate models to test whether detected trends are related to climate change. We provide information about the probability of weather events, the influence of climate change and other factors on the weather event, the uncertainties and about the vulnerability of society to extreme weather events. Not only meteorological factors but also social circumstances determine the impact of an extreme weather event.
Current activities consist of:
performing rapid attribution studies after extreme weather events
performing slower, more in depth attribution studies
research to expand current methods towards other types of extremes
contributing to an operationalization of attribution in the EWC
translating results to different audiences including the general public and decision makers
Contact Sjoukje Philip for more information about this cluster.
The main goal of the ‘Global Climate’ cluster is to provide a platform for development and a strategic knowledge base for issues related to global climate change and large-scale phenomena such as storm tracks, role of the Arctic on our climate, El Niño and Atlantic overturning circulation, and their impacts on our climate. The cluster contributes to the development of the EC-Earth Earth system model, and uses the model to study the global climate system and address specific science questions. The cluster participates in several EU research projects, and contributes to international efforts, such as the Coupled Model Intercomparison Project (CMIP).
Future changes in weather extremes
Global warming-induced atmospheric circulation changes over Europe
Physical drivers of atmospheric circulation changes (ocean, vegetation, sea-ice)
Tipping points and abrupt shifts in our climate
Contact Sybren Drijfhout for more information about this cluster.
The ‘Air Quality’ cluster develops atmospheric chemistry modules for regional (Lotos-Euros) and global models (IFS, OpenIFS, EC-Earth). These modules simulate the life cycles of pollutants, greenhouse gases and aerosols and their interactions with other components of the Earth system.
Lotos-Euros delivers daily forecasts of air quality for the Netherlands and Europe as well as (re)analyses.
The Integrated Forecasting System (IFS) delivers global atmospheric composition forecasts and (re)analyses as part of the Copernicus Atmosphere Monitoring Service (CAMS) coordinated by the European Centre for Medium-range Weather Forecasts (ECMWF).
OpenIFS, a version of the IFS model that is available to research institutes, is being included in the EC-Earth Earth system model to study interactions between chemistry, aerosols and climate, and produce climate scenarios.
The models are also applied to provide advice to policymakers on air quality and climate change.
In the cluster ‘Water’, much of the research is directly or indirectly related to threats from flooding. This can either be from rivers or from the sea. An important part of the research focuses on the question whether the (relatively short) observational records can be replaced by synthetic weather from climate models.
The most important activities of the cluster are (in arbitrary order):
Running the state-of-the-art surge model in order to supply the Watermanagementcentrum Nederland (WMCN) with ensemble forecasts up till 15 days ahead.
Derivation of the return values that are relevant for the design criteria of the sea dikes. The focus is mainly on extreme (once-in-10,000-year and beyond) wind speeds and water levels.
Derivation of the return values that are relevant for the design criteria of the river dikes (Rhine, Meuse, Vecht). The focus is mainly on extreme (once-in-1000-year and beyond) multi-day precipitation extremes and (in cooperation with Deltares) discharges.
Contribution to the KNMI climate scenario's, mainly focusing on a robust representation of (changes in) extremes.
Contact Henk van den Brink for more information about this cluster.
The ‘Sea Level’ cluster conducts research about various aspects of the climate system that impact sea level. Based on this research, we provide up-to-date and reliable sea level information to Dutch ministries (Infrastructure and Water management, Economic affairs), research institutes (Deltares, PBL) and to the Dutch Delta Program.
We develop physical and statistical models to answer important questions for the adaptation and mitigation strategy of the Netherlands: How fast is sea level rising? Why? How fast will sea level rise in the future? How fast will Antarctic ice melt in a warmer climate?
Develop sea level scenarios for the coast of the Netherlands
Understand past sea level observations from tide gauges and satellite altimetry and their relation to climate change
Improve the representation of ice sheets in climate models
Understand the interaction between climate and the melting of ice sheets
Contact Dewi le Bars for more information about this cluster.
The KNMI Climate Explorer is a web-based climate data browser, used by many students, researchers and practicioners worldwide.
Auxiliary information on the department activities is provided by a few selected researchers that present their work on their individual KNMI webpage (not all research staff are represented here). All publications of the department can be found in the KNMI publication database.