Our department is currently (1997–1999) involved in an extensive research project that aims at developing an integrated system for forest harvesting. The project is led by the acting professor of statistics, Dr. Erkki Liski.
The developing of the system has become possible due to a computer-controlled forest harvester, which measures the stems and stores the data. Since the computers of the harvester can be connected to a central system, two-way communication is possible: all data collected by the harvester can be transferred to the central system, which in turn can give instructions to the harvester.
The first concrete task of the research project is to optimize the prediction-based marking for bucking of tree stems by a forest harvester. The optimal marking for bucking of a stem means converting the stem into logs in such a way that the price of the stem is maximized. The price of the whole stem is the sum of the single log prices, and the single log price is usually determined by the log length and the top diameter of the log. The optimization can be done by dynamic programming, for instance, provided that the whole stem is known.
Since it is too slow a procedure to measure the whole stem in advance, it is sensible to replace the unknown part of the stem curve with a prediction and calculate the optimal cutting points on the basis of the prediction. The prediction methods are based on statistical models. When calculating the prediction the harvester also utilizes data from the previous stems. Then it makes the first cutting decision on grounds of the prediction and the portion measured.
The second task of the project is to develop the cutting of tree stems so that it meets the demand for different types of logs. If only the price is optimized, this may lead to an undesired log distribution. It is thus necessary that the distribution has to fulfill certain conditions with respect to the length, diameter, and quality class.
The third aim is to improve the utilization of the stand data, such as information about different stem distributions as to the tree species and density, for example. On grounds of these information the stand can be cut into logs by computer before the actual harvest. This gives an estimate of the log distribution in that stand and so tells about the demand the stand can meet.
On the basis of the stand data it is also possible to combine data from different stands within the integrated system, giving a good opportunity for a tree exchange system. The computer of a harvester is able to record a lot of information about the logs harvested. This makes it possible, for example, to produce a real-time location map of the timber for the logistics.
Research partners include forestry researchers, and a research team of spatial statistics from the University of Jyväskylä. Some companies are also involved in the project.