In December 2014 a sudden sleeting rain caused a lot of damage when it froze on trees and leveled complete forests or deprieved trees of their foliage, leaving behind needle-like bare trunks. The looks of it were already terrifying. Many photos and videos were shot of the situation. There were a few mountains, forest roads and valleys where it was near-impossible to get to even after a month because of fallen trees and deep mud.
The question was how to show the extent of the damage in the shortest time.
The job wasn't easy! -10 degrees, 800-1000 metres of altitude, closed valleys, low altitude, nearly no view of the working area from below. These made the drone's application impossible to use, because even in terrain-following mode it would have resulted in a crash due to high and turbulent winds on the side of the mountain.
Despite that our copters nicely flew over the area and mapped these tight valleys properly and precisely.
As a final result the orthophoto shows a slice of the area with the usual detailing.
From the photo we can say the following:
- the damaged and undamaged areas are indistinguishable- especially not digitally
- according to the photo, elements of 10 cm are recognizeable,
- however there is no clear border between fresh, young or old forest or ground on either of the photos - doesn't matter how detailed it is.
At this point 3D modeling becomes special and indispensable. A so called heightfield model does the following according to a simplified algorithm:
All the locations where it can't find a reference point because the ice leveled, brought down the trees or the forest is scarce then it "pushes" the poing down to ground level. On locations where it finds reference it nicely draws the foliage of the trees. The border between the two is a distinct "tear".
The next 2 figures show the difference. On the left is the heightfield on the right the textured.
The "tears" are clearly visible, even on the smallest patches.
The whole area from the sides according to the new visualization.
Modelling in this way creates other opportunities as well. By rating the dense mesh of reference points - with unique parameters - it is possible to identify specific trees and forests to differentiate the ground from the trees. Even more to see the underwood, the middle range and the foliage as well!
Browne means ground, the white puffs show a group ready to be rated. In this the green are for example is a selected foliage area. By specifying other methods it is possible to create a closed model of the rated forest area and count the volume of the forest in that specific area.
We are often asked how these models can fit into 2D GIS systems. This is the area's orthophoto in EOV split in the forestry's qQIS, mapped onto the previous summer's orthophotos and the reference map showing the borders of the area.