Evaluation Expert User

The new high-resolution ICON-LEM simulations provide novel insights into clouds and precipitation processes, but in depth an evaluation is crucial to proof the physical consistency of the model. The evaluation expert user coordinates all the different evaluation efforts throughout the project and is the contact point for all questions related to that.

All scientific sub projects focus on various aspects of the model like different cloud microphysic schemes, atmosphere-land exchange and many more, which will be compared with observations and evaluated by them. A synthesis of the different aspects as well as the identification of common features and possible contradictions of the results is essential. The valuable feedback of the community will be communicated throughout the project as well as with the model developers to improve the model and hereby our understanding of the atmosphere. Workshops are organized to exchange results and enhance collaborations.

Online model evaluation

German-wide large eddy simulations as well as of the tropical domain produce big datasets of several terabytes and need large amounts of computational power on the computer cluster of the Deutsche Klimarechenzentrum (DKRZ). Therefore, an instantaneous quality control system is crucial to check for realistic model simulations and physical consistency. In addition, easy to access quicklooks are necessary to screen the big datasets for interesting atmospheric situations and features.

Our goal is to make big data accessible for the scientific community. The online evaluation toolbox was developed within the project. The toolbox combines automatically model results with corresponding observations downloaded from different open access online databases like SAMD and NASA and uploads all quicklooks to the public available DKRZ Swiftbrowser cloud storage. The actual toolbox supplies an overarching ICON-LEM evaluation of basic atmospheric parameters like temperature, humidity, wind speed profiles, precipitation and cloud coverage using various observational products.

Example output for 26 April 2013


The figures below are produced with the evaluation toolbox that can be used to quickly assess the quality of the  high-resolution runs. The code is available at https://github.com/akiohansen/ICON_OnlineEval.

They show among other things the temperature, humidity and wind profiles of different sounding stations and compare them to the 1D meteogram output of the ICON-LEM simulation runs. A color bar represents the match between simulation and observation.

Back to top