A new method of digital wood anatomy, pixel-contrast densitometry, was proposed recently for conifers. This technique can replace the X-ray densitometer with software that scans and analyzes the image. A photograph of the wood cross-section consists of pixels, i.e., dots of equal area belonging to either the cell wall or the lumen (empty space). By measuring the number of pixels of the cell walls and lumens, the ratio of the cell wall area to the total area of the scan can be determined. The result of multiplying this ratio by the cell wall matter density is the wood density, averaged for the scan area. We used this method to obtain another alternative parameter of wood structure, wood porosity, that is, the proportion of lumen area within scan. We applied this approach to eight angiosperm species of shrubs and trees in Southern Siberia, varying from ring-porous to diffuse-porous wood. The original software was created to implement the technique automatically.

Before analyzing images, their transformations were performed: (1) the Gaussian smoothing algorithm to clean the photo from noise and (2) the Otsu method to separate the cell walls and lumens. Then, a porosity profile over the entire cross-section of the growth ring can be constructed using the program’s virtual sensor.

Good synchronicity between several measurements within ring proved that profiles of wood porosity describe the ring structure and capture inter-annual differences.

For samples of the three most long-living species, Prunus padus L., Caragana arborescens Lam., and Alnus alnobetula Ehrh., 14-32 year-long series of maximum, mean, and minimum porosity determinations and ring widths were analyzed, showing comparable inter-series correlations, common signals, and consistent inter-relations between these parameters of wood. This means that these easily measured characteristics register environmental variability during their formation. Directions for further research were also discussed.