Detelina P. Ivanova

The Fidelity of Ocean Models With Explicit Eddies

Current practices within the oceanographic community have been reviewed with regard to the use of metrics to assess the realism of the upper-ocean circulation, ventilation processes diagnosed by time-evolving mixed layer depth and mode water formation, and eddy heat fluxes in large-scale fine resolution ocean model simulations. We have striven to understand the fidelity of these simulations in the context of their potential use in future fine-resolution coupled climate system studies. A variety of methodologies are used to assess the veracity of the numerical simulations. Sea surface height variability and the location of western boundary current paths from altimetry have been used routinely as basic indicators of fine-resolution model performance. Drifters and floats have also been used to provide pseudo-Eulerian measures of the mean and variability of surface and sub-surface flows, while statistical comparisons of observed and simulated means have been carried out using James tests. Probability density functions have been used to assess the Gaussian nature of the observed and simulated flows. Length and time scales have been calculated in both Eulerian and Lagrangian frameworks from altimetry and drifters, respectively. Concise measures of multiple model performance have been obtained from Taylor diagrams. The time-evolution of the mixed layer depth at monitoring stations has been compared with simulated time series. Finally, eddy heat fluxes are compared to climatological inferences.

Scientific Computing and Big Data Analytics: Application in Climate Science

The nature of scientific and engineering problems of real world is often very complex. These problems are intrinsically multi-dimensional, multivariate, nonlinear and non-stationary in their dynamics. Solutions to these problems often necessitate the use of complex mathematical modeling, simulation and analysis which are traditionally achieved by the use of expensive high performance computing, more commonly known as Super-Computers. In this chapter we discuss the nature of problems involved in Scientific Computations, a type of use cases under the genre of Big Data Analytics, that require Super-Computers to solve, and further discuss the possibilities of addressing the same using Cloud based Distributed Computing technologies. We discuss the same using the example of Climate Analytics, which represents typical challenges of Scientific Computing to a considerable extent. In particular, we delve into the details of how significantly large-sized data from the output of a complex fluid dynamics based Earth’s Climate Model can be processed using Distributed Technology framework, Spark, in an integrated manner with the final analytics results accessed by an web application for the end users.

Fine-Resolution Global Sea-Ice/Ocean Modeling and Data Assimilation

The focus of our recent work has been to continue global simulations using the Los Alamos Parallel Ocean Program (POP) model and the sea ice model (CICE), as well as ensemble runs using the Simple Ocean Data Assimilation (SODA) and POP (SODA POP). The primary goal of the former effort is to simulate fine resolution (0.1deg) global ice fields using CICE, obtain measures of their veracity, and then move forward with a global coupled 0.1deg POP/CICE simulation. Sensitivity runs using the less computationally expensive global coupled 0.4deg POP/CICE simulation have also been conducted. In the latter effort ensemble runs with SODA-POP are being used to evaluate the relative impact of using different surface boundary conditions and assimilating different data

Towards a high-resolution global coupled navy prediction system

The focus of our recent work has been to complete the global high-resolution ocean and the eddy-permitting coupled ice/ocean simulations using the Los Alamos Parallel Ocean Program (POP) model and the sea ice model (CICE), as well as the first reanalysis using the Simple Ocean Data Assimilation (SODA) and POP (SODA POP 1.2). The high-resolution (0.1degree) global ocean model was run for 25 years (1979-2003) forced with synoptic atmospheric fluxes. The simulated ocean state for the post spin-up period (1994–2002) is examined in terms of its representation of the eddy variability and significant mesoscale processes by comparing energy levels and intrinsic scales with those from satellite altimetry. Also the role of the mesoscale in inter-basin property exchanges is examined with respect to the Indonesian Throughflow (ITF) and the Greenland- Iceland-Norwegian (GIN) Sea. Global coupled seaice/ ocean and stand-alone ice simulations on an eddypermitting 0.4 degree grid were run for the period 1979-2003; both were forced with synoptic atmospheric fluxes. The ice states have been compared statistically with available observations (passive microwave satellite fields, upwardlooking sonar, and ice buoys) in the latter part of the simulation. The results have lead to a suite of sensitivity runs, whose results will guide our final choices for the 0.1° global coupled sea-ice/ocean simulation. The spinup of global stand-alone 0.1 degree CICE is underway. The first reanalysis (SODA POP 1.2) using the 0.4 degree global POP and SODA is complete and is available to the general oceanographic community through Live Access Servers (LAS) at the University of Maryland and at the University of Hawaii. The reanalysis is being used to study various aspects of the global circulation including the Indonesian Throughflow, and the variability of the subtropical cells in the Pacific.