and predicted flood quantiles were compared at gauged sites only, treated as un-
gauged, for average recurrence intervals T of 2, 5, 10, 20,
50 and 100 years. The performance of
the prediction was evaluated by cross-validation at independent sites not used in the method de-
velopment, by calculating the mean relative error (BIAST ) and relative root mean squared error
(RMSET ):
BIAST (%) =
1
N
N
/media/vedurstofan/utgafa/skyrslur/2014/VI_2014_001.pdf