Modelling the perfect pine plantation
DEVELOPING A software model that predicts the wood quality properties (such as wood density, stiffness and growth) for how a tree will grow under a unique but changing set of environmental conditions is a complex undertaking. Dr David Drew and his research team from CSIRO and Dr Geoff Downes (now with Forest Quality Pty. Ltd.) have spent the last three years building a model of tree growth in Pinus radiata which provides explicit predictions of wood property variation based both on extremely fine level data (such as how wood cells divide and grow) and larger scale environmental inputs such as rainfall, temperature, soil quality and— perhaps most importantly to plantation managers— silviculture management practices.
With support from Forest and Wood Products Australia ( FWPA), Scion, HVP and ForestrySA, David and the research team have created ‘e- Cambium’, a processbased model that takes into account the biological knowledge of how a tree uses water and nutrients to build wood cells when affected by different climate and environmental conditions. It is designed primarily as a tool to let forest managers predict tree stem growth as well as wood density (and stiffness) responses under a range of conditions. The great value of such a process-based model is that the user can explore beyond existing data and field experience to predict stand growth responses and tree performance under future conditions for which they may not be any precedent (e.g. increasing average temperatures, or a new silvicultural intervention). The model can incorporate outputs from two existing tree-growth models: the more complex CABALA model, which gives very detailed results, or the simpler to run and use 3PG model (standing for Physiological Principles Predicting Growth), which gives slightly less accurate modeling results.
“This model is valuable in that, with the length of time it takes to grow a tree, if you do make a mistake early on you don’t have much of a chance to rectify it,” says David. “Therefore some sort of scenario-based exploration tool becomes very useful. If the tool says a particular silviculture management is the worst thing you could ever do then it provides a basis for reevaluating a given approach and exploring alternatives. That’s where this model has the potential to be a real dollar saver.”
The team measured climate data and tree growth at six sites: four in Australia and two in New Zealand. At the four Australian sites about 40 trees (aged 16 to 18 years old) were measured every day for two and a half years using automated dendrometers. These precision tools were attached to each tree to measure the rate the tree grew in 15 minute intervals, while also recording climate conditions. The data was sent wirelessly back to the research team, which was probably just as well as at one site the dendrometer cables were chewed by wombats resulting in data not being recorded. At another site the solar panel that recharged the battery of a dendrometer was damaged by a speeding kangaroo!
At the end of the two and a half year monitoring period pith-to- bark core samples were taken from each tree at, or as close as possible to, where the dendrometer had been measuring growth. These core samples were analysed using SilviScan technology (developed by Dr Robert Evans) to measure wood properties such as density. This allowed David to match wood properties with daily patterns of growth and environmental impacts (such as temperature and rainfall).
While e- Cambium may have been based on pre- existing theoretical models of tree growth it evolved significantly as the volume of incredibly detailed data grew.
‘The model started off highly theoretical but as the data came in, it gave us the opportunity to refine things and improve the quality of the framework. The model changed more than I expected as the data started to stream in, and there were some major changes we had to make to our earlier theoretical concepts to accommodate it, but that was a very good thing. It was exactly what we wanted from all that data!’ says David. ‘It’s great that the forestry industry appreciates that the modeling process is a backwards–forwards process.’
The e- Cambium model, with its algorithms derived from data from the six measurement sites, was calibrated against wood property data from another 10 sites. Excitingly, the model was able to predict around 80 per cent (using CABALA inputs) of the variability seen in wood density in trees from these sites, and around 70 per cent accuracy using 3PG. Although predicting final stand volume wasn’t the priority of the project, David says ‘The indications are strong that the model would have a similar high level of accuracy in predicting wood volumes from a site.’
‘In the “better/ faster/ cheaper” paradigm for business this research is helping build up the “better”,’ says David. ‘The stage we’re at now is that we’ve built a testable model that industry can experiment with and help modify. Currently HVP are experimenting with the model and getting reasonable results. They can see the real potential for it in their future decision making.’
The current e- Cambium model is available to all FWPA members for testing using their pine plantations and to explore the effects of different sites, silvicultural regime and weather conditions on both tree growth and wood properties. Although e- Cambium is not a fully operational model yet, David hopes that industry will embrace such process-models to get the best from Australia’s timber resource. ‘For industry to be willing to take this to the next level they need to be convinced of its commercial value,’ says David. “Perhaps one of the great advantages of these models is they help growers explore the major implications of decisions so that problems can be avoided in the future.”
Dendrometer on Radiata Pine.
Dendrometer data collecting.