I am going ro randomly select 4 trees in a sampling site. I know that the most common way to randomly select a tree is to give a number to each tree and than to use a random number generator. But sometimes the number of trees in a forest can be huge. Is there any trick to make the random selection faster and simplier?
You can select random geographic coordinates within a given study area. Yet, if there are too many trees at that random possition, you can then number the available trees within a given radius from this possition and thus select a tree with a random number generator. Hope this help :)
That method sounds good. Random selection is not always the way to go though. For example, if you are studying fruit production, random selection may result in no trees with fruit or male trees. You may just want to select suitable trees :)
The main thing is to be random, pick a random coordinate, get to that (about), then when you are about there, I pick a random direction and distance (1-30m or something like that) and then the next tree of your specified size in that direction.
Hope that helps
Colin
A caveat and a caution
A caveat: Picking an area randomly and then finding the closest stem is not exactly the same as picking a stem randomly by numbering them (in the latter each numbered stem has an equal probability of being selected).
To see why it isn't the same imagine a patchy forest where some areas are dense with closely packed stems while other areas are more open with a few widely scattered stems. Individual stems in the sparse areas are much more likely to be selected by choosing an area than are the individuals in the denser areas. Even if the per-stem differences in area are less pronounced they always occur in natural forests.
It may not matter, but sometimes it does. It will depend on your purpose. To provide guidance we would need to know more.
A caution: The first answer above is wrong. It confuses the admittedly unfortunate terminology of computer classification methods used in machine learning with real biological terms (forests and trees don't mean real forests and trees that biologists measure). Don't be confused!
It is important that you consider the distribution of the tree in the forest. This is a primary factor in selecting the appropriate method. Please, tell me this information to help you.
I followed the same path as Víctor Arroyo-Rodríguez's suggestion when finding random trees for comparison in a sleeping site selection paper. It worked well and was significantly faster than numbering all possible trees and randomly choosing.
You could try cluster sampling. That is a well researched method with variance estimation and lots of discussion. For example, see
Cochran(1977), Sampling Techniques, 3rd ed, Wiley,
and/or
Lohr, S.L(2010), Sampling: Design and Analysis, 2nd ed., Brooks/Cole.
One-stage cluster sampling says you divide an area into segments (called clusters). You select a set of them by simple random or some other good sample scheme, and you census the trees in the clusters selected. Or you could do a second stage and sample within the clusters.
But if you want to know what someone in forestry does, you could use the share button on your question to invite some to answer your question (or not).
Only 4 trees? Well, I guess this can be done on a small scale. Perhaps use a relatively large number of small clusters. Select four clusters using simple random sampling, and one tree from each of those four clusters. With only one tree in each cluster, the variance is infinite (undefined). See
https://newonlinecourses.science.psu.edu/stat506/node/44/
- Badges
- Science topic
- Similar topics
- Filing