 |
| Titel |
Browsing large natural hazard event sets |
| VerfasserIn |
Aidan Slingsby, Jane Strachan, Pier-Luigi Vidale, Jason Dykes, Jo Wood |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250053831
|
|
|
|
|
|
| Zusammenfassung |
| Utilising output from dynamical modelling of the global climate system to produce large sets
of simulated events is becoming an increasingly important means of assessing risk from
climate-related extremes. Such event sets can be generated for timescales that exceed those
available from the historical records. As a result, they may better represent natural
variability, and the impact of natural variability on extreme events. With the use of
increasingly large datasets of natural hazards, comes new challenges for validating,
analysing and presenting the results. Data visualisation has an important role in
exploratory data analysis, but these techniques are generally underused in natural hazard
modelling.
The National Centre of Atmospheric Science (NCAS) have generated thousands of simulated
storm tracks using a General Circulation Model and a storm tracking algorithm (Hodges,
1995). Data points are at six hourly intervals with vorticity and wind speed for several
atmospheric levels. Examples of storm tracks that illustrate important implications of
atmospheric risk are selected and used to accompany talks and presentations to the insurance
industry.
We developed rapid visual browsing techniques to assist in the identification and extraction of
such examples. However we found that these techniques were also effective means for
exploration analysis, for helping validate the dataset and for identifying issues worthy of
further investigation (Slingsby et al, 2010).
We present examples of how our visual browsing techniques have helped generate new
insights from the data and identify compelling examples of storm tracks that illustrate
aspects of atmospheric risk of interest to the insurance industry. These include:
spatial and temporal clustering of tracks with potentially serious implications of
risk to the insurance industry
the observation that many extratropical storms have tropical origins, significant
to the insurance industry because these types of events are usually modelled
separated, therefore not correlated in time
characteristics of storms at landfall
We are also adapting and extending the tool to help pursue other research questions.
These include:
establishing the atmospheric conditions under which particular configurations of
storm tracks occur
establishing the differences within El Nino years, La Nina years and neutral years |
|
|
|
|
|
|