Request Info
Request Info

Fridays at 11:00 am, RSMAS Auditorium (unless stated otherwise)

Jan 10 (SLAB 103): Dr. Rui Ni
Johns Hopkins University, Baltimore, Maryland

Bubbles and Spray in Turbulence
Recording Available at COMPASS ON DEMAND

Air-sea interactions are often associated with strong turbulence and two-phase mixtures, including spray in the air and bubbles entrained in water. Although attempts have been made to simplify the two-phase flow problems with single-phase-fluid-like parameterizations, there is a fundamental difference between these two systems. In this talk, I will leverage some advanced diagnostic tools developed in fluid dynamics community to address some key two-phase flow problems. In particular, I will show a simple example to extract the momentum transfer between two phases and illustrate some key parameters that have been elusive in turbulent two-phase flow. The goal of developing these small-scale laboratory experiments is to compartmentalize the complex air-sea interaction problems into several simpler questions that will hopefully lead to better parameterizations and large-scale modelling.

Dr. Ni recently joined the Johns Hopkins University as Assistant Professor of Mechanical Engineering in 2018. Before this position, he was the endowed Kenneth K. Kuo Early Career Professor at Penn State since 2015. He received his Ph.D. in Physics in 2011 from the Chinese University of Hong Kong and worked as a postdoctoral scholar at Yale and Wesleyan University. He won the NSF CAREER award in fluid dynamics and ACS-PRF New Investigator Award in 2017. His primary research focus is the development of advanced experimental methods for understanding gas-liquid and gas-solid multiphase flow as well as two-phase heat transfer problem.


Manish Devana (MPO)
Rapid Entrainment-Forced Freshening of the Iceland Scotland Overflow
Manish Devana, William E. Johns, and Sijia Zou

The Iceland Scotland Overflow (ISOW) is a major component of the Atlantic Meridional Overturning Circulation's deep limbs. Newly available mooring observations from the Overturning in the Subpolar North Atlantic Program (OSNAP) show an abrupt decline in ISOW salinity. ISOW salinity, and its variability, is governed by the combination of two distinct pathways: convection in the Nordic Seas, and entrainment along the Iceland Faroe Ridge. Previous ISOW salinity anomalies have been attributed primarily to the convective pathway acting on decadal, and longer, timescales. However, we show that entrainment allowed an upper ocean anomaly to bypass the convective pathway to drive the overflow salinity decline. This is shown by tracking propagation of the upper ocean salinity anomaly into ISOW along the entrainment pathway. We tracked the anomaly using a combination of mooring and Argo observations, surface drifter trajectories, and the FLAME numerical model. The upper ocean segment of the pathway advected the anomaly in the North Atlantic Current to the Iceland Faroe Ridge and mixed downwards to depths of active entrainment. The total upper ocean advection time was ~6 months. After being entrained into the overflow, the anomaly took 1–1.5 years to flow southwards back to the OSNAP array in the ISOW layer. A 2-year transit time from the upper ocean into the ISOW layer was found, which is significantly faster than the convective pathway involved with ISOW formation. This shows that entrainment allows interannual to sub-decadal scale upper ocean variability to directly modify the abyssal ISOW.

Simge Bilgen (MPO)
Understanding the Delayed Warming of the Southern Ocean

Here, a fully coupled model run at multiple resolutions from coarse to eddy resolving, driven by observed historical and fixed CO2 concentration is used to investigate the delayed warming of the Southern Ocean (SO). We analyze the 1941-2014 SO sea surface temperature (SST) and ocean potential temperature for the first 1 km trends simulated in the coupled general circulation model and evaluate possible causes of the model's inability to reproduce the observed 1941-2014 SO cooling. We used NCAR's Community Climate System Model version 4 (CCSM4) at both eddy resolving (0.1 degree ocean model) and eddy parameterized resolutions (1 degree ocean model) to explore the mesoscale atmosphere-ocean interactions in the SO in a fully coupled regime and to understand the role of ocean dynamics in modulating temperature response. At both resolutions the models successfully reproduce the observed warming response for the northern flank of the Antarctic Circumpolar Current (ACC). The eddy resolving simulations are able to reproduce the observed SO cooling, however in the eddy parameterized simulations, the SO SST response is inconsistent with the observations for the south of the ACC. The results presented here provide support for the hypothesis that the cooling around the Antarctic is intimately connected with ocean meso-scale processes that cannot be captured by ocean eddy parameterized models typically used for IPCC simulations.

