For validation of the LUCAS output maps for simulating land-cover change in the Little Tennessee River basin, Scenarios 4 and 1 from Table 5 can be applied in succession (beginning with the year 1975) for three (5-year) time steps, i.e., Scenario 4 (1975--1986) for two time steps and Scenario 1 (1986--1991) for the final time step. The LUCAS-generated land-cover maps for each time step can then be compared with corresponding land-cover maps available from the GRASS database (see Section 2.3). Table 7 illustrates the excellent agreement in the land-cover proportions between the actual and LUCAS-generated maps. The land-cover proportions shown for LUCAS reflect the mean over 10 replicates, and the initial 1975 land-cover map for the LUCAS simulation was taken from the GRASS database (i.e., no differences the for initial year). For the year 1986, the LUCAS-generated map from the second time step or year 1985 (two 5-year time steps) is compared with the available 1986 (not 1985) land-cover map from the GRASS database. Similarly, the actual 1991 land-cover map from GRASS is compared with the 1990 (final time step) land-cover map generated by LUCAS.
Table 7: Proportions of land covers for simulated and actual
landscapes in the Little Tennessee River basin.