VTT predicts extreme events more accurately
VTT Technical Research Centre of Finland has developed an improved method for estimating the probability of extreme events, such as storms, floods and earthquakes.
The new method will be used to update building codes and land-use regulations, and has a wide range of applications in, for example, developing artificial intelligence, economics, and medical data analysis.
To estimate the probability of extreme events, statistics of previously observed extremes are subjected to so-called extreme value analysis, with VTT researchers coming up with a more accurate such method.
“Extreme value analysis of natural phenomena aims at appropriate preparation against hazardous extreme events,” said VTT researcher Lasse Makkonen. “Therefore, it forms the foundation of all regulations that aim at securing the safety of buildings and infrastructure. Stronger constructions and protection arrangements increase costs, so that economic [optimisation] is another issue here.”
Makkonen developed the new method with VTT colleague Maria Tikanmäki and proved with numerical Monte-Carlo simulations that it was much more accurate than its widely used alternatives. The difference was particularly significant in regards to cases where the number of previously observed extremes was limited.