Jan 24: NO SEMINAR (Auditorium in use for Miami Climate Symposium)

Jan 31: John Lodise
Department of Ocean Sciences, RSMAS
(one-hour MPO student seminar)

Measuring Surface Ocean Currents Using Massive Arrays of CARTHE Drifters
Recording Available at COMPASS ON DEMAND

Very near surface currents are vital to the transport and aggregation of biogeochemical materials naturally found in the ocean, as well as the fate of buoyant pollutants, like oil and plastics. In this presentation we investigate and report on very near surface currents using the massive array of CARTHE drifters deployed during the LAgrangian Submesoscale ExpeRiment (LASER) that took place from January to March of 2016 in the Northern Gulf of Mexico. Surface currents are especially complex, due to the wide array of forcing mechanisms that drive the surface flow on many different spatial and temporal scales. Specifically, very near surface currents can be easily dominated by wind and wave forcing during moderate to severe winds associated with atmospheric fronts. However, surface currents under mild wind conditions are mainly forced by pressure gradient-driven flows set up by density fronts and varying stratification, which often develop into smaller scale, ageostrophic flows. Given the difficult task of studying these varying dynamics, the objectives of this work were to: (1) deconstruct near surface currents to isolate and describe the vertical structure of wind- and wave-driven surface flows under high wind conditions and (2) investigate the interactions between mesoscale and submesoscale structures in order to observe the processes involved with surface convergence and the vertical exchange of surface and interior waters. The major findings of the work, shed new light on the vertical structure of wind-driven currents through the use of drogued and undrogued drifters, as well as the connection between mesoscale and submesoscale flows, using a Gaussian process regression based interpolation method to calculate Eulerian estimates of divergence and relative vorticity from Lagrangian drifter data.

Feb 07: NO SEMINAR (Recruitment Weekend)


Matthew Grossi (MPO)
Predicting Particle Trajectories Using Artificial Neural Networks
Matthew D. Grossi1, Miroslav Kubat2, and Tamay M. Özgökmen1
1 University of Miami Rosenstiel School of Marine and Atmospheric Science, Miami, FL, USA
2 University of Miami College of Electrical and Computer Science, Miami, FL, USA

Artificial neural networks (ANNs) may be futuristic tools for predicting maritime oil dispersion, but only if they are capable of learning realistic particle trajectories in a turbulent ocean. We explore the predictability of 2D trajectories from a variety of flow regimes. After conducting proof-of-concept experiments consisting of simulated flows of increasing complexity, we generate realistic particle trajectories using modeled flow fields from a regional ocean general circulation model for the Gulf of Mexico. We choose as a test case of interacting scales of motion a mesoscale eddy surrounded by submesoscale dynamics. ANNs are developed to predict particles' future velocities based on their past observations. A rolling window training approach enables the ANNs to be continuously updated according to the most recent available data. ANNs are trained in two ways to predict future velocities: first, a so-called "one-to-one ANN" uses only a particle's most recently observed velocity as input, and second, a "time series ANN" uses the past 24 hours' worth of velocity observations. We compare ANN output to rudimentary persistence predictions within a 24-hour forecast window and find that, for realistic trajectories, one-to-one networks offer little to no improvement over persistence while time series ANN forecast errors are at least half those of persistence, implying that realistic trajectories do contain some inherent learnability. By always testing the simplest possible ANN, our networks have much room for further development and performance enhancement. Our results suggest that ANNs are a promising new data-driven approach to forecasting material transport in the ocean.

Kayla Besong (ATM)
Atmospheric Blocking, Forecast Model Resolution,
and Winter Weather Conditions in the U.S.

