The MIT Multidisciplinary Simulation, Estimation, and Assimilation Systems (MSEAS) group creates, develops and utilizes new mathematical models and computational methods for ocean predictions and dynamical diagnostics, for optimization and control of autonomous ocean observation systems, and for data assimilation and data-model comparisons. Our systems are used for basic and fundamental research and for realistic simulations and predictions in varied regions of the world's ocean, recently including monitoring (Lermusiaux et al., Physica-2007), naval exercises including real-time acoustic-ocean predictions (Xu et al., POMA-2008) and environmental management (Cossarini et al., JGR-2009).

• New free-surface 2-way nested primitive-equation ocean model (Haley and Lermusiaux, 2010). New open boundary conditions and tidal parameterizations. Vertical coordinate options include sigma, hybrid and multiple sigma coordinate transformations.
• Stochastic modeling component to represent sub-grid-scales (Lermusiaux, 2006)

• Three-dimensional Kalman update with specified error covariances (Optimal Interpolation, Lozano et al, 1996; Lermusiaux, MWR-1999).
• A nonlinear Bayesian-based scheme that predicts field error covariances, Error Subspace Statistical Estimation (ESSE, Lermusiaux, JAOT-2002, JCP-2006). ESSE includes: error subspace initialization, state and uncertainty prediction, minimum error variance data assimilation, adaptive error correction, smoothing, adaptive sampling, path planning and adaptive modeling using varied optimization tools (Lermusiaux, Physica-2007).
• New Dynamically-Orthogonal decomposition for uncertainty predictions using prognostic equations (Sapsis and Lermusiaux, 2009)
• Objective analysis schemes to map gappy ocean data in complex geometries based on the fast-marching method, numerical diffusion and scale estimation (Agarwal and Lermusiaux, 2010)

• barotropic tides from an inverse tidal model (Logutov, Oc. Dyn.-2008)
• nested data-assimilative barotropic tidal prediction system (Logutov and Lermusiaux, Oc. Model.-2008)

• Adaptable biogeochemical models (Besiktepe et al., JMS-2003; Tian, Eco. Model.-2006)
• Unstructured-grid biogeochemical ocean modeling based on high-order discontinuous Galerkin finite elements (Ueckermann and Lermusiaux, 2009).

• Model training via the two-fold problem of error estimation and subsequent model correction based on observational data (Logutov and Robinson, QJRMS-2005).
• Schemes that estimate the spatial distribution of errors in the individual modeling systems and fuse multi-model estimates based on Bayesian principles (Logutov, Oc. Dyn.-2008)
• Empirical neural-network calibration based on prior model misfits (Leslie et al., JMS-2008).

• Mixed Integer Linear Programming (MILP; Yilmaz et al, 2006, et al, 2008; Yilmaz and Lermusiaux, 2009),
• Genetic algorithms (Heaney et al, 2007) to solve the optimization problem.
• Combined predictive adaptive sampling and onboard adaptive routing for thermocline tracking and adaptive sampling for acoustic fields with AUVs (Wang, 2006; Wang et al., JMS-2009).

• Software for processing, analysis and management of data (Leslie, et al., 2009)

• Automated relocatable configuration and control of MSEAS codes: web-based software to configure and control multiple ocean modeling codes (Evangelinos et al., Oc. Model.-2006)
• Many-Task Computing for distributed ocean uncertainty predictions using Error Subspace Statistical Estimation (Evangelinos et al, 2009)

• NPS two-way coupled normal mode model (COUPLE, Chiu et al., JASA-1996) for 2D acoustic sections
• Range-dependent Parabolic Equation Acoustic Model (RAM, Collins, JASA-1989a,b)
• C-Snap, the one-way Coupled SACLANTCEN normal mode acoustic propagation loss model
• 3d Parabolic Equation method (FOR3D, Lee, et al., JASA-1992)
• High-resolution embedded 2-D and 3-D non-hydrostatic models developed for internal waves studies (refraction, radial spreading, scattering, dispersion)