作者: Steen Magnussen , Oswaldo Ismael Carillo Negrete
DOI: 10.1186/S13021-015-0031-8
关键词: Errors-in-variables models 、 Population 、 Covariance matrix 、 Biomass 、 Sample size determination 、 Bioinformatics 、 Tree (data structure) 、 Unit-weighted regression 、 Linear regression 、 Statistics 、 Mathematics
摘要: Biomass and carbon estimation has become a priority in national regional forest inventories. of individual trees is estimated using biomass equations. A covariance matrix for the parameters equation needed computation an estimate model error tree level biomass. Unfortunately, many equations do not provide key statistics direct errors. This study proposes three new procedures recovering missing from available estimates coefficient determination sample size. They are complementary to recently published computationally intensive Monte Carlo approach. Our recovery approach use survey data population targeted Examples Germany Mexico illustrate validate methods. Applications with robust recovered fit gave reasonable errors It good practice uncertainty any model-dependent above ground When impossible due statistics, proposed procedure first step practice. recommended offers protection against inflated precision.