![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Spatial models for climate extremes. Application to extreme snow depth in the Swiss Alps. |
VerfasserIn |
Juliette Blanchet, Michael Lehning |
Konferenz |
EGU General Assembly 2010
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 12 (2010) |
Datensatznummer |
250036095
|
|
|
|
Zusammenfassung |
In recent years, there has been a growing interest in spatial extremes within the climate
community. For many practical applications, an important issue is to model extreme events
anywhere in space, and not only at some specific locations. A naive way of doing this is to
estimate the distribution of extremes (the so-called Generalized Extreme Value -GEV-
distribution) at individual locations and to make prediction where data are not available in a
second step (using kriging for example). Such a method should nevertheless be avoided and
may lead to unrealistic results. A better approach is to model directly the GEV distribution as
a smooth function in space. With the help of good covariates, this simple method
allows to build realistic return level maps of extremes events without requiring a
posteriori interpolation. It nevertheless relies on the assumption that extremes at
different places are uncorrelated which is obviously wrong for climate extremes. The
theory of max-stable processes provides the most natural way for the specification of
spatially-dependent extremes. Max-stable processes can be seen as an extreme value analogy
of Gaussian processes. But they have rarely been applied to real climate problems and
may be too simplistic models to describe the complexity of climate extremes in
reality. We therefore propose a modified max-stable characterization of extremes
designed explicitly for climate data. It can in particular account for the presence
of climate regions and directional weather movement effects. Performance and
comparison of the different approaches discussed above are illustrated by analysing
annual maximum snow depth distributions in Switzerland. The new method shows a
higher performance and is able to provide more information than the other methods. |
|
|
|
|
|