Research projects 

Analysis of Longitudinal Data

(1.8.1990–31.7.1993) headed by Esko Leskinen (University of Jyväskylä), Erkki Liski and Gunnar Rosenqvist (Swedish School of Economics).

Developing an Integrated System for Forest Harvesting

(1993–1994) headed by Erkki Liski, funded by the Ministry of Education.

Analysis of Covariance Structures

(1995–1996) headed by Erkki Liski and Götz Trenkler (University of Dortmund), funded jointly by the Academy of Finland and Deutscher Akademischer Austausch Dienst.

Developing of Forest Harvesting Systems by Utilizing Statistical Methods

(1997–1999) headed by Erkki Liski.

The effective use of Statistical methods have become possible by the appearance of modern computer-controlled forest harvesters. There are four mutually supportive parts within this project:

1. Utilization of prior information and pre-harvest measurements
2. Modelling and prediction of tree stems and bucking in cases of incomplete information.
3. The log mix problem and selection of stands for felling
4. Sampling and experimental design
5. Forest harvesting simulator program

The focus of this research is on the application of statistical methods in the areas named above. The basis of the information system should be efficient and comprehensive data collection and data management, which offer an opportunity for the application of mathematical-statistical methods.

A relatively lightweight method is needed in order to estimate the quantity and structure of standing timber for planning the harvesting of forests. Sampling data or information obtained from comparable stumpage are used to update the prior information on the standing timber. The estimates based on the log distribution is the basis of planning forest harvesting according to desired criteria. A mixed model approach has proved repeated-measurements to be a flexible tool for stem curve modelling and prediction. The aim is to develop these methods more suitable for practical purposes. One can find more information about the system for harvesting from here.

Our group has studied optimal designs for estimation and prediction in linear random coefficient regression models and growth curve models. The results are applied to tree stem data collected by forest harvesters. We have also introduced and investigated so-called DS-optimality criterion.

The project's home page is http://www.uta.fi/laitokset/mattiet/tilasto/projekti.html.

Advanced Methods for Computer-Aided Bucking of Scots Pine

(2001–2003) headed by Tapio Nummi. Project's home page is http://www.uta.fi/laitokset/mattiet/tilasto/uusiprojekti/projekti.html.

Forest-level bucking optimization including transportation cost, product demands and stand characteristics

(2004–2006) headed by Jori Uusitalo (Finnish Forest Research Institute, Metla) and Tapio Nummi. Project's home page is http://www.joensuu.fi/~wmets03/.

Nonparametric and Robust Methods for Mixed Models

(2006–2008) headed by Tapio Nummi and Hannu Oja (University of Tampere).