Projecting Plots in LiDAR – Correcting for Slope in Remotely Sensed Data

Authors: Tom Adams, David Pont
Publication: New Zealand Journal of Forestry, Volume N.Z.J.For. 2012, Issue N.Z.J.For. 57(1) 2012, pp 25-31, May 2012
Publisher: New Zealand Institute of Forestry

Abstract: Double sampling inventory of a varied resource – such as using aerial LiDAR in conjunction with forestry field measurements – relies on the assumption that the area sampled in the field can be accurately quantified in the remote sensing data. GPS error and the horizontal shape of the ground plots both lead to mismatches between the area measured in the field and the area sampled from the remote sensing data. The horizontal shape of a ‘circular’ ground plot will vary on sloping terrain when the boundary is defined by measuring the distance the from a centre peg with a tape running parallel to the ground (normally at breast height). If a circular shape is cut out of the remote sensing data (as is usual practice) then it will not exactly match the area measured in the field. This discrepancy can be completely corrected by using slope-correcting instruments such as the vertex, and although this is becoming more common, the majority of circular plots within New Zealand on steep ground will still be measured by tape with a slope-corrected radius. The GPS accuracy – for locating the plot in the remotely sensed data – can be improved by using high grade GPS systems and antennas, collecting as many measurements as possible and differentially correcting the results. However, even with the best technology, GPS locations are only accurate to around 1m. In this study we found that the standard field measurement practice of assigning a slope-correction factor to the plot radius led to no bias of over or under sizing plots in a sample of 498 LUCAS plots. Approximating the field sampled area with a circle was correct for 97.3% of the area field measured (29.07ha); leaving only 2.7% (0.08ha) of the total LiDAR sampled area misrepresented in the field. In comparison, the GPS accuracy led to a slightly larger amount of area misrepresented (6%). As the two errors are additive, it is always beneficial to use a vertex to determine horizontal distances and removing the shape error. However, as the discrepancy is relatively small compared to GPS error, its effects are negligible in the LUCAS double sampling. LUCAS uses LiDAR derived metrics to estimate biomass carbon (Stephens, 2007; Beets et al., 2011; Stephens et al., 2011). A comparison of these metrics from both circular LiDAR plots and plots equivalent to the area actually field sampled showed that the metrics used to calculate carbon per hectare varied on average by no more than 0.22%.
Access to the full text of this article is available to members of:
  • NZ Journal of Forestry (NZI)
  • Non member Online NZ Journal of Forestry
If you're a member and should have access:
Login

Otherwise:
Register for an account