Design and acceleration of new procedures for the automated identification of atmospheric blocking situations

Atmospheric blocking is associated with high-impact weather events in mid and high latitudes. The specific location of the blocking system is of major importance for the type of meteorological extreme developed. The studies on blocking and its effects in the European continent are based on the differences in atmospheric mass between high, middle, and low latitudes and the strong reduction of the zonal winds. They are mainly based on the Tibaldi and Molteni (1990) blocking index or adapted versions of it, in which meridional gradients of geopotential height fields at 500 hPa integrated in longitudinal intervals are calculated.

In this work, we consider an alternative to those approaches, focusing on identifying the atmospheric centers of action and their relative positions and intensities using the geopotential height at 500 hPa. After filtering the local minima and maxima using a gradient criterion, the remaining maxima and minima are grouped according to proximity and intensity. The identification of omega and dipole (Rex) blocks at any instant is made by a selection algorithm that uses the relative positions of the action centers as well as the geopotential contours, identifying the blocking type, shape, and area of influence.

We have analyzed some of the most well-studied heat-wave events over Europe (August 2003, July 2019) and a blocking situation over the western Eurasian border which caused one of the strongest African dust outbreaks impacting Europe (March 2022).

Algorithm performance is crucial in various computational applications, from scientific simulations to real-time data processing. In this work, we focus on improving the performance and speed of the algorithms through parallelization techniques aimed to produce a climatology of blocking situations in the Northern Hemisphere. Specifically, we are exploring the use of OpenMP and MPI libraries for C, widely used in high-performance computing (execution time and scalability efficiency.