Cloud Microphysics

Cloud microphysics is a field of studies of processes occurring at microscales, particularly formation and evolution of cloud particles. In warm clouds (i.e. void of ice) describing rain formation is still a challenge. Droplets particles first grow by condensation, which efficiency diminishes strongly with increasing droplet size. Raindrops are formed when droplets are big enough to collide and coalesce, thus increasing their size rapidly. However, there is the range of sizes where neither condensational nor collision-coalescence growth rate is big enough. This so-called ‘gap-size’ problem is the major challenge in simulating clouds. Simulating collisions between cloud droplets is a complex task due to various factors involved. It requires considering droplet properties such as size, velocity, concentration, as well as environmental factors like turbulence.

University of Warsaw Lagrangian Cloud Model (UWLCM)

  • Anelistic Large Eddy Simulation (LES);
  • 2D or 3D simulations;
  • The Multidimensional Positive-Definite Advection Transport Algorithm (MPDATA) used to model transport of Eulerian variables;
  • SGS turbulence modeled with Smagorinsky scheme or implicit LES;
  • Models for sub-grid scale (SGS) motion;
  • Super-Droplet Method (SDM):
    • The Super-Droplets (SD) are computational objects used in Lagrangian particle-based microphysics methods;
    • Each SD represents a large number of identical real particles. SDs are described by their attributes: position, dry or wet radius and multiplicity;
    • Super-Droplets microphysics can be calculated using GPUs;

A more detailed description of UWLCM can be found in Dziekan et al.(2019) and Dziekan and Zmijewski (2022). Figure obtained by P. Dziekan in ParaView from results of UWCLM.

Modeling Collision-Coalescence in Particle Microphysics

A fundamental component of the Super-Droplet Method is the All Or Nothing (AON) algorithm, which determines attributes of SDs following collisions between two SDs. The AON algorithm ensures mass conservation during collision events. Let's consider two Super-Droplets, SD₁ and SD₂, with respective radii (r₁ and r₂) and multiplicities (ζ₁ and ζ₂). If ζ₂ exceeds ζ₁, the post-collision properties of SD₁ and SD₂ are described as follows: ζ₂ decreases by ζ₁, while r₂ remains unchanged. In the case of SD₁, ζ₁ remains unchanged, but r₁ increases in accordance with mass conservation i.e. r1new = (r13 + r23)1/3.

High Performance Computing

Using the UWLCM for research requires significant computational power, typically in the form of heterogeneous clusters consisting of CPUs and GPUs. These resources are essential for handling the complex calculations and simulations involved in my field. Throughout my research, I had the opportunity to make use of various computing resources. I made use of the resources provided by the ICM (Interdisciplinary Centre for Mathematical and Computational Modelling) at the University of Warsaw. Additionally, I benefited from the computational capabilities offered by Cyfronet, the Academic Computer Centre located in Poland. Their resources greatly contributed to the successful execution of my simulations and data analysis. Furthermore, I had the opportunity to work at the Hyogo Supercomputer Center, located at the University of Hyogo in Japan. During my 3-month internship, I had the privilege of utilizing their high-performance computing facilities to carry out advanced calculations and simulations. Throughout my internship, I extensively worked with the SCALE-SDM (Scalable Computing for Advanced Library and Environment for Simulation and Data-driven Decision Making) model developed by RIKEN. Lastly, I had the privilege of exploring the capabilities of the Fugaku supercomputer, developed by RIKEN and located in Japan. The cutting-edge technology offered by Fugaku enabled me to push the boundaries of my research and gain valuable insights. Below, you can find a glimpse of some of the noteworthy results obtained through my research. These findings highlight the impact and potential of utilizing powerful computing resources in advancing our understanding of weather and climate phenomena.

Skills

I have experience working with various programming languages and tools, including:

Python

I am proficient in Python and have utilized it for data analysis, scientific computing, and developing applications.

C++

I have a strong command of C++ and have utilized it for high-performance computing and algorithm development.

FORTRAN

I have experience with FORTRAN and have utilized it for computational physics and numerical simulations.

BASH

I am proficient in BASH scripting and have utilized it for automating tasks and working in a UNIX/Linux environment.

SLURM

I am familiar with SLURM (Simple Linux Utility for Resource Management) and have used it for job scheduling and management on HPC clusters.

GIT

I have experience with GIT version control and have used it for collaborative software development and managing code repositories.

PBS

I have experience with PBS (Portable Batch System) and have used it for job scheduling and management on HPC clusters.

LATEX

I am proficient in LATEX typesetting system and have used it for writing scientific papers and documents.

Matplotlib

I have experience with Matplotlib, a Python library for data visualization, and have used it for creating informative plots and graphs.

Get in touch