Weak and strong constraint data assimilation in the inverse Regional Ocean Modeling System (ROMS): Development and application for a baroclinic coastal upwelling system

We describe the development and preliminary application of the inverse Regional Ocean Modeling System (ROMS), a four dimensional variational (4DVAR) data assimilation system for high-resolution basin-wide and coastal oceanic flows. Inverse ROMS makes use of the recently developed perturbation tangent linear (TL), representer tangent linear (RP) and adjoint (AD) models to implement an indirect representer-based generalized inverse modeling system. This modeling framework is modular. The TL, RP and AD models are used as stand-alone sub-models within the Inverse Ocean Modeling (IOM) system described in [Chua, B.S., Bennett, A.F., 2001. An inverse ocean modeling system. Ocean Modell. 3, 137–165.]. The system allows the assimilation of a wide range of observation types and uses an iterative algorithm to solve nonlinear assimilation problems. The assimilation is performed either under the perfect model assumption (strong constraint) or by also allowing for errors in the model dynamics (weak constraints). For the weak constraint case the TL and RP models are modified to include additional forcing terms on the right hand side of the model equations. These terms are needed to account for errors in the model dynamics.Inverse ROMS is tested in a realistic 3D baroclinic upwelling system with complex bottom topography, characterized by strong mesoscale eddy variability. We assimilate synthetic data for upper ocean (0–450 m) temperatures and currents over a period of 10 days using both a high resolution and a spatially and temporally aliased sampling array. During the assimilation period the flow field undergoes substantial changes from the initial state. This allows the inverse solution to extract the dynamically active information from the synthetic observations and improve the trajectory of the model state beyond the assimilation window. Both the strong and weak constraint assimilation experiments show forecast skill greater than persistence and climatology during the 10–20 days after the last observation is assimilated.Further investigation in the functional form of the model error covariance and in the use of the representer tangent linear model may lead to improvement in the forecast skill.     

A singular evolutive extended Kalman filter to assimilate ocean color data in a coupled physical–biochemical model of the North Atlantic ocean

Within the European DIADEM project, a data assimilation system for coupled ocean circulation and marine ecosystem models has been implemented for the North Atlantic and the Nordic Seas. One objective of this project is to demonstrate the relevance of sophisticated methods to assimilate satellite data such as altimetry, surface temperature and ocean color, into realistic ocean models. In this paper, the singular evolutive extended Kalman (SEEK) filter, which is an advanced assimilation scheme where three-dimensional, multivariate error statistics are taken into account, is used to assimilate ocean color data into the biological component of the coupled system. The marine ecosystem model, derived from the FDM model [J. Mar. Res. 48 (1990) 591], includes 11 nitrogen and carbon compartments and describes the synthesis of organic matter in the euphotic zone, its consumption by animals of upper trophic levels, and the recycling of detritic material in the deep ocean. The circulation model coupled to the ecosystem is the Miami isopycnic coordinate ocean model (MICOM), which covers the Atlantic and the Arctic Oceans with an enhanced resolution in the North Atlantic basin. The model is forced with realistic ECMWF ocean/atmosphere fluxes, which permits to resolve the seasonal variability of the circulation and mixed layer properties. In the twin assimimation experiments reported here, the predictions of the coupled model are corrected every 10 days using pseudo-measurements of surface phytoplankton as a substitute to chlorophyll concentrations measured from space. The diagnostics of these experiments indicate that the assimilation is feasible with a reduced-order Kalman filter of small rank (of order 10) as long as a sufficiently good identification of the error structure is available. In addition, the control of non-observed quantities such as zooplankton and nitrate concentrations is made possible, owing to the multivariate nature of the analysis scheme. However, a too severe truncation of the error sub-space downgrades the propagation of surface information below the mixed layer. The reduction of the actual state vector to the surface layers is therefore investigated to improve the estimation process in the perspective of sea-viewing wide field-of-view sensor (SeaWiFS) data assimilation experiments.     

