Task 2.4 Development and Testing for an Optimized Nesting Strategy for Europe and Benchmarking

The experiences made in the MiKlip project (e.g. Feldmann et al., 2019) showed that refinement in a target region can improve the accuracy and reliability of decadal predictions, and provide added value for user and impact relevant variables (Mömken et al., 2021). This task aims to systematically test the options for a nesting over the European target region and assess their added value. Therefore, we will develop a strategy for dynamical downscaling for the atmospheric component of ICON, considering both the choice of the nesting steps and the optimal position of the higher resolution domains. ICON offers several possibilities for nesting or downscaling. The direct (2-way) nesting allows for a feedback of the higher resolved regions on the larger scales. In this case, the nesting steps are limited to a factor of two, which requires several successive nesting steps. An alternative way could be via the limited area mode of ICON(-LAM), which would enable larger nesting ratios over a target region, but without feedback. Starting from a global model resolution of R2B5 (80 km), the suggested set up will be first to define a nesting step to R2B6 (40 km) which covers large parts of the North Atlantic and Europe. Here, we will test the 2-way nesting variation to retain the relevant climate feedbacks. The following step will be to choose an optimal second nesting step targeting Europe. Variants with R2B7 (20 km) covering most of Europe with and without 2-way nesting will be tested, together with options for R2B8 (10 km) and R2B9 (5 km) resolution in LAM mode for smaller domains focussing in Central Europe. For the purely one-way nesting strategy option, only two steps need to be performed, namely R2B7 and R2B9. The performed simulations will be used for both, testing the stability of the model set-up for long (decadal) simulations and evaluation the performance of the model set-up for higher resolutions. The latter focuses on evaluating the added-valued due to downscaling particularly for user-specific indices for sectoral applications in cooperation with Task 2.2 and 2.3. The results will indicate an optimal starting resolution and act as a reference for a further statistical downscaling to the kilometer scale. A recommendation for future perspectives regarding an operational system will be given together with WP1, based on a balance between the added value of high spatial resolution, the largest possible ensemble and feasibility (especially in terms of computing time and storage capacities).

Contributors

Karlsruhe Institute for Technology
Prof. Dr. Joaquim Pinto
Dr. Patrick Ludwig
Hendrik Feldmann
Dr. Namendra Kumar Shahi

Max Planck Institute for Meteorology