MPAS-Ocean and the AOT Algorithm: A Novel Data Assimilation Technique Applied to Ocean Models

Elizabeth Carlson1, Luke Van Roekel2, Humberto C Godinez3 and Mark R Petersen3, (1)California Institute of Technology, Pasadena, United States, (2)Los Alamos National Laboratory, Los Alamos, United States, (3)Los Alamos National Laboratory, Los Alamos, NM, United States
Abstract:
One of the challenges to accurately simulate geophysical flows is that observational data is often lacking. Data assimilation can improve simulation fidelity by continually incorporating the observed data into the model. Recently a new approach to data assimilation known as the Azouani-Olson-Titi (AOT) algorithm introduced a feedback control term to the 2D incompressible Navier-Stokes equations (NSE) in order to incorporate sparse measurements. With observations taken from a reference solution to the 2D incompressible NSE, the AOT algorithm was shown to converge exponentially fast to the reference solution (Azouani et al., 2014). Subsequently the approach was expanded in various papers both theoretically and in a computationally ideal framework to other dissipative systems, including the primitive equations (Pei 2018). To the best of our knowledge, only one study of this algorithm utilizing real-world data has been implemented, in this case for atmospheric models (Desamsetti et al. 2019). We present results of the implementation of the AOT algorithm within MPAS-Ocean in both idealized and realistic settings. In the idealized setting, proxy observations are taken from a higher resolution MPAS-Ocean model evolved with an initial condition based on observations and then inserted into a companion simulation of the MPAS-Ocean model using the AOT algorithm, with the objective of recovering the reference simulation. In the realistic setting, we test the AOT algorithm in a climatological forcing/restoring configuration, with real data taken from drifters, and with regards to mean state climate biases.