Reiner SchlitzerReiner Schlitzer

Alfred Wegener Institute for Polar and Marine ResearchAlfred Wegener Institute for Polar and Marine Research

ODV Gridding MethodsODV Gridding Methods

Data Visualization – Honest WayData Visualization – Honest Way

Data Visualization – Gridded FieldData Visualization – Gridded Field

Requirements: Requirements: • easy-to-use,easy-to-use, fast and „reliable“, fast and „reliable“, • usable for large datasets of >100000 pointsusable for large datasets of >100000 points

Grid Value Estimation by Weighted AveragingGrid Value Estimation by Weighted Averaging

i

ii

w

dwestimate

Gridding AlgorithmGridding Algorithm

ri ew

22yx LyLxr

Variable Resolution GridVariable Resolution Grid

Data Visualization (2)Data Visualization (2)

„„Honest“ WayHonest“ Way Gridded FieldGridded Field

Gridding MethodsGridding Methods

Gridding is …Gridding is …• important and definitely neededimportant and definitely needed• challenging mathematical problemchallenging mathematical problem• obtaining reliable fields is an „art“obtaining reliable fields is an „art“

Inverse distance weightingInverse distance weighting Pros: Pros: fast, good results for homogenous data fast, good results for homogenous data coveragecoverage

Cons: Cons: erodes extrema; poor results for sparse and erodes extrema; poor results for sparse and inhomogenous data coverageinhomogenous data coverage

Objective analysisObjective analysis Pros: Pros: optimal estimationoptimal estimation

Cons:Cons: very slow, requires knowledge of data very slow, requires knowledge of data statistics, „small“ datasets onlystatistics, „small“ datasets only

MethodsMethods

Variational data Variational data interpolation (DIVA)interpolation (DIVA)

Pros: Pros: quite fast,quite fast, optimal estimation, supports optimal estimation, supports domain separation, anisotropic statistics and domain separation, anisotropic statistics and rotated correlation ellipsesrotated correlation ellipses

DIVADIVA

•Developed at U LiegeDeveloped at U Liege

•2-Step procedure:2-Step procedure:(1) Triangular mesh generation on possibly (1) Triangular mesh generation on possibly complex domain(s)complex domain(s)

(2) Variational fitting to data and estimation (2) Variational fitting to data and estimation at arbitrary pointsat arbitrary points

•Supports:Supports:(1) Variable mesh resolution(1) Variable mesh resolution

(2) Multiple sub-domains(2) Multiple sub-domains

(3) Anisotropic statistics (3) Anisotropic statistics

http://modb.oce.ulg.ac.be/projects/SeaDataNet/DivaUserGuide2008.pdfhttp://modb.oce.ulg.ac.be/projects/SeaDataNet/DivaUserGuide2008.pdf

DIVA Integration in ODVDIVA Integration in ODV

ODV will automatically…ODV will automatically…

•create all files for DIVAcreate all files for DIVA

•run DIVA mesh generation and estimationrun DIVA mesh generation and estimation

•read and display the gridded fieldread and display the gridded field

Domain Example 1Domain Example 1

Domain Example 2Domain Example 2

East AtlanticEast AtlanticWest AtlanticWest Atlantic

Example (1) - Example (1) - Separating domains…Separating domains…

Example (2) - Example (2) - Maintaining extremes…Maintaining extremes…

Example (3) - Example (3) - Filling large gaps…Filling large gaps…

AvailabilityAvailability

http://odv.awi.dehttp://odv.awi.de

Download and install latest version of ODV 4.5. 0. Download and install latest version of ODV 4.5. 0.

DIVA is included.DIVA is included.

Available for Windows, Mac OS X, and Linux.Available for Windows, Mac OS X, and Linux.