Simeon Sandaramu Chiyenda's research while affiliated with University of British Columbia - Vancouver and other places

Publications (2)

Article
Results obtained by Kozak (A. Kozak. 1970. For. Chron. 46(5): 402–404.) concerning conditions for additivity of component biomass regression equations are formalized and extended. More specifically Kozak demonstrated, using multiple linear regression equations to model three biomass components (bole, bark, and crown) for individual trees, that corr...

Citations

... At first this problem may seem trivial because the component equations can be summed to produce an equation for total tree biomass. Indeed, if the component biomasses are inde- pendent and are linear functions of the same set of traits (e.g., are all specified by Eq. (7)), then the equation for total biomass can be derived by simply summing the component equations, and the corresponding confidence and prediction intervals can be derived (Kozak, 1970;Chiyenda, 1983;Chiyenda and Kozak, 1984). However, complexities arise because (i) the component equations may be functions of different sets of traits; (ii) the com- ponent biomasses are likely to be estimated from the same set of data on the same tree, so that the error terms are actually corre- lated (Parresol, 1999); and (iii) the component equations can be intrinsically non-linear. ...
... When modeling individual biomass components, it is worth considering the logical assumption that the sum of the component of the tree estimated using equations should be equal to the estimated biomass of the whole tree. This assumption can be met by seemingly unrelated regression application [32,33,48,49]. The assumption about the additivity of the biomass model system and the use of nonlinear models was the basis of elaboration of a consistent set of additive biomass functions for eight tree species and nine components in Germany [34]. ...