I am helping foresters in the tropical forests of the Yucatan Peninsula with the analysis of inventory data. They use strip plots but often have relatively few in an inventory so their t value for a confidence interval is large and the confidence interval for volume is large also. They are considering changing to systematically arranged fixed area plots and would like to keep them relatively close. That creates the question - how close?
If the altitude level varies even adjacent plots will also be fine. However, it would be better to choose the plots at least 100 - 200 m apart. A preliminary survey of the area could be done and the plots could be chosen in a way so that they reflect the real species richness of the area.
Thanks. The Yucatan peninsula is very flat. there are little high spots but many are 40 sq m in size. The forests are very dense with undergrowth. 100 m is a fairly long way in such a thick forest but GPS should help.
Dear Carl,
To ensure your design won't make more costs for systematic inventory rather than strip sampling, you can follow this procedure:
Calculate the whole area which is typically measured by strip sampling (total measured area: TMA). Determine the proper size for the plot samples (SA). Dived TMA by SA to calculate number of samples. Dived total forest area to the number of samples. The resulted area is the area that you take one sample from as a representative sample. Now you just need to decide about the dimension of the grid. In flat area, square area can work well. So if you just calculate the square root of the area, you have the proper dimensions for your systematic design. In this way the sampling intensity will be the same for both methods.
Thanks for the thoughts but the same sample area in plots will take much more work than the area in strips as moving through the forest is slower than in more open forests and with strips they measure trees as they move along.
I like a square design but it is the least efficient in terms of time spent walking compared to a more rectangular array with more distance between lines and less between plots.
Indeed, linear sampling designs (line, belt, strip, road, track) maybe efficient, but tricky. Firstly, stratification of the area of interest may help, with spectral (RS) or with environmental strata (e.g.terrain, substrate, human impact); the preliminary survey of Gautam above. Secondly, your sampling may be combinations of linear (systematic) and random components. Thirdly, sampling design depends on the variables of interest (botanical composition; timber volume/carbon stock?). Attached a number of forest sampling designs with stratification and linear components for your consideration.
Article Sharing natural resources: Mountain gorillas and people in t...
Article Invasion of an alien shrub across four land management regim...
Article Carbon Stock of European Beech Forest; A Case at M. Pizzalto, Italy
See the link below for some excellent information on local spatial structure. In Honduras, I have used 1000m between lines and 500m within, but those were pine forests. In the wet forests of the Yucatan, 100+ m will tend to behave as if independently sampled. You can then analyze later using both simple random sampling and as cluster samples (rows as the primary sampling units) to check for the difference in resulting variance.
http://www.biogeosciences.net/11/6827/2014/bg-11-6827-2014.pdf
chip. 100 m might be doable. they could enlarge their plots. variability is very high, at least species. in 2.5 ha of forest there are about 100 species of trees. the forests are relatively dry (for tropical forests) with a history of usage for slash and burn that goes back 2000+ years plus hurricanes and the odd large fires that often follow hurricanes.
It is better to estimate the minimum size of the sample (quadrat). After determining the size of the sample, determine the number of the sample required to inventory the forest. Multiply the number with the size of the sample. This will give you the area required to sample the forest for proper inventory.
You can distribute the samples either at random or as per randomized table. You can also distribute the samples as semi-controlled random number.
For determining minimum size and number required to be sampled, you can refer any standard Ecological or Forest ecology manual or field book.
thanks and i agree with you. unfortunately, the people tend to take a 1 or 2 percent cruise by tradition. it is often too small to achieve required precision but they accept that for economic reasons. also, random location in a tropical forest is much more work than in temperate forests. they usually do strip plots as they can measure trees while moving through the forest. it is more efficient. though they have been receptive to the idea of taking larger plots at fixed distances - systematic sampling along lines.
- Badges
- Science topic
- Similar topics
- Correction