Congressi Nazionali SISEF

• Home 
• Cerca Contributi 
• Risultati Ricerca 
• Contributo #c11.3.6 

Presentazione orale

Giannetti F, Puliti S, Gobakken T, Næsset E, Travaglini D, Chirici G

DTM-independent variables to predict forest inventory variables using 3D UAV photogrammetric data

Riassunto: This study presents a novel approach for the extraction of a new set of explanatory variables from 3D UAV photogrammetric data without the relying of any Digital Terrain Model (DTM) to normalize the data (i.e. to obtain relative heights above ground). The set of DTM-independent variables was used to predict five forest inventory variables: growing stock volume, basal area, stem number, Lorey’s height and dominant height. To gain further insights in the applicability of the approach across different regions, the assessment of the DTM-independent variables was performed across two different forest types, namely a temperate mixed forest in Italy and a boreal forest in Norway. The use of DTM-independent variables was compared against two more traditional sets of variables: (i) statistical, height and density variables from UAV photogrammetric data normalized using an ALS DTM (Image-DTM_ALS), and (ii) statistical, height and density variables from normalized ALS echoes (ALS variables).Multivariate linear regression models were fitted with the forest inventory variables as response and the three different sets as explanatory variables, i.e., (i) DTM-independent variables, (ii) Image-DTM_ALS variables, and (iii) ALS variables. The mean accuracy across all the studied forest inventory variables found for the models using the DTM-independent variables was 19,6%, which was similar or smaller than the RMSE% of the other approaches. For the Image-DTM_ALS variables RMSE% was 19,7% and for the ALS variables 21,6%. Interestingly, as the terrain and forest structure complexity increased (mixed forests) the DTM-independent variables yielded smaller average RMSE% (19.1%) than ALS (23,2%). Our results suggest that 3D UAV photogrammetric data may be used effectively for forest inventories even when high resolution DTMs are not available.

Parole Chiave: Photogrammetry, Structure from Motion, Airborne Laser Scanner, area-based approach, Unmanned Aerial Vehicle (UAV), Unmanned Aerial System (UAS), forest inventory

Citazione: Giannetti F, Puliti S, Gobakken T, Næsset E, Travaglini D, Chirici G (2017). DTM-independent variables to predict forest inventory variables using 3D UAV photogrammetric data. In: XI Congresso Nazionale SISEF “La Foresta che cambia: ricerca, qualità della vita e opportunità in un paese in transizione“ (Roma, 10-13 Ottobre 2017), Abstract-book, Contributo #c11.3.6. - [online] URL: https://congressi.sisef.org/?action=paper&id=501

Dettagli