Data Assimilation for Ocean Drift Trajectories Using Massive Ensembles and GPUs
Conference object
Accepted version
View/
Date
2020-06-10Metadata
Show full item recordCollections
Original version
Holm HH, Sætra ML, Brodtkorb A: Data Assimilation for Ocean Drift Trajectories Using Massive Ensembles and GPUs. In: Klöfkorn R, Keilegavlen E, Radu FA, Fuhrmann J. Finite Volumes for Complex Applications IX - Methods, Theoretical Aspects, Examples, 2020. Springer p. 715-724 https://doi.org/10.1007/978-3-030-43651-3_68Abstract
In this work, we perform fully nonlinear data assimilation of ocean drift trajectories using multiple GPUs. We use an ensemble of up to 10000 members and the sequential importance resampling algorithm to assimilate observations of drift trajectories into the underlying shallow-water simulation model. Our results show an improved drift trajectory forecast using data assimilation for a complex and realistic simulation scenario, and the implementation exhibits good weak and strong scaling.