Premium
On the risk of obtaining misleading results by pooling streamflow data for trend analyses
Author(s) -
Viviroli D.,
Schädler B.,
SchmockerFackel P.,
Weiler M.,
Seibert J.
Publication year - 2012
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/2011wr011690
Subject(s) - pooling , flooding (psychology) , flood myth , streamflow , environmental science , precipitation , econometrics , artifact (error) , statistics , computer science , geography , meteorology , mathematics , cartography , drainage basin , artificial intelligence , psychology , archaeology , computer vision , psychotherapist
Floods have broad impacts on nature, society, and the economy. The frequency and intensity of flood events are generally believed to increase with the anticipated changes in temperature and precipitation. Trend analyses are important tools to quantify these changes, but often, they provide inconclusive results, partly because of the limited data availability. One way to overcome this limitation is to pool data from different gauging stations. However, pooling data from different stations may lead to misleading results. For example, using pooled flood data Allamano et al. (2009a) found a considerable increase of flooding risks for Switzerland. Here we demonstrate that the previous finding of increased flooding risks was an artifact of the pooling of stations and the fact that the longer time series came from larger catchments, which tend to have lower values for specific peak flows than smaller catchments. Our results demonstrate the risk of obtaining incorrect statistical conclusions when statistical analyses and data selection are not considered with due care.