Dr. Stefan Rüdisühli
Dr. Stefan Rüdisühli
Staff of Professorship for Atmospheric Circulation
ETH Zürich
- web_assetn.ethz.ch/~ruestefa
- Tel: +41 44 632 75 66
- Tel: +41 44 632 82 27
- call_made0000-0003-1582-1138
- contactsV-Card (vcf, 1kb)
Additional information
Additional information
My main interests are in extratropical weather systems and their analysis in high-resolution simulation data. Specifically, I am investigating fronts and precipitation over Europe, and how they relate to each other.
A decade of European weather at 2.2 km
My PhD project is part of the interdisciplinary project crCLIM, whose goal is to run decade-long convection-resolving (2.2. km) simulations over Europe using the COSMO modelcall_made on GPUs, with most analysis conducted on-the-fly to avoid immense amounts of data output. I am analyzing weather phenomena such as fronts and precipitation in these simulations from a feature-based perspective, in order to create deade-long high-resolution climatologies.
Precipitation
To investigate precipitation events, I track surface precipitation features over time (see below). Attribution to fronts (see below) allows me to distinguish frontal and nonfrontal precipitation. As a first result, the following figure shows diurnal cycle composites of precipitation anomalies. Nonfrontal precipitation shows a diurnal cycle with prominent maxima over the continents in the afternoon and evening, and weaker maxima over the oceans during the night. The frontal precipitation composites show no diurnal cycle, which validates our attribution technique.
Fronts
To automatically detect fronts fronts, I first apply an established method based on ∇θe (Jenkner et al. 2010) to obtain raw front fields, to which I subsequently apply feature tracking (see below). Track analysis then allows me to distinguish between meso-/synoptic-scale fronts, in which I am interested, and near-surface ∇θe structures that develop near orography or convective cells due to the high model resolution.
Feature tracking
To obtain information about the temporal development of features, I track them, which involves following them through time and space. Due to the high temporal and spatial resolution of our simulation, features change little in position and size between consecutive analysis timesteps. This allows me to base our purely diagnostic tracking algorithm on a relatively simple combination of overlap and size criteria. To account for interactions of multiple features, the algorithm support mergings and splittings. The main components of a feature track are illustrated in the following schematic.