Premium
Modelling extreme rainfall data within a catchment region
Author(s) -
Cocchi Daniela,
Scagliarini Michele
Publication year - 2003
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.554
Subject(s) - extreme value theory , drainage basin , generalized extreme value distribution , environmental science , catchment area , covariate , maxima , data set , statistics , maximum likelihood , hydrology (agriculture) , climatology , mathematics , geography , geology , cartography , geotechnical engineering , art , performance art , art history
The present work examines the extreme behaviour of monthly rainfall in the Reno River catchment area (Italy). The data are recorded by a network of monitoring sites for which a single model, valid for the whole region, can be built. As a first step, a generalized extreme value (GEV) model is fitted to monthly rainfall maxima for each site. As a further step, we have introduced a second level into the model whereby the GEV site‐specific parameters depend on covariates by means of hyperparameters common to the area. The maximum likelihood method is used for the purposes of estimation. We employed an extra‐sample set of monitoring sites in order to compare the results obtained using the catchment model with the single‐site models estimated by classical one‐level extreme value analysis. Copyright © 2003 John Wiley & Sons, Ltd.