Data input quality and resolution effects on regional and local scale Eucalyptus globulus productivity predictions in north-east Tasmania

To examine the effect of multi-scaled soils and climate data on a process-based productivity model, we apply the same forest productivity model (ProMod), at five resolutions. We compare point growth predictions using site information sampled in a representative 10 m×10 m area within 26 commercial Eucalyptus globulus plantations in north-east Tasmania with the corresponding measured site and growth data gathered at the same resolution. Five input data sets are generated at increasingly coarse grid scales from climate and soils databases (50, 100, 200, 500 and 1000 m). Our results indicate that forest productivity models like ProMod are not scale dependent and even with the use of broad-scale input data the model provides an unbiased estimate of the mean productivity for studies covering large geographic extents. However, our comparison using fine-scale data shows that sub-grid variance at broad scales can be very large, particularly for soil variables that can change abruptly over short distances. We conclude that although ProMod may be used over a range of scales with some confidence, specific applications will require different scales of data input and quality.