An atmospheric block is defined as a large-scale obstruction of zonal flow in the form of 500 mb quasi-stationary cyclones and anticyclones lasting a minimum of five days. Their persistent displacement of the jetstream coincides with a shift in storm tracks, influencing regional weather patterns, often in the form of temperature and precipitation extremes. With resulting impacts from extremes on human and natural systems at large, the significance in predictability of blocks is highlighted. Climate models are notorious for lack of skill in accurately capturing atmospheric blocking, primarily with strong underestimations of wintertime blocking frequencies over the North Atlantic basin. Suggestions to decrease model biases relating to blocking include increasing horizontal resolution and the use of a fully coupled ocean-atmosphere model. Therefore both 1.0º×1.0º and 0.5º×0.5º retrospective forecasts of the Community Climate System Model, version 4 (CCSM4) have been evaluated in their ability to capture January-March blocking frequency, duration, and consequential up- and downstream impacts on the mean flow with associated regional precipitation and temperature extremes. Duration of blocking events, mean blocked flow and resulting regional impacts were well represented by the model. However, blocking frequencies were poorly captured for both higher and lower versions of CCSM4, with a strong underestimation over the North Atlantic. Differences between resolutions are minimal for all analysis, suggesting that increasing horizontal resolution does not improve blocking frequency bias nor does it increase confidence in accurately predicting impacts caused by blocking events.

Tyler Fenske (ATM)
The Relationship Between the Pacific Decadal Oscillation
and the Atlantic Multi-Decadal Oscillation in a Multi-Ensemble

We explore the potential relationship between the Atlantic Multi-decadal Oscillation (AMO) and the Pacific Decadal Oscillation (PDO), the leading climate modes in their respective basins. Their drivers are generally not well understood; current leading theories suggest that the PDO is primarily driven by internal variability, while the AMO is primarily driven by externally forced variability. Both modes have downstream teleconnections that can affect weather patterns in North America and Europe. These teleconnections can also reach other ocean basins, thus allowing for the possibility of these modes being linked by their teleconnections. Our observational results suggest that statistically significant correlations exist between the two modes when the PDO leads by 14 years and the AMO leads by 24 years. Previous studies also find statistically significant correlations, but with shorter time lags where the AMO leads by 12 years and the PDO leads by 1 year. We further analyze this link with a novel approach by using CLIVAR's "ensemble of ensembles", a set of six large-ensemble climate model runs. These results show a nearly perfect negative correlation with no time lag between the two in the ensemble means for each model. Additionally, they all show a negative trend in the PDO and a positive trend in the AMO over the last 30 years. Both of these findings suggest that a forced connection between these modes may be present. Future work will determine whether this connection is indeed forced, as well as determining a mechanism for this connection.

Feb 21: NO SEMINAR (week of AGU Ocean Sciences)

Feb 28: Romain Chaput
Department of Ocean Sciences, RSMAS
(one-hour OCE student seminar)

Quantification of the Impact of Biological Uncertainties on Estimates of
Fish Population Connectivity in the Florida Keys

This work investigates the impacts of uncertain biological input parameters on the estimates of a biophysical model, the Connectivity Modeling System (CMS). These input parameters are used to model biological traits and behaviors; they are, however, often poorly known and characterized. The source of these uncertainties is the natural variability between individuals. In this study, to account for it, we characterized biological inputs with probability density functions informed by previous studies on our modeled species, Abudefduf saxatilisAfter propagation of the uncertainty to the CMS outputs, we build a surrogate to quantify the impact of the input biological traits. We find that behaviors and biological traits affect connectivity differently depending on distance of dispersal and release location. A sensitivity analysis shows that swimming speed and orientation accuracy are the mostly likely to influence the settlement abundance in the Florida Keys. Accuracy of estimates of marine connectivity is important for decision makers, whether it be for conservation, establishment of Marine Protected Areas, or fisheries management. The approach used in this study allows to quantify the individual and combined impact of the biological traits on the different metrics used to characterize the connectivity. The propagation of uncertainty through the model allows us to 1) determine the main contributors to the output uncertainties, 2) test hypothesis based on observations, and 3) guide the focus of future modeling effort toward including some biological uncertainty.


Kaycie Lanpher (OCE)

Rebecca Evans (MPO)

Shannon Doherty (OCE)

Mar 13: NO SEMINAR (Spring Recess)


Joseph Anderson (OCE)

Yueyang Lu (MPO)

Nektaria Ntaganou (MPO)

Mar 27: Tiago Bilo
Department of Ocean Sciences, RSMAS
(one-hour MPO student seminar)


Samantha Shawver (OCE)

Yu Gao (MPO)

Kelsey Malloy (ATM)


Wei-Ming Tsai (ATM)

Xingchen Yang (MPO)

Glorianne Rivera (OCE)


Haozhe He (ATM)

Luna Hiron (MPO)

Houraa Daher (OCE)


Di Sang (OCE)

Szandra Peter (MPO)

Hanjing Dai (AMP)