Skip to content

The project development aims to interpolate the seawater temperature, salinity three-dimensional structure, and to understand the physical oceanography in the turbulent region Gulf Stream by exploiting the latent regression model and deep regression neural networks.

License

Notifications You must be signed in to change notification settings

v18nguye/gulfstream-lrm

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

50 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Introduction

The ocean is a dynamical turbulent and chaotic system, so very complexly to interpret and predict. This is particularly true in the western regions of the mid-latitude oceans, where very large-scale currents narrow to intense fronts, meanders and numerous vortices with a width of around 100km. These regions are called "extensions of the west edge currents". Because they are turbulent, these regions are one of the main sources of uncertainty in assessing the role of the ocean in climate and their behavior under the influence of global climate change is still limited. These uncertainties can be removed if we improve our ability to diagnose the temporal evolution of the three-dimensional structure of the extensions of the west edge currents.

In our project, we introduced the Gaussian Mixture Model resolved by the classical Expectation-Maximization algorithm, and deep regression neural networks to characterize the Gulf Stream turbulence region. Our analysis were conducted to analyse the three dimensional distribution of the seawater temperature and salinity in the region. Moreover, we tried also to capture the seasonal variation of these ocean's characteristics, and to represent dominant currents exiting in the region, which have had great impact on physical oceanography in the region.

The project including several file units in the dev document.

  • data processing library
  • result evaluation library
  • model training library
  • data analysis notebook
  • result analysis notebook
  • trained model notebook
  • For more details about data, please contact me through van-khoa.nguyen@imt-atlantique.net.

About

The project development aims to interpolate the seawater temperature, salinity three-dimensional structure, and to understand the physical oceanography in the turbulent region Gulf Stream by exploiting the latent regression model and deep regression neural networks.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published