The use of low-cost, efficient GPU-based parallel computing in lightning modelling

• The major approaches in lightning propagation modelling are briefly introduced.
• A short summary on calculations done during lightning simulation is given.
• The main challenges of parallel computations are described also a brief summary on GPU based parallel computation is given.
• A comparison is given between single and parallel calculations of the typical tasks in lightning propagation modelling.

Lightning modelling became a very important topic in lightning research in the last decade. More and more complex calculations were made available due to the rapid development of computing capabilities. Numerical models were implemented providing important and interesting results regarding both primary and secondary effects of lightning. The implementations of these models were bottlenecked with the available computing resources, as they required resources available in top-of-the-line supercomputers. Nowadays though, even desktop PCs are able to fulfil such tasks. This paper aims to raise awareness on the availability of low-cost efficient computing tools, which are nowadays widely available in standard PCs in the form of graphical cards, GPUs. This paper focuses on briefly introducing the use and the efficiency of the GPU aided parallel computation in lightning modelling tasks. After a review of the current GPU architecture, some currently used complex physical lightning models are briefly discussed, and the ‘typical’ computational tasks are collected. Then a performance analysis of parallelisation is given to show the incredible advantage of the use of parallel computing.