2014). A strong
correlation between a reconstruction of the NAO index and detrended relative salt marsh sea level
records as well as a positive correlation between Reykjavík tide gauge data and air pressure and
wind speed suggest that wind pattern variations connected to shifts in NAO control Icelandic sea
level variability. Long periods of a positive NAO index, that is with strong Icelandic
/media/vedurstofan-utgafa-2020/VI_2020_005.pdf
with the observed AMF series indicate that the two regions are ho-
mogeneous, although Region 1 could be slightly heterogeneous. When the H statistics are cal-
culated with the simulated AMF series within each catchment, the values are negative, indicating
the presence of positive cross-correlation between the different sites frequency distributions or
an excessive regularity in the data (Hoskings
/media/vedurstofan/utgafa/skyrslur/2014/VI_2014_001.pdf
part of Upper Tisza 3 4 3 10
Lower Rhine - Rivierenland, Netherlands 2 6 2 10
Fig. 2 Overview of the number of experts (per stakeholder group)
consulted in each case-study
6 More details on statistical analyses can be found in Huntjens et al.
(2007, updated 2009).
7 More details can be found in Huntjens et al. (2007, updated 2009).
8 A complete overview of correlation coefficients has been
/media/loftslag/Huntjens_etal-2010-Climate-change-adaptation-Reg_Env_Change.pdf
the correlation between the
different predictors:
MD(u) =
q
(X(t) X(u))TS 1(X(t) X(u)) (5)
14
where X(t) (o1(t); ::;o j(t); :::;ok(t)) and X(u) (o1(u); ::;o j(u); :::;ok(u)) denote the multivariate
feature vectors at times t and u and S the covariance matrix.
The N best analogues correspond to those N days with the lowest ED or MD score. The ED
score has traditionally been used with nearest-neighbor
/media/vedurstofan/utgafa/skyrslur/2013/VI_2013_008.pdf