Spatio-temporal modelling of biomass

Abstract

Environmental problems include erosion, salinisation, eutrophication, carbon allocation and rising C02 in the atmosphere. Environmental modelling, mapping, research and management are part of the solution to biophysical degradation. However, field data are usually limited and alternative data sources such as modelled or remotely sensed data must be calibrated. The resolutions between tht? different data sets must also be matched. Therefore there is a need to develop spatio-temporal models at an appropriate resolution to enhance limited field data. Such models need to be linked to the terrain surface (the spatial data) and incorporate climate (time varying) data. Preferably these models would maintain the integrity of source data (physical catchment attributes), have a predictive capacity and reflect catchment processes. In southern and eastern Australia catchments are mostly cleared, particularly those in low relief landscapes. These catchments have limited spatio-temporal vegetation data and therefore monitoring, research and management are constrained. Digital Elevation Models (DEM) can supply accurate spatial information about the terrain shape if appropriate source data, resolution and accurate interpolation methods are used. Hutchinson (1988) developed a locally adaptive algorithm which automatically calculates ridge and stream lines from points of locally maximum curvature on contour lines (chapter 2). Further developments by Hutchinson ( 1996) have provided a smoothing method, which has yielded useful error estimates for grid DEMs and a criterion for matching grid resolution to the information content of the source data. DEMs are essential input data for modelling terrain effects, which directly influence the surface conditions for plant growth. Climate is another dominant control over vegetative growth and climate data can also be limited. Climatic data can be modelled using interpolation methods developed by Hutchinson ( 1997). In this thesis, three approaches are developed to model the spatio-temporal distribution of biomass. These models are referred to as the Sub-catchment model, the Satellite model and the Topo-climate models. The Sub-catchment model calibrates the GROWEST model to biomass averaged over three separate sub-catchments (chapter 4). Combining catchment averaged climate data with disaggregated temperature and biomass GROWEST produced growth indices at each sub-catchment for 13 and 26 week growth accumulation periods. The 26 week growth accumulation period matched observed biomass data with greater accuracy than the 13 week period. The Satellite model simply calibrates biomass data with observed satellite data (chapter 3). Satellite data although spatially extensive requires atmospheric corrections and normalisation over time if direct comparisons are required. These models have limited predictive capacity, although they can be good for monitoring instantaneous catchment condition and structural features in the landscape. The third approach develops full spatio-temporal models, which simultaneously include effects of terrain (the spatial component) and climate (the temporal component) on biomass distribution (chapters 4, 5). The Topo-climate models are fitted using thin plate smoothing splines (Hutchinson 1999) (chapter 7). The Topo-climate models form a process based approach to spatio-temporal biomass modelling. They were successful in achieving spatio-temporal modelling of biomass in this catchment. They also have excellent predictive capacity, requiring only standard climate data. Model validation and statistical model comparisons were examined to determine the degree of parameterisation and accuracy of the different models. Model veracity is discussed and different applications for the various model types are suggested. Further research includes land management and research areas of vegetation modelling and carbon allocation. Predictive modelling of landscape processes such as the topo-climate models developed in this thesis, help to address environmental problems by providing spatio-temporal biomass data under varying climatic conditions for management and research purposes. Show more