P.J. Haley, Jr., W.G. Leslie,
O. Logutov, Jinshan Xu,
Arpit Agarwal, Lisa Burton, Themis Sapsis, Matt Ueckermann
Massachusetts Institute of Technology
This research is concerned with the fundamental understanding and modeling of complex physical, acoustical and biogeochemical oceanic dynamics and processes. New mathematical models and computational methods are created, developed and utilized for: i) ocean predictions and dynamical diagnostics, ii) data assimilation and data-model comparisons, and, iii) optimization and control of autonomous ocean observation systems. The regional dynamics involves interactions of sub-mesoscale and mesoscale ocean processes in the littoral as well as effects from large-scale processes in ocean basins. Such interactions and feedbacks with scales smaller and larger than the mesoscale need be better quantified. The technical approach is rooted in the comparison and optimal combination of measurements and models via nonlinear data assimilation (DA), including the development of adaptive modeling and adaptive sampling schemes based on Error Subspace Statistical Estimation. Our research group is updating and renewing our previous approaches and computational schemes and systems. We will keep and modernize the strengths of our methods and codes, but we will also progressively utilize other ocean dynamical models, or parts thereof, and explore novel numerical systems.
The research topics that are specific to the present effort include:
Uncertainty, data assimilation and dynamics in a coastal ecosystem: the Lagoon of Venice.ESSE was used to investigate the seasonal ecosystem dynamics of the Lagoon of Venice in 2001, combining a rich data set with a physical-biogeochemical numerical estuary-coastal model (Cossarini et al, 2009). Novel stochastic modeling components were developed to represent uncertainties in the internal ecosystem dynamics model, measurement model and boundary forcing by rivers, open-sea inlets and industrial discharges. The formulation and parameters of these new additive and multiplicative stochastic error models were optimized based on data-model forecast misfits. The sensitivity to initial and boundary conditions was quantified and analyzed. Half-decay characteristic times were estimated for key ecosystem variables and their spatial and temporal variability studied. The new error models were used in the ESSE scheme for ensemble uncertainty predictions and data assimilation, and an optimal ensemble dimension was estimated. The seasonal biogeochemical-ecosystem fields and their uncertainties were estimated using ESSE and used to guide local environmental policies.
The uncertainty analyses show that boundary forcing and internal mixing have a significant control on the seasonal dynamics of the Lagoon of Venice and that data assimilation is needed to reduce their prior uncertainties. Overall, higher uncertainties are predicted in the central and northern regions of the Lagoon. Based on the dominant singular vectors of the ESSE ensemble, the two major northern rivers are the biggest sources of DIN uncertainty in the Lagoon. Other boundary sources such as the southern rivers and industrial discharges can dominate uncertainty modes on certain months. For DIP and phytoplankton, dominant modes are also linked to external boundaries, but internal dynamics effects are more significant than for DIN. Our ESSE estimates of the seasonal biogeochemical fields and their uncertainties in 2001 cover the whole Lagoon and provide the means to describe the ecosystem and guide local environmental policies. Specifically, our findings and results based on these fields include the: temporal and spatial variability of nutrient and plankton gradients in the Lagoon; dynamical connections among ecosystem fields and their variability; strengths, gradients and mechanisms of the plankton blooms in late-spring, summer and fall; uncertainties of the field predictions, their monthly reductions by data assimilation and thus a quantification of data impacts and data needs; and, finally, an assessment of the water quality in the Lagoon in light of the local environmental legislation.
|Evolution of the a posteriori fields for DIN [mg/l] (maps),
and observation (overlaid colored circles) for the Lagoon of Venice in 2001.
Color scale is set to 0 to 1 mg/l for all months.
|Percentage of the year during which the Lagoon of Venice is above the limit of the Water Quality Target for DIN.
(A) data, (B) OA estimate, (C) ESSE estimate.
Multi-grid data assimilation into regional tidal models:A new method has been developed for nested data assimilation in barotropic tidal models resolving the topographic and coastal features (Logutov and Lermusiaux, 2008; Logutov, 2008). The method is designed to reduce representativeness errors by fitting the resolved dynamics to data consistently, across a multi-grid computational system. The set of control parameters are presently chosen to be the OBCs of the outer domain. The uncertainty from the outer domain OBCs is propagated to model tidal fields at observation locations through the set of nested domains using efficient low-rank error covariance representations. An analysis increment for these outer OBCs is computed to optimally steer the multi-grid system towards observations by minimizing the (weighted) variance of the observation-minus-forecast residuals.
Our new methodology based on nested domains in complicated coastal regions allowed fitting the full local tidal data only where it made sense, i.e. where the resolution of the model nests was sufficient to fully resolve the tidal dynamics. The method (Logutov and Lermusiaux, 2008, Logutov, 2008) can avoid artificial steering of the solution towards unresolved observations which is, in general, degrades accuracy. The presence of representativeness errors in data-model misfits was detected through sensitivity experiments with model resolution. In some cases, the observation-minus-forecast residuals were found to be highly sensitive to resolution. For example, in the Phillipines region, a high-resolution (1-minute resolution) nested domain was setup around the Sulu, Bohol, Visayan, and Sibuyan seas, where representativeness errors were found, and the assimilation of ADCP and Topex/Poseidon data in those areas was carried out using the nested computation.
Ocean dynamics, modeling and assimilation: New nesting schemes and open-boundary conditions for nested free-surface primitive-equation ocean models were implemented. New data assimilation schemes for barotropic tidal estimates were published (Logutov and Lermusiaux, 2008). Contributions were made to our Monterey Bay research (Haley et al, 2009; Ramp et al, 2009). A manuscript on the verification and training of models for real-time forecasting was published (Leslie et al, 2008). A special edition of Ocean Dynamics on Multi-Scale modeling: nested grid and unstructured mesh approaches was edited with a refereed editorial (Deleersnijder and Lermusiaux, Guest Eds, 2008).
Acoustic predictions. Coupled ocean-acoustic fields were forecast at sea in real-time (Lam et al, 2009). A manuscript on acoustically focused adaptive sampling and onboard routing for marine rapid environmental assessment was completed (Wang et al, 2008). The spatial and temporal variations in acoustic propagation during the PLUSNet07 Exercise in Dabob Bay was described (Xu et al, 2008).
Adaptive sampling.A manuscript on the quantitative planning of the paths of AUVs using Mixed Integer Linear Programming (MILP) was published (Yilmaz et al, 2008). The second part of this work, which selects the ocean sampling paths based on the ESSE-MILP forecasts of data impacts, is finalized (Yilmaz and Lermusiaux, in prep). Schemes for adaptive sampling based on genetic algorithms were evaluated based on Observation System Simulation Experiments (manuscript in prep).
Return to top of document
Return to the MSEAS home page