Assessing the Accuracy of Lidar as a Forest Inventory Tool Using Individual Stem Identification and Attribution

Lidar is fast becoming one of the most widely used and accurate remote sensing tools for forest inventory. The means by which the lidar data is used to accomplish these inventories varies greatly. This study examines the use of individual tree detection and attribution to assess various forest characteristics, along with testing two alternative methods of determining canopy cover from lidar. Individual stem detection was accomplished using a local maxima algorithm. The total number of stems detected by lidar was 6% lower than field tallied stems, with regression analysis yielding an R2 of 0.59 and RMSE of 188/ha. The difference in number of lidar versus field trees also differed by density. For those plots with a density greater than 600 trees/ha, the number of stems detected by lidar was 26% lower than field measured stems. For those plots with less than 600 trees/ha, lidar detected a much larger number of small stems (< 35 cm diameter), leading to a larger estimate of tree density overall (~25% more lidar trees). Field based measures of mean and maximum height were highly correlated with the lidar data, resulting in R2 values of 0.85 and 0.89, and RMSE values of 1.6 and 1.7 respectively. Overall mean diameter and basal area were closely estimated by lidar, with both field and lidar mean diameter at 27.8 cm and total basal area at 77.3 m2 for lidar and 77.4 m2 field data. Plot-level comparisons of these 2 attributes showed some variation however. Mean for the diameter had an R2 of just 0.56 and an RMSE of 4.7 cm. For the plot level basal area, R2 was 0.57 with an RMSE of 0.8 m2/0.1 ha. Canopy cover was found to be most accurately estimated using the intensity (i.e. returned energy) of the lidar pulses and calculating the ratio of canopy intensity to total intensity. The high correlation between lidar-based estimates and field-based estimates suggests that lidar data can be effectively used to help provide complete wall-to-wall data for key forest inventory attributes.

Name thesis:degree_name

Master of Science (MS)

Grantor dc:publisher

University of Montana

Year

2010

Author dc:creator

• Floch, Luke

• basal area
• canopy cover
• DBH
• forest inventory
• lidar
• stem detection

Repository record dc:identifier

https://scholarworks.umt.edu/etd/65

OAI identifier oai:identifier

oai:scholarworks.umt.edu:etd-1084