Coarse woody debris sampling
intensity considerations
by
John Parminter
Ministry of Forests, Research Branch
Victoria, British Columbia
August 1998
Introduction
Coarse woody debris (CWD) refers to the larger dead and down woody material on the
forest floor in natural stands as well as that found in harvested areas and managed
forests. Standing dead trees (snags) and stumps are usually excluded. Coarse woody debris
is most often sampled along a line transect of varying length arranged singly, in pairs
(emanating from a common point or crossing at their midpoints) or as a triangle.
CWD sampling in the Vegetation Resources Inventory
The CWD sampling that will be carried out at the provincial level in British Columbia
is part of the Vegetation Resources Inventory (VRI). Thus, the sampling intensity used
stems from the VRI system for delineating polygons, determining which will be sampled and
how many samples or plot clusters will be put in each one.
In what is called Phase 1, the polygons are delineated by airphoto interpretation. The
minimum polygon size for areas with distinct boundaries is 2 ha and for areas with
indistinct boundaries it's 5 ha. Indistinct boundaries are caused by irregular tree
heights, repetitious ridges and swales, and patchy forest alternating with rock outcrops
(for example).
In preparation for the ground sampling (Phase 2 or plot establishment) the identified
polygons are sorted and selected. A 100 by 100 m grid overlays the entire province and the
sampling points (or plot clusters) are located at the grid points. The end result is a
minimum sampling intensity of one sample (plot cluster) per 2 ha, the expected minimum
polygon size.
For larger polygons the number of samples would depend on how many of the 100 by 100 m
gridpoints in the polygon were selected. For example, there would be nine plots in a 9.1
ha polygon so the maximum intensity would be one sample per ha. Each sample would have two
24 m lines for CWD sampling.
How many plots are required in logging slash?
Most analyses have concerned logging slash sampling, which involves more fine material
than CWD sampling and potentially more variation due to logging effects (utilization
standards and piece orientation). There is more smaller material in logged areas and
usually less larger material in forested situations, with the exception of accumulations
caused by windthrow or snowbreak.
The recommended procedures for logging slash sampling in B.C. use two triangles with 30
m sides per stratum per cutblock (Trowbridge et al. 1989). A stratum was defined as a
combination of topography, ecosystem type, fuel type and fuel loading but no sampling
density per se was recommended.
Triangles are easy to lay out and avoid piece orientation bias caused by slope and/or
disturbances, even though some pieces may be sampled twice if they lie across a corner.
Another approach is to randomly arrange individual transect lines through a cutblock. If
you're not tied to another sampling scheme (the VRI CWD lines are also used for shrub
sampling) it may be easier to use triangles or single randomly-placed lines.
For logging slash situations, McRae et al. (1979) recommended 90 m of sampling line per
20 ha, this being one-tenth of the VRI sampling intensity for CWD. But remember that the
greater the amount of material, the shorter the sampling line required.
The most detailed analysis of line intersect sampling was done by Pickford and Hazard
(1978). Their Figure 1 shows the relationship between the number of sample lines required
to meet different levels of precision. A less detailed analysis was done by Howard and
Ward (1972) and they concluded that, for logging slash situations, attainment of precision
levels of +/- 15 to 20% is quite feasible but to go below +/- 15% would require
substantial effort. They did not test for sensitivity to line length but stated that lines
greater than 60 m might yield lower sampling errors.
How many plots are required for CWD in other situations?
Pickford and Hazard (1978) calculated the intensity of sampling required to produce CWD
volume estimates to different levels of precision. For instance, estimation of residue
volume to 95 + 10% confidence would require 363 sample lines of 22.86 m. Placement
of 363 sample lines within a particular fuel/forest type would not always be possible and
to sample this many would be extremely time-consuming and result in massive quantities of
data.
Unfortunately inadequate analysis has been done regarding sampling requirements for CWD
in natural forests and managed stands. Brown et al. (1982) recommended 304 metres of
sample line for every 20 ha with "...high diversity in amount and distribution of
fuel and vegetation..." (in this case "nonslash" or "naturally fallen
material" greater than 7.6 cm in diameter). By comparison, the VRI sampling scheme of
48 m per ha would convert to 960 m per 20 ha, three times what Brown et al. recommended.
Two papers point to the need for longer CWD transects. Harmon and Sexton (1996)
included a graph based on Brown (1974), indicating that transects of 100 m in length are
required to produce a good estimate for larger material. The more material present, the
shorter the lines needed and vice versa.
For CWD sampling in forested situations, it seems that one triangle (with 30 m sides)
per ha would be quite adequate, or three scattered 30 m lines. But because some
researchers have recommended 100 m lines, if it’s feasible to put in one such line
per ha this may be a better alternative.
Conclusion
Statisticians have ways of estimating the amount of sampling needed to achieve certain
levels of precision and can do so after some test data have been collected. Beyond that
one has to consider the size of the area that's being inventoried, the amount of expected
variation, the purposes to which the data will be put and the time, energy and cost
implications of different levels of sampling.
And alternative to the line transect is the complete census sample. In this approach,
the end diameters and lengths of all CWD pieces are measured within a fixed area plot.
Clark et al. (1995) compared the line transect and complete census sampling schemes and
concluded that both yield similar results and show no evidence of systematic bias. The
line transect system was preferred because it is less subjective and considerably faster.
Literature cited
Brown, James K. 1974. Handbook for inventorying downed woody material. USDA Forest
Service General Technical Report INT-16. Ogden, Utah. 24 p.
Brown, James K., R.D. Oberheu and C.M. Johnston. 1982. Handbook for inventorying
surface fuels and biomass in the interior west. USDA Forest Service General Technical
Report INT-129. Odgen, Utah. 48 p.
Clark, Donald F., P.J. Burton and J.A. Antos. 1995. A comparative study employing
different methods for the inventory of coarse woody debris. Symbios Research and
Restoration, Smithers, B.C. 7 p., App.
Harmon, Mark E. and J. Sexton. 1996. Guidelines for measurements for of woody detritus
in forest ecosystems. Publication No 20, U.S. LTER Network, College of Forest Resources,
University of Washington, Seattle, Wa. 73 p.
Howard, James O. and F.R. Ward. 1972. Measurement of logging residue - alternative
applications of the line intersect method. USDA Forest Service Research Note PNW-183.
Portland, Oregon. 8 p.
McRae, Douglas J., M.E. Alexander and B.J. Stocks. 1979. Measurement and description of
fuels and fire behavior on prescribed burns: a handbook. Canadian Forestry Service Report
O-X-287. Great Lakes Forest Research Centre, Sault Ste. Marie, Ontario. 44 p., Appendices.
Pickford, Stewart G. and J.W. Hazard. 1978. Simulation studies on line intersect
sampling of forest residue. Forest Science 24(4):469-483.
Trowbridge, R., B. Hawkes, A. Macadam and J. Parminter. 1989. Field handbook for
prescribed fire assessments in British Columbia: logging slash fuels. FRDA Handbook No.
001. 63 p.
Author: John Parminter
Submitted by: John Parminter
Peer review: No
|
