Multifidelity Computer Model Emulation with High-Dimensional Output: An Application to Storm Surge

Abstract Hurricane-driven storm surge is one of the most deadly and costly natural disasters, making precise quantification hazard great importance. Surge often performed through physics-based computer models surges. Such can be implemented with a wide range fidelity levels, computational burdens varying by several orders magnitude due to nature system. The threat posed makes greater highly desirable, however, such their high-volume output tend come at cost, which make detailed study coastal flood hazards prohibitive. These needs development an emulator combining high-dimensional from multiple complex different levels important. We propose parallel partial autoregressive cokriging model predict accurate surges in computationally efficient way over large spatial domain. This has capability predicting as accurately high-fidelity given any characteristics