Data Assimilation for MPAS

Data assimilation is a crucial part of MPAS development and testing.  Interfaces exist for both the Data Assimilation Research Testbed (DART) and for the Gridpoint Statistical Interpolation system (GSI), thus providing two distinct implementations of the Ensemble Kalman filter (via both DART and GSI) and an ensemble-variational solver (GSI) for MPAS.  We compare the implementations for the two systems and present results for extend periods of cycling data assimilation on both uniform- and variable-resolution meshes.