Cycling the representer algorithm for variational data assimilation with a nonlinear reduced gravity ocean model

The representer method was used by [Ngodock, H.E., Jacobs, G.A., Chen, M., 2006. The representer method, the ensemble Kalman filter and the ensemble Kalman smoother: a comparison study using a nonlinear reduced gravity ocean model. Ocean Modelling 12, 378–400] in a comparison study with the ensemble Kalman filter and smoother involving a 1.5 nonlinear reduced gravity idealized ocean model simulating the Loop Current (LC) and the Loop Current eddies (LCE) in the Gulf of Mexico. It was reported that the representer method was more accurate than its ensemble counterparts, yet it had difficulties fitting the data in the last month of the 4-month assimilation window when the data density was significantly decreased. The authors attributed this failure to increased advective nonlinearities in the presence of an eddy shedding causing the tangent linear model (TLM) to become inaccurate. In a separate study [Ngodock, H.E., Smith, S.R., Jacobs, G.A., 2007. Cycling the representer algorithm for variational data assimilation with the Lorenz attractor. Monthly Weather Review 135 (2), 373–386] applied the cycling representer algorithm to the Lorenz attractor and demonstrated that the cycling solution was able to accurately fit the data within each cycle and beyond the range of accuracy of the TLM, once adjustments were made in the early cycles, thus overcoming the difficulties of the non-cycling solution. The cycling algorithm is used here in assimilation experiments with the nonlinear reduced gravity model. It is shown that the cycling solution overcomes the difficulties encountered by the non-cycling solution due to a limited time range of accuracy of the TLM. Thus, for variational assimilation applications where the TLM accuracy is limited in time, the cycling representer becomes a very powerful and attractive alternative, given that its computational cost is significantly lower than that of the non-cycling algorithm.     

Preconditioned Optimizing Utility for Large-dimensional analyses (POpULar)

I present the derivation of the Preconditioned Optimizing Utility for Large-dimensional analyses (POpULar), which is developed for adopting a non-diagonal background error covariance matrix in nonlinear variational analyses (i.e., analyses employing a non-quadratic cost function). POpULar is based on the idea of a linear preconditioned conjugate gradient method widely adopted in ocean data assimilation systems. POpULar uses the background error covariance matrix as a preconditioner without any decomposition of the matrix. This preconditioning accelerates the convergence. Moreover, the inverse of the matrix is not required. POpULar therefore allows us easily to handle the correlations among deviations of control variables (i.e., the variables which will be analyzed) from their background in nonlinear problems. In order to demonstrate the usefulness of POpULar, we illustrate two effects which are often neglected in studies of ocean data assimilation before. One is the effect of correlations among the deviations of control variables in an adjoint analysis. The other is the nonlinear effect of sea surface dynamic height calculation required when sea surface height observation is employed in a three-dimensional ocean analysis. As the results, these effects are not so small to neglect.     

Fast dynamical spin-up of ocean general circulation models using Newton–Krylov methods

Numerical models of the ocean play an important role in efforts to understand past climate variability and predict future climate changes. In many studies, ocean models are driven by forcings that are either time-independent or vary periodically (seasonally) and it is often highly desirable or even essential to obtain equilibrium solutions of the model. Existing methods, based on the simple, expedient idea of integrating the model until the transients have died out, are too expensive to use routinely because the ocean takes several thousand years to equilibrate. Here, we present a novel approach for efficiently computing equilibrium solutions of ocean models. Our general approach is to formulate the problem as a large system of nonlinear algebraic equations to be solved with a class of methods known as matrix-free Newton–Krylov, a combination of Newton-type methods for superlinearly convergent solution of nonlinear equations, and Krylov subspace methods for solving the Newton correction equations. As an initial demonstration of the feasibility of this approach, we apply it to find the equilibrium solutions of a quasi-geostrophic ocean model for both steady forcing and seasonally-varying forcing. We show that the matrix-free Newton–Krylov method converges to the solutions obtained by direct time integration of the model, but at a computational cost that is between 10 and 100 times smaller than direct integration. A key advantage of our approach is that it can be applied to any existing time-stepping code, including ocean general circulation models and biogeochemical models. However, effective preconditioning of the linear equations to be solved during the Newton iteration remains a challenge.     

