Radar-based nowcasting of severe thunderstorms: A better understanding of the dynamical influence of complex topography and the sea

Natural disasters of hydro-meteorological origin are the biggest risk worldwide. In Catalonia (NE of the Iberian Peninsula), severe weather and flash floods occur each year, resulting in major damage to property, losses in agriculture, and also of human lives. To reduce its impact, we need to improve the early warning systems and storm short-term forecasting. There’s a need to gain in-depth knowledge of severe thunderstorm dynamics, since the current accused conditions of global warming can impact in factors triggering these storms. The main objective of the present thesis is to enhance the knowledge of severe storms dynamics and to improve their identification and monitoring in real time, in order to help prevent their surface effects on the citizens. The project addresses the unresolved problem of storm anomalous motion, as it becomes a great challenge to predict their evolution and impact in the next few hours. For this purpose, the area of Catalonia has been chosen as the study region of this project, due to the proximity to the sea and complex topography, which are often key factors in varying the weather at a local scale. There is also the advantage of having good radar coverage, which will be the essential tool for characterizing storms. We first propose a methodology that identifies potentially convective days from daily cumulative rainfall fields, selects them to search for storms, and determines if their motion is anomalous. We have found that the area with the highest convective activity between 2008-2015 in Catalonia was located in the eastern Pre-Pyrenees, due to the possible creation of a convergence line. It has also been identified that there are more convective structures with possible anomalous propagation in summer and spring, with the main patterns being related to splitting, merging, stationarity and elongated storms. Once the study sample is defined, we have developed an algorithm to improve the identification and tracking of these thunderstorms, especially those with anomalous propagations. The keys of improvement have been based on proposing new techniques in the three main modules; 2D, 3D identification and tracking. In addition, it incorporates alerts before possible cell splitting or merging. These changes have shown that the algorithm is able to faithfully reproduce storm life cycle, correctly identify in advanced anomalous motion, and correctly distinguish storms in highly dense convective situations. The algorithm has been ver