Table 1.
Summary statistics according to the system of conventional and linear moments.
Fig 1.
The relation of conventional skewness coefficient CS versus conventional variation coefficient CV for some two-parameter distributions commonly used if FFA plotted with the Polish data of 90-year annual peak flow series.
Distributions: Ga–gamma, We–Weibull, LN–log-normal, LL–log-logistic, LG–log-Gumbel, Exp–exponential.
Fig 2.
The relation of linear skewness coefficient LCS versus linear variation coefficient LCV for some two-parameter distributions commonly used if FFA plotted with the Polish data of 90-year annual peak flow series.
Distributions: Ga–gamma, We–Weibull, LN–log-normal, LL–log-logistic, LG–log-Gumbel, Exp–exponential.
Fig 3.
Map of 38 Polish gauging stations.
Table 2.
Origin and basic characteristics of 38 Polish gauging stations.
Fig 4.
The relation of conventional skewness coefficient CS versus conventional variation coefficient CV for two-parameter inverse Gaussian, IG, and generalized exponential, GE, distributions plotted with the Polish data of 90-year annual peak flow series.
Fig 5.
The relation of linear skewness coefficient LCS versus linear variation coefficient LCV for two-parameter inverse Gaussian, IG, and generalized exponential, GE, distributions plotted with the Polish data of 90-year annual peak flow series.
Table 3.
Basic characteristics of two-parameter IG and GE distributions.
Fig 6.
Probability density functions of GE and IG distributions for μ = 1.0 and selected values of CV and thus CS.
Fig 7.
Probability of correct selection [%] for competing GE and IG distributions by the K discrimination procedures.
Fig 8.
Probability of correct selection [%] for competing GE and IG distributions by the QK discrimination procedures.
Fig 9.
Probability of inconsistent selection [%] for competing GE and IG distributions by the K or QK discrimination procedures.
Fig 10.
Probability of correct selection [%] for competing GE and IG distributions by the KS discrimination procedure.
Fig 11.
Probability of correct selection [%] for competing GE and IG distributions by the R1 discrimination procedures.
Fig 12.
Probability of correct selection [%] for competing GE and IG distributions by the R2 discrimination procedures.
Fig 13.
Relative asymptotic bias [%] of from T = GE distribution, assuming F = IG model.
Fig 14.
Relative asymptotic bias [%] of from T = IG distribution, assuming F = GE model.
Table 4.
The 1% quantile estimates for selected gauging stations in Poland, assuming GE and IG distributions, respectively.
Table 5.
Distribution choice by the four discrimination procedures for annual maximum records of selected gauging stations.
Table 6.
The aggregation of 1% quantile of annual maximum flow series for selected gauging stations in Poland.