sub header2

[View PDF]

 

Han-Wei Shen, OSU

Application

  • Climate modeling by Dr. Ruby Leung and Dr. Samson Hagos at PNNL

Simulation Goal: Understand MJO

  • A complex cloud system in multiple scales over the Indian and Pacific oceans
  • An important weather phenomenon related to the tropical intra-seasonal change

Requirement: Time-varying Feature Tracking

  • Visualizing the movement of MJO-related quantities (eg. cloud population and precipitation)
    • Understand how MJO is formed
    • Understand how MJO is related to different physical quantities
  • Feature identification to filter the major cloud movement of MJO

Challenges

  • Large data size generated by high resolution simulations
  • Various temporal scales of MJO
    • make traditional techniques that perform per time step analysis and visualization ineffective
    • require global analysis over time other than local feature tracking

Cloud rendering on a virtual globe

Teh average water vapor mixing rate, indicating the existence of two MJO cycles

(Top): Cloud rendering on a virtual globe
(Lower): The average water vapor mixing rate,
indicating the existence of two MJO cycles

 


Time-Varying Data Analysis with Time Activity Curves


 

Technology

  • Robust global time-varying feature descriptors
    • Model the time-varying feature based on the change of value over time at each location
  • Visualize complex time-varying phenomena with detailed multivariate  temporal trend analysis

Result/Impact

  • Flexible interface to allow scientists to interact with data at their desktop
  • Replace tedious viewing of animations with more efficient visualization of spatio-temporal features in still images
  • Assist in summarizing and verifying major temporal trends such as MJO in extreme scale data sets
TAC-based (Time Activity Curve) region clustering

Clustering TACs from all 2D location

(Top): TAC-based (Time Activity Curve) region clustering
(Bottom): Clustering TACs from all 2D location.

Shen.OSU.Ana.MJO.climate-fig5

Linked views between TAC-based overview and virtual globe.
(Top Left): TACs per longitude as Hovmoller diagram.
(Top Right) Cloud animation with Feature Tracking