Nonstationary regional flood inundation projection under climate change: a case study of Athabasca River Basin, Canada

Abstract:

To reveal regional flood-inundation variations and impacts under changes in extreme climatic conditions, a nonstationary regional flood inundation projection method (NSRFIP) is developed in this study. The method consists of four progressive modules: (1) projection of precipitation extremes under representative concentration pathway (RCP) scenarios through bias-corrected statistical downscaling of general circulation models (GCMs) (i.e., FGOALS, ESM2M, and CM5AMR); (2) projection of floods through nonstationary flood frequency analysis (i.e., generalised additive models) taking antecedent extreme precipitation as covariates; (3) projection of flood inundation maps through high-resolution digital elevation mapping (20 m), lumped hydrodynamic modeling (i.e., Height Above Nearest Drainage), and remote-sensing-based model verification; (4) assessment of flood-inundation impacts on social economies under climate change. NSRFIP is applied to a representative catchment (i.e., Pembina River Watershed) in Athabasca River Basin, Canada. Results reveal higher impacts of GCMs than RCP scenarios on extreme precipitation projections. Variation of floods with RCP scenarios is associated with the selection of GCMs, and timing characteristics of design floods significantly vary with RCP scenarios. NSRFIP is effective at high-resolution regional flood inundation modeling over the catchment. Flood-inundation timing significantly varies with GCMs and RCP scenarios, while flooding areas do not to a relative extent. Flooding would seriously threaten socio-economic security of the catchment; for instance, approximately 28% of current buildings might be flooded by the maximum daily precipitation in 2080 under the CM5AMR-RCP4.5 modeling scenario. This study provides a scientific decision support tool for strategic flood adaptation and mitigation at large scales and high resolutions in the changing environment.