Derivation of Plantation Type Maps
The focus of the Australian Greenhouse Office (AGO) Land Cover Change program has been on the identification of the timing and extent of land cover change (to or from forest) as input to the carbon modelling process used to calculate emissions. In areas where new forest growth has been identified, the carbon modelling can be improved by using growth data specific to the new forest type. The forest categories of interest are hardwood and softwood plantations, environmental planting and native regrowth.
Existing available GIS data for forest / plantation class is of variable quality, inconsistent within and between states, irregularly updated and often at inappropriate spatial scales for integrating with the AGO Landsat forest cover mapping. Excellent data is held privately for some areas by forestry related businesses. Nationally consistent forest type classifications are simply not available at an appropriate spatial and temporal scale for use in the carbon modelling.
Research was undertaken to examine whether an operationally feasible methodology for labelling the type of new forest cover could be developed using the existing time series of Landsat imagery. At local scales, conventional classification of single date imagery can produce plantation type maps, however they do not extrapolate well across space and time.
This paper describes how the time series was analysed to investigate the spectral growth patterns of different plantation types to identify the strategy to best separate them from each other and from native regrowth. At different times during growth the different forest types are separable. A compositing technique, incorporating change flags from the Land Cover Change mapping, is used to integrate age information into a single classifier that is feasible to apply operationally to derive a national product. The derivation of the methodology is discussed and some results from the national mapping program are presented.