User: | Open Learning Faculty Member:
For background, I am looking at the density of a particular “bunchgrass” in a field on hilltops, slopes, and troughs, to see if there is any correlation.
For this initial sampling, I used simple random sampling. I made a grid over a map of my field in Photoshop, and randomly generated Cartesian coordinates, throwing out pairs that fell on the highway or residential areas adjacent. I took these locations and found their coordinates on Google Maps. I then made a 2ft x 2ft quadrat out of PVC piping I had around and used that to count the number of plants per square at the locations I generated.
I would have liked to have used stratified sampling, in order to get an even distribution of the three types of terrain I am looking at. However, since hills and slopes are not easy to see on a map, with the exception of a few very large troughs, I found this to be infeasible.
I intend to continue using this method. However, in order to obtain enough of each terrain type, I intend to throw out coordinates of terrains for which I have ten samples already. This means that I will need to generate more locations than I will actually use, but it will save me time in the field and ensure that I do not end up with, say, 20 hilltops and 2 slopes in my final data.
My data is about on par with what I would have guessed. There tended to be fewer or no plants in troughs (0-3). There does not yet appear to be any difference between hilltops and slopes (5-6).
