Abstract

Decision-makers in areas such as agriculture, urban-planning, and civil engineering frequently need estimates of environmental data that are specific to particular sites or small regions, rather than broad spatial aggregates. Many applications require joint information on multiple environmental variables, such as temperature and precipitation, wind speed and humidity, or combinations of air pollutants. Furthermore, the applications are often concerned with estimates of extremes, for example in the design of buildings, roads, bridges and other infrastructure.

From a modelling perspective, the setting is as follows: we are interested in the marginal distribution, or the joint marginal distribution, of one or more environmental variables at any location in space. We have a focus on correctly capturing the tails of the distributions, and it is important to consider if they are non-stationary in time.

In this talk, we will discuss the modelling challenges associated with this setting, how some of the challenges may be solved, and current open questions.

The talk is based on joint works with Peter Craigmile, Max Drains Dahl, Peter Guttorp, Alex Lenkoski, Geir Storvik, Kushagri Tandon and Silius M. Vandeskog.