A VEGETATION MODEL FOR THE SAYAN MOUNTAINS, SOUTHERN SIBERIA

A mountain vegetation model driven by regional climatic parameters was developed for predicting the geographic distribution of the numerous ecosystems of the Sayan Mountain system in southern Siberia. Meteorolo gical data are drawn from normal records of monthly means. Four climat ic variables were interpolated to a 10' by 10' grid: temperature, prec ipitation, vapor pressure, and cloudiness. In addition, albedo was der ived by 100-m elevation classes for both windward and leeward slopes a s a function of the general type of vegetation surface and the dates o f snow occurrence. Ecosystems were classified using a two-dimensional climatic ordination. The first dimension was growing degree-days above a 5 degrees C threshold, and the second was Budyko's dryness index, w hich is based on annual radiation balance and annual precipitation. Th e patterns: of predicted vegetation corresponded rather well to observ ed mapped vegetation. Overall kappa statistics indicated agreement wit h observed vegetation that varied from fair at the finest resolution ( 10' by 10' cells) to good at intermediate resolution (30' by 30' cells and 1 degrees by 1 degrees cells) to very good at the coarsest resolu tion (2 degrees by 2 degrees cells). Focusing on individual vegetation classes, agreement was good to very good for each vegetation class at intermediate resolutions and very good to excellent for each vegetati on class at the 2 degrees by 2 degrees resolution. Thus, the general l ocation, pattern, and overall distribution of these mountain ecosystem s was accurately predicted. This approach for predicting the spatial d istribution of biomes was successfully applied at both the global and subcontinental scale. The successful extension of Budyko's radiation b alance approach to mountainous terrain on a finer resolution regional scale indicates the generality of the approach.