Research Project #20: Lagrangian atmospheric transport simulations in: addressing the challenges related to ECMWF’s ERA5 reanalysis
Lagrangian transport models are indispensable tools for studying atmospheric transport processes. Lagrangian transport models simulate the dispersion of trace gases or aerosols by means of trajectory calculations for a number of infinitesimally small air parcels or “particles” following the fluid flow. A major advantage of the Lagrangian approach is that it does avoid numerical diffusion that would be immanent in Eulerian models due to the determination of parameters in spatially fixed grid cells. Due to this low artificial diffusivity, the method is well capable of representing small-scale features such as filaments of tracers associated with long-range transport. Because of their distinct advantages, Lagrangian transport models found a variety of operational and research applications.
Lagrangian transport simulations are driven by external data from meteorological reanalyses. Meteorological data sets provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) are among those data frequently used for Lagrangian transport simulations. In 2006, the ECMWF implemented the ERA-Interim reanalysis (Dee et al., 2011), which has since then been successfully applied in numerous research applications. About a decade later, ECMWF implemented the successor of ERA-Interim, its fifth-generation reanalysis, referred to as ERA5 (Hersbach and Dee, 2016). This new reanalysis opens completely new perspectives for Lagrangian modeling. It comes with many improvements compared to ERA-Interim, most notably higher spatial and temporal resolution, but also other aspects, such as a better representation of geophysical processes in the forecast model and more extensive observational input to the data assimilation system.
However, the new ERA5 data pose significant technical challenges for Lagrangian transport model simulations (Hoffmann et al., 2019). The application of ERA5 at its full spatiotemporal resolution comes along with a substantial increase in computing resources and storage requirements. In particular, the use of ERA5 now poses a “big data” challenge, as the total amount of data increases by a factor ~80 as we progress from ERA-Interim to ERA5. The capability to conduct comprehensive global simulations, long-term simulations for climate studies, or ensemble runs for inverse modeling studies is hampered by these demands. Lagrangian transport simulations driven by ERA5 data require new and extended numerical kernels, data handling, and workflows to cope with the increase in computational requirements.
The particular benefits that come along with using the next-generation ECMWF reanalysis need to be carefully evaluated. First case studies for the Asian Monsoon are considered demonstrating that ERA5 data show faster and stronger vertical transport than ERA-Interim primarily due to ERA5’s better spatial and temporal resolution, resolving more convective events (Li et al., 2019). The Asian monsoon circulation provides an effective pathway for tropospheric trace gases such as pollutants, gaseous aerosol precursors, and aerosol particles into the lower stratosphere (e.g. Vogel et al, 2019). Changes of the chemical composition of this part of the Earth’s atmosphere can have a significant impact on surface climate. Within this project Lagrangian transport simulations will be performed to evaluate the vertical transport in ERA5 in the region of the Asian monsoon.
The main aim of this PhD project is to evaluate Lagrangian transport simulations driven by the new ERA5 data. The work will focus on two complementary Lagrangian transport models, the Chemical Lagrangian Model of the Stratosphere (CLaMS) (Pommrich et al, 2014 and references therein) and Massive-Parallel Trajectory Calculations (MPTRAC). The PhD candidate will spend 50% of her/his working time at the Jülich Supercomputing Centre (JSC) to address computational science aspects related to the study. The remaining 50% will be spent at the Institute of Energy and Climate Research (IEK-7) to focus on the atmospheric science aspects. The new PhD project will foster existing collaborations between IEK-7 and JSC and is considered as an important contribution to the upcoming PilotLab and JointLab projects on Exascale Earth System Modeling of the Helmholtz Association.
In particular, the PhD project will comprise the following work packages:
- Identification, performance analysis, and optimization of “numerical kernels” in the Lagrangian transport models (at JSC)
- Analysis of data structures, memory management and I/O patterns in the models, implementation of parallel I/O concepts, and tuning of simulation workflows (at JSC)
- Intercomparison of 3-dimensional Lagrangian transport simulations using ERA-Interim and ERA5 data (at IEK-7)
- Evaluation of vertical transport in CLaMS using ERA5 data and aircraft measurements in the region of the Asian monsoon (at IEK-7)
 Hoffmann, Lars, et al. "From ERA-Interim to ERA5: the considerable impact of ECMWF's next-generation reanalysis on Lagrangian transport simulations." Atmospheric Chemistry and Physics 19.5 (2019): 3097-3124.
 Rößler, Thomas, et al. "Trajectory errors of different numerical integration schemes diagnosed with the MPTRAC advection module driven by ECMWF operational analyses." Geoscientific Model Development 11.2 (2018): 575-592.
 Li, Dan, et al, “Dehydration and low ozone in the tropopause layer over the Asian monsoon caused by tropical cyclones: Lagrangian transport calculations using ERA-Interim and ERA5 reanalysis data.” Atmospheric Chemistry and Physics Discussions (2019): https://doi.org/10.5194/acp-2019-816.
 Pommrich, Robert, “Tropical troposphere to stratosphere transport of carbon monoxide and long-lived trace species in the Chemical Lagrangian Model of the Stratosphere (CLaMS), Geosci. Model Dev., 7, 2895–2916, https://doi.org/10.5194/gmd-7-2895-2014, URL http://www.geosci-model-dev.net/7/2895/2014/, 2014
M.Sc. degree in Computational Science, Mathematics, Physics, Meteorology, Geophysics (interest in atmospheric and computational science and very good coding skills are a necessity for this project; e.g. MPI, OpenMP, Fortran90, C,...)
Dr. B. Vogel
tel.: +49 2461 61-6221
Dr. L. Hoffmann
tel.: +49 2461 61-1978