NCC presenting the success story
NCC-Bulgaria has been founded by the Institute of Information and Communication Technologies at
the Bulgarian Academy of Sciences, Sofia University “St. Kliment Ohridski” and the University of
National and World Economy.NCC-Bulgaria is focused on:
- Creating a roadmap for successful work in the field of HPC, big data analysis and AI,
- Analyzing the existing competencies and facilitating the use of HPC/HPDA/AI in Bulgaria
- Raising awareness and promoting HPC/HPDA/AI use in companies and the public sector.
Scientific partners involved:
This work was developed in collaboration with colleagues from IICT-BAS, the Technical University of Sofia, and the University “Prof. Dr. Asen Zlatarov” in Burgas. These institutions combine expertise in high-performance computing, numerical analysis, and environmental sciences
Technical/scientific Challenge:
Air pollution remains one of the most critical challenges of modern society, affecting public health, agriculture, ecosystems, and the climate. Predicting the levels and their dynamics in time (due to transport and chemical transformations) of the most dangerous air pollutants requires solving a complex system of partial differential equations that represent advection, diffusion, emissions, chemical reactions and deposition processes in the atmosphere.
Traditional modelling approaches are limited by computational resources and cannot efficiently explore uncertainties in physical or chemical parameters. Therefore, a highly parallel and scalable implementation was needed to allow realistic high-resolution simulations and global sensitivity analyses based on Monte Carlo and quasi-Monte Carlo methods.
Solution:
The team is working on development and optimization of some highly efficient parallel supercomputer versions of the Danish Eulerian Air Pollution Model (DEM), in particular, its unified High Performance Computing (HPC) implementation UNI-DEM and its Sensitivity Analysis extension (SA-DEM). The model was parallelised by using MPI-based domain decomposition and operator splitting techniques to handle advection-diffusion, chemistry, and deposition modules efficiently [1].
The SA-DEM framework integrates various stochastic sampling algorithms (Sobol, MSS-1, MSS-2, OPTL, and quasi-Monte Carlo sequences) in order to calculate multidimensional integrals used to evaluate global sensitivity indices for some of the chemical reaction rates, as well as for the main input emission levels (more precisely, nitrogen oxides NOₓ, sulphur dioxide SO₂, ozone O3, and volatile organics VOₓ).
Comprehensive scalability experiments were performed on two EuroHPC petascale systems: IBM MareNostrum III at the Barcelona Supercomputing Center and Discoverer at Sofia Tech Park. The results demonstrated nearly perfect scaling of the chemical module, slight superlinear speedup due to cache-size effects at moderate process counts, and efficient performance up to hundreds of thousands of cores.
This achievement confirms that advanced environmental simulations can now be performed in realistic time frames, enabling policy-oriented forecasting and large-scale uncertainty quantification [2].
Scientific impact:
Using HPC transformed the computational workflow from week-long serial runs into hour-long parallel simulations.
On both systems, scalability tests confirmed the model’s efficiency for ensemble and uncertainty studies. The integration of Monte Carlo and Quasi-Monte Carlo methods provided robust estimation of global Sobol sensitivity indices, allowing researchers to rank chemical reactions by their influence and to simplify the most complex modules without losing accuracy.
This work contributes to more reliable air-quality forecasts, improved understanding of pollutant dynamics, and accelerated policy support for emission-reduction strategies. It also demonstrates how Bulgarian and European HPC infrastructures jointly advance environmental research under the EuroHPC initiative.
Benefits:
- Reduction of simulation time from weeks to hours;
- Execution of multiple “what-if” emission scenarios in parallel;
- Enhanced accuracy and robustness through global sensitivity analysis;
- Support for evidence-based clean air policy development;
- Demonstration of EuroHPC resources in environmental decision support.
Success story # Highlights:
Keywords: air pollution, environmental modelling, sensitivity analysis, parallel computing, supercomputing, Monte Carlo, Quasi-Monte Carlo, Sobol indices, high-performance computing, scalability
Technology: HPC, HPDA
Industry sector: Environment /Climate/Weather; Energy; Public Services/Civil Protection
Figure 1: Speed-up (Sp) of Advection–Diffusion (blue), Chemistry (green), and the total (red) for the 2D SA-DEM model on the finest grid, computed on the IBM MareNostrum III supercomputer (BSC, Barcelona, Spain).
Figure 2: Speed-up (Sp) of Advection–Diffusion (blue), Chemistry (green), and the total (red) for the 2D SA-DEM model on the finest grid, computed on the Discoverer supercomputer (Sofia Tech Park).
Contact:
- Tzvetan Ostromsky, Institute of Information and Communication Technologies at the Bulgarian Academy of Sciences, ceco[at]parallel.bas.bg
- Silvi-Maria Gurova, Institute of Information and Communication Technologies at the Bulgarian Academy of Sciences, smgurova [at]parallel.bas.bg
- Meglena Lazarova, Department of Mathematical Modelling and Numerical Methods, Faculty of Applied Mathematics and Informatics, Technical University of Sofia and the University “Prof. Dr. Asen Zlatarov” in Burgas, meglena.laz[at]tu-sofia.bg
- Venelin Todorov, Institute of Information and Communication Technologies at the Bulgarian Academy of Sciences, venelin[at]parallel.bas.bg