How do we measure the age of the forest?

We use a combination of remote sensing datasets to estimate the likely age and biomass of the forest. We use a variety of datasets to mitigate the effects of some uncertainties in the remote sensing datasets. We also use a tree species map, which allows us to estimate the maturity of the trees on a per-species basis. We are continually updating our methods to improve the accuracy of our estimates.

Why do we have the 3HA limit?

Note: This limit is somewhat arbitrary, but reflects the likelihood that smaller plots have higher uncertainties. For example, one source of uncertainty is the so-called boundary effect. The forest cartoon below shows a land defined by three pixels, each 30 by 30 m in size (figure 1 below). The green ellipse corresponds to the forest represented as a single patch. The pixels that lie on the forest-no-forest boundary have a higher likelihood of being classified incorrectly. This is shown in figure 2 and figure 3 below. Assuming that the forest should cover at least 50% of the pixel area to be classified as a forest, figure 2 shows the results of classification with zero errors. We call respectively the correctly classified forest and forest-free pixels as True Positives (TP) and True Negatives (TN). Figure 3, on the other hand, shows five incorrectly classified pixels: three false positives shown as the green pixels in the three corners labeled as FPs, and two false negatives shown as the white pixels with FN labels.

The cartoon of the land covers 90x90=8100 square meters or roughly one hectare (100 by 100 meters; note that we need 11 pixels to cover 1 ha of land). If we take instead of 9 pixels, representing here 1 ha, 33 pixels to fit 3 ha of land, there will be more pixels covered entirely by forest, and therefore there will be fewer incorrectly classified pixels. As mentioned above, 3 ha is somewhat arbitrary. E.g. by choosing 6 ha, we will have even higher classification accuracy, but then smaller landowners cannot be onboarded by Single.Earth.