ForestCO2 project: Goč in-situ data

This is a repository of in-situ data from the Faculty of Forestry test field located on the Goč mountain, collected during the ForestCO2 project (EO and in situ based information framework to support generating Carbon Credits in forestry), which was funded by the Science Fund of the Republic of Serbia, grant no. 6686. Description of the data contained in each file:

    
Goc_Block_0.las   and  Goc_Block_1.las   - Parts of the Goc forest point cloud.

Goc_chm_1m.tif   and  Goc_dtm_1m.tif    -  Canopy Height and Digital Terrain Model, both products derived from the point cloud with 1m spatial resolution.

DTM_20cm_Goc.tif, DSM_20cm_Goc.tif and CHM_20cm_Goc.tif    -  Digital Terrain Model, Digital Surface Model and Canopy Height Model, products derived from the point cloud with 0.2m spatial resolution based on detailed classification of point clouds.

Tree_inventory-extracted_features.gpkg   -   Tree inventory as 3D point features obtained from LiDAR point cloud extraction, based on detailed classification of point clouds.

Goc_measurements_all.xlsx and Goc_insitu.xlsx -  Spectrometer soil scans, and lab analysis results for the same soil samples.

Detailed description of the datasets:

• Soil chemical properties analysis

Soil analysis was performed by an accredited laboratory in Ruma. The dataset contains ~200 samples from the Faculty of Forestry test field located on the Goč mountain. The samples contain only the Hummus (%) content, and every 10th sample has also ph measurements.

• Soil spectral properties analysis

All of the above-described soil samples were scanned with the NeoSpectra handheld NIR Scanner (developed by Si-Ware). This instrument captures the soil spectral reflectance in the 1350 -2500nm spectral range. All samples were scanned 6 times with a 10s scan time. The scanner was not moved between scans.

• Forest LiDAR point cloud

These are the measurements taken with the DJI ZENMUSE L2 instrument mounted on the MATRICE 350 RTK drone, in the Faculty of Forestry test field on the Goc mountain. The measurement campaign was performed in September 2024. Point cloud covers approximately 850 000m2 and average point density is ~800 points/m2.

• DTM and CHM derived from the point cloud

Digital Terrain Model and Canopy Height were derived from the point cloud using the LidR software library in the R programming language. Both products were derived at 1m spatial resolution. The DTM product was constructed using the TIN (triangulation) algorithm, while CH was obtained using the point to raster algorithm (highest point in every raster cell is taken as the canopy height).

• High resolution DTM, DSM and CHM derived from point cloud

Digital Terrain, Digital Surface and Canopy Height Model were derived from the point cloud using the Global Mapper software. All products were derived at 0.2m spatial resolution. The DTM product was constructed based on classified ground points using the Binning method implemented in the software, made up of the lowest points in each area. Binning is a processing data modification technique that takes point data and creates a grid of polygons, or bins. The value of each bin or polygon is representative of the point values within it. An inverse weighted distance algorithm is used to fill in the gaps.  The DSM product was constructed based on classified vegetation points using the Binning method implemented in the software, made up of the highest points in each area. The Canopy Height Model is the difference in height between the DTM and the DSM. After subtraction, the resulting raster stores the values of tree height instead of elevation and the Unsigned option is used (absolute value), to ensure that all of the values is positive and all of the trees are above ground.

• Tree inventory - extracted features

Tree inventory - extracted point features were obtained from the classified LiDAR point cloud using the Global Mapper software. Workflow consists of the following steps: classify the point cloud to identify tree points and extract the point cloud into individual tree features. A tree point is created at the center of each segment and contains the tree’s measured attributes, such as height, canopy spread, and classification. The total number of 31 707 features - trees were extracted from point cloud in the area of interest.