Cycling the representer algorithm for variational data assimilation with a nonlinear reduced gravity ocean model

The representer method was used by [Ngodock, H.E., Jacobs, G.A., Chen, M., 2006. The representer method, the ensemble Kalman filter and the ensemble Kalman smoother: a comparison study using a nonlinear reduced gravity ocean model. Ocean Modelling 12, 378–400] in a comparison study with the ensemble Kalman filter and smoother involving a 1.5 nonlinear reduced gravity idealized ocean model simulating the Loop Current (LC) and the Loop Current eddies (LCE) in the Gulf of Mexico. It was reported that the representer method was more accurate than its ensemble counterparts, yet it had difficulties fitting the data in the last month of the 4-month assimilation window when the data density was significantly decreased. The authors attributed this failure to increased advective nonlinearities in the presence of an eddy shedding causing the tangent linear model (TLM) to become inaccurate. In a separate study [Ngodock, H.E., Smith, S.R., Jacobs, G.A., 2007. Cycling the representer algorithm for variational data assimilation with the Lorenz attractor. Monthly Weather Review 135 (2), 373–386] applied the cycling representer algorithm to the Lorenz attractor and demonstrated that the cycling solution was able to accurately fit the data within each cycle and beyond the range of accuracy of the TLM, once adjustments were made in the early cycles, thus overcoming the difficulties of the non-cycling solution. The cycling algorithm is used here in assimilation experiments with the nonlinear reduced gravity model. It is shown that the cycling solution overcomes the difficulties encountered by the non-cycling solution due to a limited time range of accuracy of the TLM. Thus, for variational assimilation applications where the TLM accuracy is limited in time, the cycling representer becomes a very powerful and attractive alternative, given that its computational cost is significantly lower than that of the non-cycling algorithm.     

The impact of mean dynamic topography on a sea-level anomaly assimilation in the South China Sea based on an eddy-resolving model

The sea-level anomaly (SLA) from a satellite altimeter has a high accuracy and can be used to improve ocean state estimation by assimilation techniques. However, the lack of an accurate mean dynamic topography (MDT) is still a bothersome issue in an ocean data assimilation. The previous studies showed that the errors in MDT have significant impacts on assimilation results, especially on the time-mean components of ocean states and on the time variant parts of states via nonlinear ocean dynamics. The temporal-spatial differences of three MDTs and their impacts on the SLA analysis are focused on in the South China Sea (SCS). The theoretical analysis shows that even for linear models, the errors in MDT have impacts on the SLA analysis using a sequential data assimilation scheme. Assimilation experiments, based on EnOI scheme and HYCOM, with three MDTs from July 2003 to June 2004 also show that the SLA assimilation is very sensitive to the choice of different MDTs in the SCS with obvious differences between the experimental results and observations in the centre of the SCS and in the vicinity of the Philippine Islands. A new MDT for assimilation of SLA data in the SCS was proposed. The results from the assimilation experiment with this new MDT show a marked reduction (increase) in the RMSEs (correlation coefficient) between the experimental and observed SLA. Furthermore, the subsurface temperature field is also improved with this new MDT in the SCS.     

Euler–Lagrange equations for the spectral element shallow water system

We present the derivation of the discrete Euler–Lagrange equations for an inverse spectral element ocean model based on the shallow water equations. We show that the discrete Euler–Lagrange equations can be obtained from the continuous Euler–Lagrange equations by using a correct combination of the weak and the strong forms of derivatives in the Galerkin integrals, and by changing the order with which elemental assembly and mass averaging are applied in the forward and in the adjoint systems. Our derivation can be extended to obtain an adjoint for any Galerkin finite element and spectral element system.We begin the derivations using a linear wave equation in one dimension. We then apply our technique to a two-dimensional shallow water ocean model and test it on a classic double-gyre problem. The spectral element forward and adjoint ocean models can be used in a variety of inverse applications, ranging from traditional data assimilation and parameter estimation, to the less traditional model sensitivity and stability analyses, and ensemble prediction. Here the Euler–Lagrange equations are solved by an indirect representer algorithm.     

