It's difficult to provide a single solution without knowing what the format of the data is. Here's some suggestions.
if you have XYZ data (or polar co-ordinate data that can be converted) then Didier Bur's excellent pointscloud ruby script (Didier's site is http://rhin.crai.archi.fr/RubyLibraryDepot/) will import spreadsheet data and provide some level of triangulation/skinning. I use it mostly for terrain/ DEM data for which it works well.
If you have the contour data for the illustrated diagrams in digital form, then TIG's suggestions are the way to go. You may need to tweak the interpolations.
If all you have is the graphs, you could import these as images, trace each and then scale as required and follow TiG's suggestion. Another program that is good at accurately digitising images of graphs is Engauge (http://digitizer.sourceforge.net/). The interface is not slick, but its very capable.
Hope that helps.
If not, SU may not be the right tool - you may want to look at dedicated data visualisation programs. I haven't used these much, but GGOBI (http://www.ggobi.org/) integrates with the R data/statistical environment (http://cran.r-project.org/), or OpenDX seems well regarded (http://www.opendx.org/). I know nothing about Axiis (http://www.axiis.org/) but it might be another option.