The prospect of climatic change and its impacts have brought spatial statistics of extreme events into sharper focus. The so-called “water bombs” or “explosive cyclones” are predicted to become more frequent in the extra-tropical regions, and, actually, they raise serious concerns in some regions of the Mediterranean area. However, quantitative statistical methods to properly account for the probability of occurrence of these super-extreme events in formerly untouched areas are still lacking. This is due to their rare occurrence and to the limited spatial scale at which these events occur. In order to overcome the lack of data, different studies concerning flood frequency analysis underline the importance of combining local flood data with additional types of information, to improve the quality of the estimation of the exceedance probability for a given discharge. We propose to apply such a kind of approach in extreme rainfall frequency analysis, with the adoption of a Bayesian framework, aimed at combining local gauge information, discontinuous in space and time, with climatic regional information. The identification of this type of hierarchic relationship can complement local information, conditioning the exceedance probability to the large and meso-scale characteristics of the system, allowing the simulation of design rainfall extremes in sites where historical evidence of that hazard is lacking. The Bayesian approach allows us also to keep track of all the uncertainties involved in the prediction process, producing a measure of uncertainty associated to the estimates. The case study refers to a database of daily rainfall measurements extracted from the NOAA GHCN-Daily dataset, recorded during the 20th century by 700 rain gauges distributed in the Mediterranean basin. First, to identify the conditional variables, we analyse the large-scale environment associated with the different intense precipitation systems in the Mediterranean area, exploiting the reanalysis of the European Centre for Medium-Range Weather Forecasts (ERA-20). With the aim to define the hierarchical relationships between the events and their type, we relate the daily precipitation with different parameters both atmospherical and local. Despite the high variability, different climatic configurations that combined with the local morphology and the seasonal condition of the Mediterranean Sea can trigger very intense precipitation events are identified. Once defined the hierarchical relationships, the parameters are calibrated and the methodology is tested on a subset of daily series provided by local authorities, evenly distributed on the whole domain. The results, compared with those obtained with the classic techniques of frequency analysis and spatial interpolation, demonstrate an increased knowledge coming from climate and local factors, ensuring more reliable and accurate spatial assessment of extreme thunderstorm probability.
Simulating design rainfall extremes in locations with limited observational records / Libertino, Andrea; Sharma, Ashish; Marshall, Lucy. - ELETTRONICO. - (2015), pp. 381-381. (Intervento presentato al convegno MODSIM2015, 21st International Congress on Modelling and Simulation. tenutosi a Gold Coast (Queensland, Australia) nel 29 November - 4 December 2015).
Simulating design rainfall extremes in locations with limited observational records
LIBERTINO, ANDREA;
2015
Abstract
The prospect of climatic change and its impacts have brought spatial statistics of extreme events into sharper focus. The so-called “water bombs” or “explosive cyclones” are predicted to become more frequent in the extra-tropical regions, and, actually, they raise serious concerns in some regions of the Mediterranean area. However, quantitative statistical methods to properly account for the probability of occurrence of these super-extreme events in formerly untouched areas are still lacking. This is due to their rare occurrence and to the limited spatial scale at which these events occur. In order to overcome the lack of data, different studies concerning flood frequency analysis underline the importance of combining local flood data with additional types of information, to improve the quality of the estimation of the exceedance probability for a given discharge. We propose to apply such a kind of approach in extreme rainfall frequency analysis, with the adoption of a Bayesian framework, aimed at combining local gauge information, discontinuous in space and time, with climatic regional information. The identification of this type of hierarchic relationship can complement local information, conditioning the exceedance probability to the large and meso-scale characteristics of the system, allowing the simulation of design rainfall extremes in sites where historical evidence of that hazard is lacking. The Bayesian approach allows us also to keep track of all the uncertainties involved in the prediction process, producing a measure of uncertainty associated to the estimates. The case study refers to a database of daily rainfall measurements extracted from the NOAA GHCN-Daily dataset, recorded during the 20th century by 700 rain gauges distributed in the Mediterranean basin. First, to identify the conditional variables, we analyse the large-scale environment associated with the different intense precipitation systems in the Mediterranean area, exploiting the reanalysis of the European Centre for Medium-Range Weather Forecasts (ERA-20). With the aim to define the hierarchical relationships between the events and their type, we relate the daily precipitation with different parameters both atmospherical and local. Despite the high variability, different climatic configurations that combined with the local morphology and the seasonal condition of the Mediterranean Sea can trigger very intense precipitation events are identified. Once defined the hierarchical relationships, the parameters are calibrated and the methodology is tested on a subset of daily series provided by local authorities, evenly distributed on the whole domain. The results, compared with those obtained with the classic techniques of frequency analysis and spatial interpolation, demonstrate an increased knowledge coming from climate and local factors, ensuring more reliable and accurate spatial assessment of extreme thunderstorm probability.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11583/2651258
Attenzione
Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo