Premium
Statistical downscaling of precipitation for a selection of sites in Ireland employing a generalised linear modelling approach
Author(s) -
Fealy Rowan,
Sweeney John
Publication year - 2007
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/joc.1506
Subject(s) - downscaling , precipitation , climatology , environmental science , quantitative precipitation estimation , climate change , selection (genetic algorithm) , climate model , meteorology , computer science , geography , geology , oceanography , artificial intelligence
Producing plausible outputs from statistically downscaled models of daily precipitation remains a significant challenge when producing climate scenarios for impact assessments. This challenge arises primarily as a consequence of a number of difficulties encountered when working with higher resolution daily precipitation data. Precipitation receipts at a site tend to be largely heterogeneous over space and time as local factors, such as relief, play an important role in determining whether it rains or not. More crucially from the point of view of statistical analysis, daily precipitation data is rarely, if ever, normally distributed, resulting from a high frequency occurrence of low‐fall events and a low frequency of high‐fall events. Additionally, modelling precipitation requires a two‐step procedure. First, precipitation occurrence must be modelled: then a model is fitted to precipitation quantities which describes the rainfall distribution for days on which precipitation occurs. This paper presents a technique that overcomes some of the difficulties encountered and produces plausible precipitation amounts for a selection of 14 sites in Ireland. Difficulties still exist with predicting extreme precipitation events, which tend to be underestimated by the methodology employed. Copyright © 2007 Royal Meteorological Society