A braided gravel-bed river provides essential habitats for various plants and animals. However, human regulation of rivers, such as dams, channelization, and other engineering works, alter flow and sediment regimes, which generally cause the degradation of river and riparian ecosystems. One of the prominent changes prevailing in Japanese rivers is forest expansion over gravel bars, and many native plants and animals that depend on gravel-bar habitat are now endangered. The Satsunai River is typical of rivers experiencing forest expansion, so the Japanese government launched a restoration project in 2012 to partially restore its riparian ecosystems. This is a large-scale experiment developed jointly by an interdisciplinary science team and river managers, who conduct monitoring and evaluation under an adaptive management scheme. The main measure to restore gravel bed habitat was an artificial flood regime, releasing a maximum water volume of 120 m3/s, which was a 2-year return period flood before dam construction. A unique feature of the project is that we considered the role of high-magnitude floods with recurrence intervals greater than 20 years after dam construction. We hypothesize that the artificial floods can be timed seasonally to create sites for regeneration and nesting of riparian species, and the high-magnitude floods contribute to maintaining a shifting mosaic structure of floodplain forest and unvegetated gravel-bar patches. We also used critical non-dimensional shear stress analysis to define “flood-disturbance areas” that can be disturbed under the artificial flow regime created by a dam. Artificial floods have been initiated once a year since 2012 at the end of June, synchronized with the seed dispersal period of Salix arbutifolia, which is endangered and a high conservation priority in the project. Thus far, the idea of setting floodplain-disturbance areas and the strategy of using both artificial and high-magnitude floods to restore a shifting mosaic of floodplain habitat patches is appropriate, as we found successful regeneration of S. arbutifolia and an exponential decay curve of the age distribution of floodplain forest patches. However, the sediment regime regulated by the dam was not addressed in this research, so future monitoring should track changes in river morphology associated with reduced sediment supply caused by the dam.