Fast spin up of Ocean biogeochemical models using matrix-free Newton–Krylov

A novel computational approach is introduced for the efficient computation of equilibrium solutions of seasonally forced ocean biogeochemical models. The essential idea is to formulate the problem as a large system of nonlinear algebraic equations to be solved with a class of methods known as matrix-free Newton–Krylov (MFNK). MFNK is a combination of Newton-type methods for superlinearly convergent solution of nonlinear equations, and Krylov subspace methods for solving the Newton correction equations. The basic link between the two methods is the Jacobian-vector product, which may be probed approximately without forming and storing the elements of the true Jacobian. To render this approach practical for global models with O(10⁶) degrees of freedom, a flexible preconditioning strategy is developed. The result is an essentially “black-box” numerical scheme than can be applied to most existing biogeochemical models. The method is illustrated by applying it to find the equilibrium solutions of two realistic biogeochemical problems. Compared with the conventional approach of direct time integration, the preconditioned-MFNK scheme is shown to be roughly two orders of magnitude more efficient. Several potential refinements of the basic algorithm that may yield further performance gains are discussed. The numerical scheme described here addresses a fundamental challenge to using ocean biogeochemical models more effectively.     

Effectiveness of data assimilation using probabilistic techniques in a model for synoptic velocity field evolution

The results of numerical experiments involving data assimilation by linear and non-linear models for synoptic ocean dynamics are presented. This paper considers the model's response to data assimilation at diverse values of the relative error in the determination of the initial fields and space-time discreteness of data assimilation. The data obtained using the optimal filtering are compared with those provided by the optimal interpolation (OI). A qualitative difference in the response of the linear and non-linear models to data assimilation is noted. The limiting values of the time-space discreteness of observations assimilation are determined, which allow correct application of OI.Translated by Vladimir A. Puchkin.     

Bifurcation analysis of wind-driven flows with MOM4

In this paper, the methodology of bifurcation analysis is applied to the explicit time-stepping ocean model MOM4 using a Jacobian–Free Newton–Krylov (JFNK) approach. We in detail present the implementation of the JFNK method in MOM4 but restrict the preconditioning technique to the case for which the density distribution is prescribed. For a prescribed density field case, we present bifurcation diagrams, for the first time in MOM4, for the wind-driven ocean circulation. In addition, we show that the JFNK method can reduce the spin-up time to a steady equilibrium in MOM4 considerably if an accurate solution is required.     

Investigating transport pathways in the ocean

The ocean is a very complex medium with scales of motion that range from thousands of kilometers to the dissipation scales. Transport by ocean currents plays an important role in many practical applications ranging from climatic problems to coastal management and accident mitigation at sea. Understanding transport is challenging because of the chaotic nature of particle motion. In the last decade, new methods have been put forth to improve our understanding of transport. Powerful tools are provided by dynamical system theory, that allow the identification of the barriers to transport and their time variability for a given flow. A shortcoming of this approach, though, is that it is based on the assumption that the velocity field is known with good accuracy, which is not always the case in practical applications. Improving model performance in terms of transport can be addressed using another important methodology that has been recently developed, namely the assimilation of Lagrangian data provided by floating buoys. The two methodologies are technically different but in many ways complementary. In this paper, we review examples of applications of both methodologies performed by the authors in the last few years, considering flows at different scales and in various ocean basins. The results are among the very first examples of applications of the methodologies to the real ocean including testing with Lagrangian in-situ data. The results are discussed in the general framework of the extended fields related to these methodologies, pointing out to open questions and potential for improvements, with an outlook toward future strategies.     

A diagnostic stabilized finite-element ocean circulation model

A stabilized finite-element (FE) algorithm for the solution of oceanic large scale circulation equations and optimization of the solutions is presented. Pseudo-residual-free bubble function (RFBF) stabilization technique is utilized to enforce robustness of the numerics and override limitations imposed by the Babuška–Brezzi condition on the choice of functional spaces. The numerical scheme is formulated on an unstructured tetrahedral 3d grid in velocity–pressure variables defined as piecewise linear continuous functions. The model is equipped with a standard variational data assimilation scheme, capable to perform optimization of the solutions with respect to open lateral boundary conditions and external forcing imposed at the ocean surface. We demonstrate the model performance in applications to idealized and realistic basin-scale flows. Using the adjoint method, the code is tested against a synthetic climatological data set for the South Atlantic ocean which includes hydrology, fluxes at the ocean surface and satellite altimetry. The optimized solution proves to be consistent with all these data sets, fitting them within the error bars.The presented diagnostic tool retains the advantages of existing FE ocean circulation models and in addition (1) improves resolution of the bottom boundary layer due to employment of the 3d tetrahedral elements; (2) enforces numerical robustness through utilization of the RFBF stabilization, and (3) provides an opportunity to optimize the solutions by means of 3d variational data assimilation. Numerical efficiency of the code makes this a desirable tool for dynamically constrained analyses of large datasets.     

