Using Remote Sensing for Quantity Analysis of Chip Pile Inventory in Mill Yards
Master of Science
DisciplineNatural Resources Management
Subject3D model capture
Mill yard inventory
MetadataShow full item record
An integral part of proper wood chip inventory management is the ability to accurately monitor wood chip quantities. This thesis examines the use of a new method of capturing the volume of mill yard wood chip piles through the utilization of aerial drones. The drones are used to capture images and the images are converted into digital 3D models, which are then capable of measuring pile volume. This process allows for conversion of the volume into an accurate mass estimate by compensating for compression factors within the chip pile. These factors can change the volume by a maximum of 9.46%, but on average during simulations and real world applications, most piles exhibit a change in volume in the range of 1% to 6% difference. By performing the estimation procedure multiple times and averaging the results this method is able to generate a result that is more precise, timely and less labour intensive than the previous methods of using a ground survey to determine volume and applying a linear volume to mass conversion for the quantity of wood chips. The results suggest that this averaging technique can improve the standard deviation spread from over 5% variation in the measurement to less than 2%. This new method combines multiple techniques to improve both overall accuracy and precision. Each stage of the new method was examined to determine the accumulated degree of error. This included looking at operator error of about 2.4%, considering the precision of 3D volume capture, which adds on average of 5% to 10% error, understanding the variation in bulk density due to pile shape, and size, which adds 1% to 6% error, using different 3D software modeling for measuring pile volume, which adds about 4% error. Combined together in extreme cases, these errors can skew the results by over 20%. The results of this examination provides research-based recommendations as to how to collect the images, generate the models, and process the data for mass estimation and improve error reduction at all stages.