Statistical methods for Estonian forest resource assessment using remote sensing data developed

This summer concluded the almost two-year study to develop statistical methods for forest resource assessment using remote sensing data. The study was concucted by researchers from the University of Tartu Institute of Mathematics and Statistics and Tartu Observatory.

Tartu Observatory's Senior Research Fellow in Remote Sensing of Forests Mait Lang cooporated with Associate Professors of the Institute of Mathematics and Satistics Raul Kangro and Märt Möls and Professor Emeritus Kalev Pärna to develop the statistical methology for using lidar, or laser scanning as well as satellite photos to evaluate Estonian forests.

"Researching this field is important, because remote sensing enables to receive data cheaper and faster than in situ measurements," Project Manager Pärna stated. "At the same time, remote sensing results can be quite prone to errors, which makes it difficult to use this data for forest resource assessment."

The statistical methology was developed for raster charts (charts made out of pixels), which show tree species and the height and growing stock of the forest. For each chart, informative measures were identified, and based on these, necessary predictions were done using machine learning methods. As a result of the project, an innovative way of calculating the prediction error was designed. The new methodology is based on both statistical and data mining methods. University of Tartu statisticians also offered ideas for further improvement of the project, highlighting ways of using remote sensing data, improving year by year, for making excisting assessments more accurate.

The study was a part of the Ministry of Environment's procurement project "Integration of remote sensing image processing methods for forest resource assessment", which was carried out by company Reach-U and University of Tartu.

Additional information:

Kalev Pärna, Professor Emeritus in Probability Theory, University of Tartu, 737 5452, [email protected]
Mait Lang, Senior Research Fellow in Remote Sensing of Forests, University of Tartu, 737 4528, [email protected]