作者: W. K. Lauenroth , A. A. Wade , M. A. Williamson , B. E. Ross , S. Kumar
DOI: 10.1007/S10021-005-0072-Z
关键词: Statistics 、 Statistical error 、 Environmental science 、 Uncertainty analysis 、 Primary productivity 、 Estimation 、 Primary production 、 Data variability 、 Propagation of uncertainty 、 Monte Carlo method
摘要: Net primary production (NPP) is a fundamental characteristic of all ecosystems and foundational to understanding the fluxes energy nutrients. Because NPP cannot be measured directly, researchers use field-measured surrogates as input variables in various equations designed estimate ‘true NPP’. This has led considerable debate concerning which most accurately influenced efforts assess grasslands, with often advocating more complex avoid underestimation. However, this approach ignores increase statistical error associated estimates greater number parameters mathematical functions are introduced into equation. Using published grassland data Monte Carlo simulation techniques, we assessed relative variability obtained using six different estimation that varied both intricacy operations. Our results indicated may result uncertainty without reducing probability The amount was by well nature For example, due belowground than aboveground data, tended have NPP. An analysis were standardized allowed us isolate details calculations from assessing propagation uncertainty. made clear product terms potential magnify inputs although relationship complicated interactions parameters. suggest can necessarily risk never tested comparison “true NPP”, recommend include an assessment when evaluating ‘best’ method.