Test of a system to predict productivity of eucalypt plantations in Tasmania

Summary A test was undertaken of a system which predicts eucalypt plantation wood volume yields in Tasmania from a set of eight topographic, climatic and soil variables measured at a site. The system was tested in 32 plots located in plantations of E. globulus and E. nitens in northern Tasmania. When the system predicted that a site had high productivity (maximum mean annual increment ≤20 m3ha−1yr−1), there was a fairly high probability (≈80%) that the site indeed had that productivity. However, when it predicted lower productivity, it incorrectly predicted a substantial proportion of sites (≈50%) as being of lower productivity when they were in fact of high productivity. It was shown that, on average, the system underestimated productivity by 7%. In practice, the system could be used to predict average productivity over a region within a range bounded by -4% to +18% of the prediction (95% prediction interval). The system had a similar predictive ability to two other systems developed in Australia to estimate plantation productivity.