Numerical models to simulate storms
Modeling techniques simulate storms numerically and are applied in a joint effort by CliSAP-2 members at HZG and at Universität Hamburg. In addition the STORM project, coordinated and executed at the Max-Planck-Institute of Meteorology, is a partner. Simulations necessary for the project are run on the HLRE-2 supercomputer hosted by the DKRZ.
Compilation of a high resolution longterm numerical hindcast
Multi-source observation data of many atmospheric parameters are routinely recorded and analysed by the national weather services in many countries. Mainly for scientific purposes these data are merged to so-called re-analyses, comprehensive data sets of the global atmosphere, and provided to the public. Re-analyses divide the atmosphere into several grid boxes and calculate the meteorological state in each of them, thereby assimilating the available observations. Due to computational constraints reanalysis data are only provided at a relatively coarse resolution, typically about 200km in the horizontal.
At HZG we have started to use a global model to downscale NCEP/NCAR re-analysis data set to a resolution of about 50km. We expect the such gained data set to even represent smaller types of storm and want to use the data set for multi-decadal and comprehensive investigations of their changing characteristics.
Adjusting previously developed tracking methods
At HZG there is expert knowledge in storm tracking methods, i.e. methods that are capable of automatically finding individual storm events in the large amounts of data generated by the models. These methods typically apply various filtering techniques to extract information at relevant scales to detect storms.
However, previous studies normally applied Regional Climate Models, which only focused on a confined region in the centre of interest. Due to computational constraints, global simulations were too cost intensive. Technically these regionally confined fields of the previous research had different properties than the global data generated in our new approach.
Hence in CliSAP-2 these methods need to be adopted to take account for the different technical properties of the global fields used here.
Analysing the longterm model hindcasts
Tests of the new high resolution longterm numerical hindcast and of the new tracking procedure are currently almost finished and the longterm calculations are underway. Once these are finished the model output will be searched for the different types of storm. The inventory of single cases will statistically be analysed for any changes in their characteristics and linkages.
Nonlocality of tropical cyclone activity in the climate system
This topic is part of CliSAP-B4, but is also associated to the
CRG Dynamical Systems. This topic is aiming at detecting and explaining nonlocal responses of tropical cyclone activity. The term nonlocal in this context refers to a parameter in one region exerting an influence on tropical cyclone activity in another. Here it is hypothesized that the sea-surface temperature (SST) in one ocean basin influences tropical cyclone activity in another basin.
Led by Prof. Thomas Frisius idealized experiments were conducted applying the Planet Simulator of the University of Hamburg. These experiments were set with an Earth consisting of two idealized continents separated by two idealized oceans. Indeed a nonlocal response of the tropical cyclone activity was found (Figure 1), i.e. when SST is perturbed in the ocean to the right, not only more tropical cyclones (TC) are found here, but also a reduced number is found over the left ocean.
These results are confirmed by current analyses using continents of the real world. They found a similar response in the Atlantic when the SST in the Pacific is artificially heated. In further studies the sensitivity of TC activity to changed land surface properties will be analyzed.