Premium
An evaluation of statistical models for downscaling precipitation and their ability to capture long‐term trends
Author(s) -
Benestad R. E.,
HanssenBauer I.,
Førland E. J.
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.1421
Subject(s) - downscaling , precipitation , climatology , environmental science , scale (ratio) , term (time) , statistical model , meteorology , statistics , mathematics , geography , geology , physics , quantum mechanics , cartography
Abstract Large‐scale changes in the sea‐level pressure do not necessary reflect changes in the atmospheric moisture budget, and hence may not give a good representation of changes in precipitation as a result of a global warming. Statistical models that use both sea‐level pressure and large‐scale precipitation as predictors are evaluated for a number of locations in Fennoscandia. The statistical models in most cases were capable of capturing 60–80% of the year‐to‐year seasonal variations in precipitation, and a correlation analysis over independent data indicated predictive correlation scores in the range 0.2–0.5. A comparison between statistical models based on large‐scale precipitation, sea‐level pressure, and a mixture of these, indicated similar skills in terms of variance and predictive skill of inter‐annual variations. Analyses of their ability to capture recent precipitation trends reveal potential problems regarding reconstructing long‐term changes in the past. One explanation for the statistical models not giving similar past trend values as given by the station observations may be partly because the precipitation trends during the most recent 50 years are not well defined since the interval is not sufficiently long. This is supported by the fact that trend analysis for station observations based on two different data products, and different trend analysis strategies, do not correspond well with each other. An analysis for possible non‐stationarities between large and local spatial scales does not indicate any significant presence of non‐stationarities. Copyright © 2006 Royal Meteorological Society