Assimilation of hydrological observation data for calculating currents in seas and oceans

This paper considers the main steps in improving the methods for calculating the ocean (sea) dynamics on the basis of observational data on sea-water temperature and salinity. The results of diagnostic and adaptation calculations for the near-equatorial area of the West Atlantic in the area of the Lomonosov countercurrent formation are presented. We consider the problem of the complex use of measurements of temperature, salinity, and current velocity in the POLYMODE polygons with their assimilation into the model using a Kalman filter. The results of calculations of the coordinated fields with the mechanism of geostrophic adaptation and using asynchronous measurements obtained by the Razrezy program are given. We discuss further modifications of the assimilation algorithms for hydrological observation data in models of sea dynamics and the principles of adaptation of hydrophysical fields that made it possible to reconstruct the climate fields of the Black Sea and to reproduce the basin dynamics for 23 years.     

The Regional Ocean Modeling System (ROMS) 4-dimensional variational data assimilation systems: Part I - System overview and formulation

The Regional Ocean Modeling System (ROMS) is one of the few community ocean general circulation models for which a 4-dimensional variational data assimilation (4D-Var) capability has been developed. The ROMS 4D-Var capability is unique in that three variants of 4D-Var are supported: a primal formulation of incremental strong constraint 4D-Var (I4D-Var), a dual formulation based on a physical-space statistical analysis system (4D-PSAS), and a dual formulation representer-based variant of 4D-Var (R4D-Var). In each case, ROMS is used in conjunction with available observations to identify a best estimate of the ocean circulation based on a set of a priori hypotheses about errors in the initial conditions, boundary conditions, surface forcing, and errors in the model in the case of 4D-PSAS and R4D-Var. In the primal formulation of I4D-Var the search for the best circulation estimate is performed in the full space of the model control vector, while for the dual formulations of 4D-PSAS and R4D-Var only the sub-space of linear functions of the model state vector spanned by the observations (i.e. the dual space) is searched. In oceanographic applications, the number of observations is typically much less than the dimension of the model control vector, so there are clear advantages to limiting the search to the space spanned by the observations. In the case of 4D-PSAS and R4D-Var, the strong constraint assumption (i.e. that the model is error free) can be relaxed leading to the so-called weak constraint formulation. This paper describes the three aforementioned variants of 4D-Var as they are implemented in ROMS. Critical components that are common to each approach are conjugate gradient descent, preconditioning, and error covariance models, which are also described. Finally, several powerful 4D-Var diagnostic tools are discussed, namely computation of posterior errors, eigenvector analysis of the posterior error covariance, observation impact, and observation sensitivity.     

Ocean current estimation using a Multi-Model Ensemble Kalman Filter during the Grand Lagrangian Deployment experiment (GLAD)

In the summer and fall of 2012, during the GLAD experiment in the Gulf of Mexico, the Consortium for Advanced Research on Transport of Hydrocarbon in the Environment (CARTHE) used several ocean models to assist the deployment of more than 300 surface drifters. The Navy Coastal Ocean Model (NCOM) at 1 km and 3 km resolutions, the US Navy operational NCOM at 3 km resolution (AMSEAS), and two versions of the Hybrid Coordinates Ocean Model (HYCOM) set at 4 km were running daily and delivering 72-h range forecasts. They all assimilated remote sensing and local profile data but they were not assimilating the drifter’s observations. This work presents a non-intrusive methodology named Multi-Model Ensemble Kalman Filter that allows assimilating the local drifter data into such a set of models, to produce improved ocean currents forecasts. The filter is to be used when several modeling systems or ensembles are available and/or observations are not entirely handled by the operational data assimilation process. It allows using generic in situ measurements over short time windows to improve the predictability of local ocean dynamics and associated high-resolution parameters of interest for which a forward model exists (e.g. oil spill plumes). Results can be used for operational applications or to derive enhanced background fields for other data assimilation systems, thus providing an expedite method to non-intrusively assimilate local observations of variables with complex operators. Results for the GLAD experiment show the method can improve water velocity predictions along the observed drifter trajectories, hence enhancing the skills of the models to predict individual trajectories.     

