|
|
 |
|
|
|
|
|
| |
Information - IS40 - NP3.04/HS1.8/NH10.5 Geophysical Extremes: scaling versus nonstationarity (co-organized by NP, HS & NH, co-listed in AS)
 |
|
|
|
Event Information |
|
|
|
|
|
|
Probabilistic treatment of geophysical processes is typically based on the classical statistical paradigm, which is greatly influenced by the assumptions of independence and Gaussian distribution. Even when dependence is considered, this is typically done by a Markovian assumption, so that the dependence weakens rapidly as we move from smaller to larger lags or scales. Currently, however, it has become clear that geophysical processes do not correspond to this typical model, and exhibit scaling behaviours in time, space and state. The scaling behaviours imply non-Markovian, long range, dependence, as well as non-Gaussian distribution. To make the conventional statistics compliant with such a behaviour, the notion of nonstationarity is often invoked, which is, more often than not, unjustified, since its key condition that the change is a deterministic function of time (or space) may not be fulfilled. Thus, a better probabilistic and statistical framework is needed, which should incorporate the empirically observed scaling properties in a theoretical framework. In addition, a clarification of the conditions that may justify the nonstationarity notion is necessary.
The session aims to discuss the notion of scaling on empirical, as well as theoretical, probabilistic grounds, and to shed light on the notion of nonstationarity within such a new framework. Geophysical extremes provide an ideal domain for the focus of this discussion as they are known to exhibit different types of scaling behaviour in time, space and state.
|
|
|
|
|
|
|
 Back to Session Programme
|
|
|
|
