Our research project is looking to do a classification of vegetation in a dynamic environment. The environment is an annually flooded floodplain and the dominant vegetation types are annual and perennial grasses. The aim is to create distribution maps of the different grasses at different temporal points during the flood cycle. We have some ground-truthed transect data but would like to extrapolate that to the whole floodplain. The goal is to describe changes in the quality and quantity of vegetation over the years for the whole floodplain and link these to changes in the flooding patters.
We are looking for any guidance on: (i) the methodological approach: (ii) the remote sensing data (we are currently using Landsat); and (iii) how to deal with the potential dynamics in the relationship between the annual growth cycle and spectral signature of the grasses in the floodplain.