H. Madsen et D. Rosbjerg, GENERALIZED LEAST-SQUARES AND EMPIRICAL BAYES ESTIMATION IN REGIONAL PARTIAL DURATION SERIES INDEX-FLOOD MODELING, Water resources research, 33(4), 1997, pp. 771-781
A regional estimation procedure that combines the index-flood concept
with an empirical Bayes method for inferring regional information is i
ntroduced. The model is based on the partial duration series approach
with generalized Pareto (GP) distributed exceedances. The prior inform
ation of the model parameters is inferred from regional data using gen
eralized least squares (GLS) regression. Two different Bayesian T-year
event estimators are introduced: a linear estimator that requires onl
y some moments of the prior distributions to be specified and a parame
tric estimator that is based on specified families of prior distributi
ons. The regional method is applied to flood records from 48 New Zeala
nd catchments. In the case of a strongly heterogeneous intersite corre
lation structure, the GLS procedure provides a more efficient estimate
of the regional GP shape parameter as compared to the usually applied
weighted regional average. If intersite dependence is ignored, the un
certainty of the regional estimator may be seriously underestimated an
d erroneous conclusions with respect to regional homogeneity may be dr
awn. The GLS procedure is shown to provide a general framework for a r
eliable evaluation of parameter uncertainty as well as for an objectiv
e appraisal of regional homogeneity. A comparison of the two different
Bayesian T-year event estimators reveals that generally the simple li
near estimator is adequate.