Questions from the Field:
When is there “an Edge” Requiring a Correction?
This is a really good question. The recent “walkthrough method” is a good correction for “edge effect”, but when does it have to be used? When does “an edge” exist? If you are doing logging that leaves patches of trees (and therefore do not want to measure them in your sample) then is there an edge around them? What if individual trees are left? Do strata boundaries count?
You have “an edge” whenever you have bordering areas where:
1) you would not put a plot, or
2) you will sample differently.
If you would not put a plot into a small swamp, or into a patch of trees that will be left, then there is a border around that area that constitutes an edge. The “walkthrough” method is a great way to correct for this, but there are other methods we have explained in earlier articles. In general, whenever you are putting plots into an area that borders another area where you would not put plots, an edge exists where correction methods should be applied. You have to be able to recognize and measure that area, then delete it when you multiply area times volume per acre to get total volume. That can be a real hassle.
If you do not put a boundary around those retained trees, choosing instead to record a zero or lower tree count by ignoring those trees, no edge correction is necessary. You might have a slightly more variable answer, but the area measurement hassle is probably more trouble than putting in a few extra plots to compensate for any extra variability.
Strata boundaries which are both sampled the same way are a bit different. If you do not mind an occasional Douglas‑fir in a pure hemlock strata, you do not need to treat the edge differently. Plots along either edge will get an occasional tree from the other strata, but these will balance in the long run and there is probably no cause for concern.
If you absolutely do not want to mix the results of several strata, you can treat each strata edge as a boundary, correcting whenever you get close to it. Is this necessary? That’s for you to decide. In many cases it is not. Just be prepared to explain when a sampled plantation of planted apples with a fence around it has an occasional fir tree recorded (and the same with a few apples in the fir stand).
If you are using different sampling methods on adjoining patches of ground, then it makes sense to treat the borders of the areas as edges needing correction when plots fall near them. If you are doing 3P sampling in the riparian zone and Variable Plot sampling for the rest of the area, then you should correct the Variable plots near the riparian zone boundary. No edge correction is needed with 3P sampling, of course.
What if you are doing Variable Plots with different BAFs? Well, there is a very small technical bias in a case like this, but it is really trivial and can be ignored for practical purposes.
We want to thank Norm Shaw of BCIT in Vancouver BC for bringing forward this interesting question, which came up when his class was doing a project on cruise system comparisons.
Originally published November, 2004
Back to Contents