Dissertation in the field of Control Engineering, Pekka Juhana Rossi

The title of thesis is Object-Oriented Analysis and Nowcasting of Convective Storms in Finland

Severe convective weather causes hazardous events and damage throughout the world. The hazards produced by intense winds, flash floods, hail, tornadoes, and lightning have significant societal impacts. Since an individual storm may have a lifespan of a few tens of minutes and the spatial extent of a few kilometers, continuous real-time measurements are needed to identify convective storms, to assess their potential impacts, and to forecast where the storms are likely to occur.

Remote sensing instruments such as weather radars and lightning location systems are capable of fulfilling the requirements for high temporal and spatial resolution. These measurements are often applied to identify convective storms automatically and to forecast their future occurrence. A common approach for the automatic identification and short-term forecasting of these storms is object-oriented storm tracking, which follows the movement of individual storms from remote sensing data and then extrapolates the storms based on the tracking information.

This thesis demonstrates that object-oriented storm tracking provides a well-established and versatile approach for analyzing properties of individual convective storms. Further, it introduces new methods that can be applied to quantify the severity of tracked convective storm objects together with multiple complementary data sources such as various radar and lightning-based parameters. Moreover, tracked storms can be supplemented with information

sources such as weather-related emergency reports from the general public to further characterize the hazardous properties of the storms in real time.

In addition to the analysis of storm severity, this thesis is concerned with the short-term forecasting of storms. Because of the small size and short lifecycle of a typical convective storm, forecasting is challenging and subject to significant uncertainty. Conventional object-based storm forecasting methodologies provide forecasts only in a deterministic fashion without any guidance regarding related uncertainties. This work makes an important contribution by proposing a new method for probabilistic forecasting of individual convective storm objects that explicitly considers the uncertainties of the forecast. The thesis demonstrates that the probability forecasts produced by the method are reliable and exhibit an improved accuracy, measured by Brier score, over deterministic forecasts produced by a conventional object-based nowcasting method.

Opponent: Dr. Alan Seed, Bureau of Meteorology

Supervisor: Professor emeritus Heikki Koivo, Aalto University School of Electrical Engineering, Department of Electrical Engineering and Automation