Visualization

1
Visualization

• Polly Baker
• Randy Heiland
• Rob Stein
• Ed Bachta
• Steve Pietrowicz
• Visualization and Interactive Spaces Lab,
• and Scientific Data Analysis Lab,
• Pervasive Technology Labs at Indiana University
• June 3, 2004

Visualization and Interactive Spaces
Lab Pervasive Tech Labs at Indiana University
Visualization and Interactive Spaces
Lab Pervasive Tech Labs at Indiana University
2
Activity Report

• Last year
• 1D Viewer Java applet
• DTD definition
• Covariance visualization
• VisPort vis tool
• Last night
• Covariance visualization (revisited)

3
1D Viewer Java Applet

• For IOM instructional web

4
DTD Definition

• Motivated by last years discussion re saving
  output from GUI XML or netCDF?
• An XML snippet
• lttitlegtIOMlt/titlegtltbodygtIOM is a frameworklt/bodygt
• DTD (document type definition) is formal
  definition of an XML file
• Many communities defining community-wide XML (and
  documenting that in a DTD) as an answer to data
  dictionary issues, or for capturing run
  parameters, metadata, etc.

5
DTD Definition

• Formalizes and documents GUI outputs and inputs
  
• A step towards commonality, interoperability
  between GUI and Vis

6
DTD Snippets

• lt!ELEMENT IOM_output (RunDescription,
  SimulationModel, CovarianceInfo, IterationInfo,
  SimulationOutput) gt

• Element IOM_output
• Children
• RunDescription (Required)
• SimulationModel (Required)
• CovarianceInfo (Required)
• IterationInfo (Required)
• SimulationOutput (Required)
• Attributes
• NONE
• Description This is the top-level node in the
  data description. It defines the document
  structure by linking to a number of lower level
  elements described below.

7
DTD Snippets

• lt!ELEMENT RunDescription EMPTY gt
• lt!ATTLIST RunDescription
• Author CDATA REQUIRED
• DateTime CDATA REQUIRED
• Platform CDATA IMPLIEDgt

• Element RunDescription
• Children
• NONE
• Attributes
• Author (Required)
• DateTime (Required)
• Platform (Required)
• Description General Meta-data for the run.

8
DTD Snippets

• lt!ELEMENT SimulationOutput ( Fields?,
• Residuals?,
• Statistics?,
• PriorErrorCov?,
• PostErrorCov?,
• ArrayAssesment? ) gt

Element FieldsChildren OutputDimension,Backgrou
ndField (O11-1), FirstGuessField (O11-2),
BestEstimateField (O11-3), ClippedBestEstimateFiel
d (O11-4), BestEstimateParams (O12)Attributes
NONE Description This is the root node for all
the field output data. The actual data is
contained in external files of some defined
format (netCDF, HDF, vtk, ascii, binary, etc...)
You may include 0 or 1 of any of the child tags.
The child tags here are tagged against the IOM
DSL design document.
9
DTD Definition

• DTD is defined just once
• Serves as the shared understanding of
  terminology, roles of various objects,
  relationships among objects
• Could serve as communication mechanism between
  GUI and Visualization tools

10
Covariance Visualization

• Data from Boon (from IPEZ?)
• Temperature field 5 days, 100 samples
• Until Thursday, 400 p.m. Choose randomly
  selected groups of samples, average them, and
  visualize (using 20 samples, or 40, or 60, etc.)

11
Covariance Vis Tests
12
Covariance Vis Tests
13
K-L Decomposition
14
Covariance Visualization

• Thursday, by 500 p.m., after conversation with
  Boon
• Use the samples to compute the covariance (oh, of
  course) and figure out how to visualize
• Newly enlightened a wireless Internet
  connection

15
VisPort

• A framework and portal for access to
  community-specific visualization capability

16
VisPort Motivation

• Easy-to-use vis capability
• Access visualization capability remotely
• Handle large data sets
• Encapsulate and reuse vis code
• Customize capabilities for community needs
• Add informatics layer to visualization

17
VisPort Architecture

• Multi-tier architecture, using web services and
  protocols (XML, XML-RPC, etc.)

Information management
Visualization services
Data services
Middleware
User interfaces
18
VisPort Architecture

• One possible scenario

Information management
Visualization services
Data services
NCAR
Middleware
PHP, XML, SOAP, etc.
Your desktop browser
User interfaces
Java applets
19
Status

• V.1 back-end databases defined and implemented
• Back-end services for some visualization types
• Developers interface in place
• Users interface exists, but doesnt connect
  dynamically to databases and services

20
Covariance Visualization

• Revisited

21
Error Covariance

• Andrews book, p.72, Eq (3.2.50)
• Cu(xn,tl,xp,tq)1/K S vknl vkpq
• Samples v, K of them

K
K1
IUPUI
22
As a function

• Julia offered a functional form
• C(x,y,t,s)C0exp(-(x-y)2/L2)exp(-t-s/tau)
• and gave us the actual 4-D cov data.
• Take slices and obtain 3-D surfaces.

23
(No Transcript)
24
Computing Cov from samples

• (x,y,z)(172,90,30)
• t 5-day inversion
• 100 samples
• ? CTT(172,90,30,5, 172,90,30,5)
• Equatorial Pacific ? surface layer of interest

i.e., 500 files of size
25
A reduced dataset

• Originally CTT(172,90,30,5, 172,90,30,5)
• Fix a day, fix z (surface)
• Subsample surface data

26
Consider a spherical cow

• Actually, make that linear

(3.2.50) becomes
T11
T21
T12
T22
C(2,2)?
T1100
T2100