I have a data set showing vegetation change over a 100 year period. The data is categorical (ie. Conifer, Deciduous, Mixedwood, Grassland, Shrub, Open Canopy forest, non vegetated). Each cell in my data set represents a hectare. The data is discontinuous (I can only see 59% of the actual landscape, as some parts are hidden from view), and there is no data source to fill the holes.
I can account for all of the landscape change that is reversed succession (ie. anything that went from Coniferous to grassland) with ancillary data about known disturbances (fire, development, roads etc). I now want to understand what has driven the forward succession of some key transitions (ie. areas that were grassland in 1909 may now be Conifer, or Deciduous, etc). I have topographic data, solar insolation, natural subregions as predictors.
I am currently stuck trying to understand what analysis method I can use to model this relationship (ie. For Vegetation = Grassland in 1909, and for Vegetation = more advanced than grassland in 2008, what are the driving forces and correlates).
I am seeking ideas for analysis methods... preferably ArcGIS, but I can dig into other software if necessary