Assimilating satellite SST/SSH and in-situ T/S profiles with the Localized Weighted Ensemble Kalman Filter

The Localized Weighted Ensemble Kalman Filter(LWEnKF) is a new nonlinear/non-Gaussian data assimilation(DA) method that can effectively alleviate the filter degradation problem faced by particle filtering, and it has great prospects for applications in geophysical models. In terms of operational applications, along-track sea surface height(AT-SSH), swath sea surface temperature(S-SST) and in-situ temperature and salinity(T/S) profiles are assimilated using the LWEnKF in the northern South China ...     

Coupling a two-way nested primitive equation model and a statistical SST predictor of the Ligurian Sea via data assimilation

A primitive equation model and a statistical predictor are coupled by data assimilation in order to combine the strength of both approaches. In this work, the system of two-way nested models centred in the Ligurian Sea and the satellite-based ocean forecasting (SOFT) system predicting the sea surface temperature (SST) are used. The data assimilation scheme is a simplified reduced order Kalman filter based on a constant error space. The assimilation of predicted SST improves the forecast of the hydrodynamic model compared to the forecast obtained by assimilating past SST observations used by the statistical predictor. This study shows that the SST of the SOFT predictor can be used to correct atmospheric heat fluxes. Traditionally this is done by relaxing the model SST towards the climatological SST. Therefore, the assimilation of SOFT SST and climatological SST are also compared.     

A reduced-dynamics variational approach for the assimilation of altimeter data into eddy-resolving ocean models

A new method of assimilating sea surface height (SSH) data into ocean models is introduced and tested. Many features observable by satellite altimetry are approximated by the first baroclinic mode over much of the ocean, especially in the lower (but non-equatorial) and mid latitude regions. Based on this dynamical trait, a reduced-dynamics adjoint technique is developed and implemented with a three-dimensional model using vertical normal mode decomposition. To reduce the complexity of the variational data assimilation problem, the adjoint equations are based on a one-active-layer reduced-gravity model, which approximates the first baroclinic mode, as opposed to the full three-dimensional model equations. The reduced dimensionality of the adjoint model leads to lower computational cost than a traditional variational data assimilation algorithm. The technique is applicable to regions of the ocean where the SSH variability is dominated by the first baroclinic mode. The adjustment of the first baroclinic mode model fields dynamically transfers the SSH information to the deep ocean layers. The technique is developed in a modular fashion that can be readily implemented with many three-dimensional ocean models. For this study, the method is tested with the Navy Coastal Ocean Model (NCOM) configured to simulate the Gulf of Mexico.     

A spectral barotropic model of the wind-driven world ocean

A global spectral barotropic ocean model is introduced to describe the depth-averaged flow. The equations are based on vorticity and divergence (instead of horizontal momentum); continents exert a nearly infinite drag on the fluid. The coding follows that of spectral atmospheric general circulation models using triangular truncation and implicit time integration to provide a first step for seamless coupling to spectral atmospheric global circulation models and an efficient method for filtering of ocean wave dynamics. Five experiments demonstrate the model performance: (i) Bounded by an idealized basin geometry and driven by a zonally uniform wind stress, the ocean circulation shows close similarity with Munk’s analytical solution. (ii) With a real land–sea mask the model is capable of reproducing the spin-up, location and magnitudes of depth-averaged barotropic ocean currents. (iii) The ocean wave-dynamics of equatorial waves, excited by a height perturbation at the equator, shows wave dispersion and reflection at eastern and western coastal boundaries. (iv) The model reproduces propagation times of observed surface gravity waves in the Pacific with real bathymetry. (v) Advection of tracers can be simulated reasonably by the spectral method or a semi-Langrangian transport scheme. This spectral barotropic model may serve as a first step towards an intermediate complexity spectral atmosphere–ocean model for studying atmosphere–ocean interactions in idealized setups and long term climate variability beyond millennia.