> Research

Comparison of Lidar and Field Measurements of Loblolly Pine

Brice Young and David L. Evans

Many new remote sensing tools are being made available to natural resource managers. One such tool is small footprint Light Detection and Ranging (LIDAR)system. This is an active sensor that directly measures elevation. LIDAR is a 3-dimensional point sampling system and does not suffer from many of the common limitations of traditional imaging systems.

Traditional forest inventories are often labor intensive and cumbersome. One potential application is the assessment of basic forest stand parameters such as average height, average crown area, and trees per acre. Other potential applications of this system include large-scale, multi-species forest inventory and canopy structure modeling in forested ecosystems. With the advent of laser altimeters, "rapid and accurate assessment of vertical properties can be made" in both forested and non-forested systems (Ritchie et al, 1993). Lasers can be used to address many aspects of a forest's canopy, including stand height, biomass, percent cover, and other critical stand characters that are very useful to forest managers (Weishampel et al, 1996).

The objective of this study is to determine if LIDAR can be used to characterize basic stand attributes needed for forest inventory accurately and efficiently. Of most interest to this study will be the investigation of three key stand characteristics required to perform basic mid-rotation pine plantation inventories: 1) Mean Stand Height, 2) Trees Per Acre (TPA), and 3) Mean Crown Area.

Mid-rotation loblolly pine plantations (age 9 - 22 years), in eastern Mississippi and coastal Georgia, were sampled using both traditional field sampling and LIDAR in 1999. Traditional field measurements were taken with the assistance of International Paper field personnel and are being compared to interpolated, small-footprint scanning LIDAR image measurements for statistical differences. Field data were collected from at least nine one-fortieth acre circular plots within each individual stand. Diameter at breast height, total tree height, and 4 crown radii were recorded on each tree. Plot means were calculated for the field data and are being compared to means of the interpolated LIDAR image measurements. If LIDAR data can be used to precisely predict basic stand parameters, then the data could also be calibrated to accurately predict stand volume.



The figure above is a scatter plot of first return raw LIDAR data values. Data has been scaled. View is oriented to look across a 60 meter swath of LIDAR data. Note the penetration ofLIDAR data through the main canopy (those points which fall under the majority).



The figure above is a perspective view of the first return data depicted in the lower left picture. View has a vertical exaggeration of three times. Stepped focal filtering and thresholding were used to automatically detect assumed treetops within the interpolated LIDAR data. Focal ranking was used to determine each pixel's relative value when compared to its neighbors. Pixels ranking in the top 10% of relative elevation values were grouped as assumed crowns. The maximum value within each assumed crown was determined using GIS (geographic information systems) techniques.

Plot centers were determined through real-time global positioning systems. A buffer zone of 18.6 feet, corresponding to field plot size, was established around each plot location and maximum values for all identified crowns were extracted for use in t-tests with corresponding field data.

The analyses results indicate that LIDAR underestimates stem height and crown diameter but adequately estimates stem count.

References:

Ritchie, J. C., D. L. Evans, D. Jacobs, J. H. Everitt, M. A. Weltz , . 1993. Measuring Canopy Structure with an Airborne Laser Altimeter. Transactions of the ASAE. 36(4): I235-I238.

Weishampel, J. F., K. J. Ranson, and David J. Harding. 1996. Remote sensing of forest canopies. Selbyana 17:6-14.