Changes during twelve years in three mature hemiboreal stands growing in a radiation model intercomparison test site, Järvselja, Estonia

For the fourth international radiation model inter-comparison experiment (RAMI) three forest stands were selected at Järvselja, South-East Estonia (Kuusk et al., 2013; Widlowski et al., 2015). In each of the three forests a one-hectare-large square-shaped sample plot was established in 2007 and the positions of all trees with stem diameter at breast height d greater than 4 cm were mapped. The three sample plots represent typical examples of hemi-boreal forests – birch and spruce stands are growing on fertile soil and the pine stand is located on a transitional bog.

Since 2007, regular measurements of canopy gap fraction have been carried out in the stands (Lang et al., 2010; Lang et al., 2017; Kuusk et al., 2018; Lang & Pisek, 2019). There have not been any substantial stand structure alternating disturbances in the stands. Only in the birch stand we removed the understory trees that had d less than 4 cm to prepare the site for terrestrial laser scanning experiment in 2018. During the regular visits to the sites we have been observing occurrences of single tree mortality in the stands, changes in canopy conditions and also tree growth. Also, more precise coordinates of sample plot corners can be measured with differential GPS now. Therefore, in the summer of 2019 all three stands were fully inventoried to update the RAMI database, create a checkpoint for modeling experiments and validate remote sensing-based forest structure estimates.

Measurements with differential GPS on the corner points of the test sites provided coordinates that were used to transform tree location maps into the Estonian basic map coordinate system (Table 1). There are small (less than 0.5 m) deviations from the planned ideal 100 by 100 m square which is explained by the decreased precision of GPS under forest canopy and also with small errors made in 2007 when the test sites were established.

In the birch stand (Figure 1) basal area G has increased from 23.3 m² ha^-1 to 28.2 m² ha^-1 in the upper layer and from 4.2 m² ha^-1 to 6.0 m² ha^-1 in the lower layers. The number of trees N in the upper layer has decreased from 654 to 565 ha^-1 and increased in the lower layers from 377 to 401 ha^-1 (Table 2). In the lower layer the increase in stand density is controlled by the cleaning in 2018, when most of the dense understorey and regeneration was removed. The natural mortality is occurring mainly on smaller trees and mean diameter of the dead trees is 10.4 cm (Figure 2, Table 2). The substantial change in mean tree crown length L_cr and crown radius R_cr of spruce trees in the regeneration layer is because in 2007 we used a model from the spruce stand, but in 2019 a model based on sample trees from the birch stand was applied. Compared to model prediction the mean value for relative live crown length for spruce sample trees in 2007 was 0.74 (n=3) and 0.87 (n=6) in 2019. As there is sufficient amount of light for Norway spruce regeneration trees in the birch stand, live crown base rise is small and live crowns extend almost to the ground.

Figure 1.

Figure 2.

In the spruce stand (Figure 3) basal area has increased from 30.9 m² ha^-1 to 35.4 m² ha^-1 in the upper layer and from 6.8 m² ha^-1 to 7.1 m² ha^-1 in the lower layers. The number of trees in the upper layer has decreased from 774 to 724 ha^-1 and substantially decreased in the lower layers from 912 to 577 ha^-1 (Table 3, Figure 4). We distinguished two separate classes of spruces in the regeneration layer based on their competition status according to the shape of the crown and increment of the top shoot.

Figure 3.

Figure 4.

While H and G increased as in the birch stand, the crown radius decreased and live crown length remained almost the same for the spruces in the upper layer and second layer between the two measurements (Table 3). Heights to live crown base were measured with hypsometer and although random errors are possible, there is probably no systematic error. Measurements of crown diameter are prone to systematic errors as the interpretation of crown projection may have been slightly different in 2007 and 2019. However, it is not excluded that live crown diameter has decreased considering the conical shape of the Norway spruce crown (Kantola & Mäkelä, 2006) and the live crown base rise with stand age (Valentine & Mäkelä, 2005), and that the mean live crown length in the spruce stand remained almost the same as in 2007.

During preparations for field measurements in the pine stand we identified one tree that was measured in 2007, but due to error had its location coordinates determined outside the test site. The tree was left out from the report by Kuusk et al. (2013). Here the tree is included and therefore for 2007 the number of trees in the upper layer is 1116 (Table 1). The pine stand (Figure 5) has the simplest structure of the three stands. There is basically only one layer of Scots pine trees and few small trees that matched the minimum d rule. As in the pine and spruce stand, we can see an increase in basal area, mean tree diameter at breast height and stand mean height. The number of trees decreased with smaller trees having a higher probability of dying (Figure 6).

Figure 5.

Figure 6.

The three sample plots in Järvselja forest district are similar to the examples in Harvard forest (Eisen & Barker Plotkin, 2015) and to the fiducial reference site in Speulderbos that was established for the validation of satellite measurements-based predictions of forest biophysical variables (Brede et al., 2016). Although for the validation of methods for a wide area map (Dostálová et al., 2018) construction the number of validation observations is more important (Copernicus HRL, 2019) than the sample plot size and tree-level detailed data, large sample plots are better suitable for experiments like RAMI described by Widlowski et al. (2015). Large sample plots are the only means to study the impact of sample plot size on the estimation of forest inventory variables regarding tree location patterns (Maleki & Kiviste, 2015) and also for testing global estimates of forest biophysical products like MODIS BRDF (Pisek et al., 2015).

Our repeated measurements in the three sample plots showed that stand basal area, stand height and mean tree diameter at breast height have been increasing since 2007. The smallest increments were observed in the pine stand which is explained by the poor growth conditions i.e., excess water and limited availability of nutrients in the thick peat soil layer. An interesting finding in the spruce stand was that crown diameter of the spruce trees decreased compared to 2007. As interpretation errors in tree crown outer boundary estimation may be one cause to the result, we expect more solid evidence from terrestrial laser scanning data from 2013 (Kuusk et al., 2015) and 2019 measurements.