A comparison between the proposed Swedish spillway design floods and historic flood marks made at lake Siljan in central Sweden, is shown. Frequency analysis is performed incorporating pregauge information on water levels together with a sensitivity analysis of modelling assumptions.

A water level of 0.42 to 0.75 metres above the highest historic flood mark (166.10 m.a.sl., 1659) was obtained when routing the design spring flood through lake Siljan. The design autumn flood lifted the lake to 1.56 to 1.52 metres below the highest flood mark. Return period for the design spring and autumn flood was estimated to about 1,000 years. The uncertainty in frequency analysis proved to have larger impact than modelling assumptions on estimating the risk of the design flood.