Flood hazard model calibration using multiresolution model output

Riverine floods pose a considerable risk to many communities. Improving flood hazard projections has the potential inform design and implementation of management strategies. Current are uncertain, especially due uncertain model parameters. Calibration methods use observations quantify parameter uncertainty. With limited computational resources, researchers typically calibrate models using either relatively few expensive runs at high spatial resolutions or cheaper lower resolutions. This leads an open question: is it possible effectively combine information from low resolution runs? We propose Bayesian emulation–calibration approach that assimilates outputs multiple As case study for riverine community in Pennsylvania, we demonstrate our LISFLOOD-FP model. The multiresolution results improved inference over single scenarios. Results vary based on values number available runs. Our method general can be used other dimensional computer improve projections.