14. Charcoal sites - VUKOZ-OEL/3d-forest-classic GitHub Wiki

Project

Methods grouped in this section are created specifically for scientific project "TL02000160 Úloha milířišť z hlediska kulturního dědictví a ochrany krajiny.!" as part of research focused on charcoal site as specific cultural heritage and ecological spot.

Methods

For charcoal sites detection is crucial detailed terrain model, which can be product of TLS, ALS or UAV lidar scanning. Thus 3D Forest seems to be ideal platform for site detection since it has implemented methods for terrain analysis like slope, curvature, aspect or hillshade. Charcoal sites can be easily identified on terrain slope, curvature or hillshade layer. Our main goal here was to implement methods, that can automatically detect charcoal sites and compute basic parameters like area, main axis size or centroid.

For understanding whole workflow for charcoal detection, here you will find its description for base cloud to charcoal features.

Terrain segmentation

For point cloud segmentation into vegetation and terrain we use methods described on page Terrain analysis Our recomended step would be "Terrain from Octree". This sites produce two layers - vegetation and terrain. After visual inspection of terrain you can manually adjust this layer by removing point that do not belong to terrain using "Manual adjustment" method.

Slope computation

In terrain analysis you can find method "Slope, that computes vertical diference of surrounding points and compute its value in percentage or in degrees. By selecting input parameter to "radius search" instead of number of neighbor you can define neighborhood from which slope is computed. This step affect grain of slope. For charcoal sites I recommend neighborhood size of 3m (input value:300 cm). Computed values are stored in "Intensity" field and can be displayed by mouse left button click on layer - "Color By Field" and select "Intensity". Slope layer with charcoal sites Visually now we can see spots where slope is close to 0 value unlike the rest of terrain. Using next steps we can more focus on changes of terrain slope, since on artificial sites tends to be horizontal or vertical. Such places can be easily see on curvature layer.

Curvature

Terrain curvature is defined as second derivate of terrain. The first derivate is slope, so our task here is to find differences in slope (input layer is slope result). Another parameter is define its neighborhood similar to Slope method. After computation of curvature result stored in "intensity"field of curvature result seems to be with wrong scale: Curvature of terrain

So for visual check we can select only point where change is close to values -5 - 5. Such result gives us better understanding where to find charcoal sites. For point selection based on its values can be found in Menu "Other Features"->"Remove Selected points from Cloud".

curvature-selected values

Now we can start automatically extract charcoal sites from curvature layer.

CharCoal Menu

In separate Menu "CharCoal" can be found three methods - one for computing Charcoal sites, "Feature table" for displaying computed values of sites and "Export Features" for Exporting computed values into text files for further processing.

Charcoal sites

This method is crucial for site extraction. As we see in previous steps, we are looking for places that has curvature value close to zero. Input cloud is thus Curvature layer, from which we are looking only for points of minimal value about -2 and maximal value around 4. After selecting points Method starts grouping points into clusters based on average point distance - select only neighbor points. Clusters are filtered by input values - point count, area, and axis length. Those clusters that fulfill all conditions are stored as features in table and also create virtual layers (not saved) that can be visualized.

Feature table

After computing sites user can see its value in table clicking on Menu "CharCoal" -> "Feature table" table will appear on the bottom of application.

Export Features

Computed features (sites) can be exported into text files - one with coordinates of centroid and computed values, next files contains concave and convex hulls of sites.