-
Direct flow mapping
- / Tobler 1987
- Long history
- Produces maps familiar to many users
- Doesn't scale well
-
Filtering can help
- / Tobler 1987
- / FlowMapLayout 2005 (selecting a dest or origin)
- Requires intelligent selection or dynamic interaction
- Difficult to provide a visual overview
-
OD Matrix
- Spatial properties are lost
- Better usability than node-link views (Ghoniem 2004)
- Except for 'path finding' and similar geo tasks
-
/ Marble 1997
- OD matrix as 3D surface: highly questionable
-
Sorting/reordering the matrix
- May make clusters more apparent
-
ways of
-
Clustering strongly connected nodes
- / Jarvis 1973
-
Revealing spatial clusters
- / Andrienko 2008, Guo 2006, 2007
-
Preserving spatial proximity relations
- / Marble 1997
-
Fails to preserve spatial configuration
-
to overcome
- Link to a map
- / Guo 2006,2007
-
Summarizing
-
Aggregation of flows of high density ("bundling")
- / Cui 2008
- / Holten 2009
- Spatial distortion may result in loss of detail
- The problem of long trajectories occluding short ones still retains
-
Flow density surfaces
- / Rae 2009
- Density does not indicate the density of origin and destination -> arbitrary patterns
- Suitability depends on the importance of the paths of the trajectories.
-
Regionalization
- / Guo 2009
-
Alternatives preserving some of the spatial props
-
Space-filling Morton ordering
- / Marble 1997
-
Aggregation into identifiable spatial regions
- / Andrienko 2008
- Limited in the ability to integrate other geo data
-
OD Map
- / Wood 2010
- Eliminates the cluttering=> Scalable
- Retains much of the geo space
- Aggregation => Potential loss of detail
- Problems with long/thin regions
- Aliasing when grid cells don't reflect geo structure