We simulate vegetation distribution using frames that represent alternative states of upland vegetation (up- land tundra, white spruce forest, broad-leaved deciduous forest, and dry grassland) found in subarctic Alaska (Figure 1). The model consists of a suite of individual frame submodels. Within each submodel frame, those factors (biotic and abiotic) that may be responsible for a switch to another ecosystem type (frame) are simulated. When a switch in ecosystem type occurs, the new frame model is activated. The ecosystem types were chosen as the simplest possible representation of the complex vegetation mosaic occupying uplands in the circumpolar arctic and boreal zones (Solomon 1992), and ignore the substantial variation in species composition within these and other intermediate vegetation types (Payette 1992; Starfield and Chapin 1996). The model operates on a 10-yr time step and a spatial scale of 2×2 k m g r i d cells; each cell has eight immediate neighbors (the ‘queen’ option for cellular automata). The time step is the average frequency of severe fire years in the North American borealforest (Flanniganand Harring- ton 1988) and allows replicated modeling of vegetation change over time scales of decades to centuries (NRC 1994; Starfield and Chapin 1996). The spatial scale is appropriate for interfacing with mesoscale climate models (Starfield and Chapin 1996). We deliberately chose temporal and spatial scales that are larger than those used to simulate the behavior of in- dividual fires, because our primary objective was to simulate the long-term average changes in regional vegetation rather than responses to specific fires. Our model is calibrated to provide realistic values of fire number and area burned for particular combinations of vegetation and climate at the temporal and spatial scales used in ALFRESCO. The model is a step for- ward toward including disturbance in dynamic global and regional vegetation models that operate at a coarse scale(0.5◦). These coarse-scale models cannot include realistic topographic effects on fire probability and spread (Kittel et al. in press).
Maps of climate, initial vegetation type, and topography (elevation) can be hypothesized or input into the model from data sets in a geographic information system (GIS) database. In our model, elevation data are used to identify barriers to migration (seed dispersal) and disturbance (fire spread) (Hadley 1994). Topography exerts a strong influence on vegetation Distribution and successional trajectory at northern latitudes, through its effects on microclimate (Van Cleve et al. 1991, 1996). The probability of high moose browse pressure is a model input that influences rate of succession (Pastor and Naiman 1992; Kielland 1997) and hence fuel loads. In this study, we assume that moose browse probability is identical for each grid cell in the landscape.
All model simulations were conducted on virtual landscapes (240×240 km or 42×20 km) that represent a typical Alaskan subarctic forest-tundra ecotone. The driving variables in ALFRESCO are cli- mate (growing-season temperature and precipitation), disturbance, and seed dispersal (Figure 1). These variables, each discussed below, then determine tree establishment, succession, and other processes controlling landscape-level vegetation change.
Quoted from : A frame-based spatially explicit model of subarctic vegetation response to climatic change: comparison with a point model