Its goodness of fit was 0.00164. In this study, furthermore, the two cosimulation
procedures used required that the parameters of the model be standardized so that the
sill value, that is, nugget plus structure parameter, equaled one unit (Figure 6.3b):
g h = 0:76 + 0:24 1:5
h
5:2
− 0:5
h
5:2
3
"
#
(6.5)
Its goodness of fit was 0.0032. The nugget and structure parameters were 0.76
and 0.24, respectively. These relatively large nugget parameters meant that in this
study the spatial autocorrelation of above-ground forest carbon was relatively weak.
FIGURE 6.3 Spatial autocorrelation g(h) of above-ground forest carbon modeled using
spherical model for (a) original data and (b) standardized data, where h is distance in km.
118
UPSCALING WITH CONDITIONAL COSIMULATION FOR MAPPING
procedures used required that the parameters of the model be standardized so that the
sill value, that is, nugget plus structure parameter, equaled one unit (Figure 6.3b):
g h = 0:76 + 0:24 1:5
h
5:2
− 0:5
h
5:2
3
"
#
(6.5)
Its goodness of fit was 0.0032. The nugget and structure parameters were 0.76
and 0.24, respectively. These relatively large nugget parameters meant that in this
study the spatial autocorrelation of above-ground forest carbon was relatively weak.
FIGURE 6.3 Spatial autocorrelation g(h) of above-ground forest carbon modeled using
spherical model for (a) original data and (b) standardized data, where h is distance in km.
118
UPSCALING WITH CONDITIONAL COSIMULATION FOR MAPPING