[ai-geostats] Sample area size
I am new to this list (and the procedures), so please excuse possible unprofessionalities. This is my question/problem:
I am studying the semi urban area around a Flemish city. Goal is to study certain properties and behaviour of landuse/landcover variables and landscape metrics. The study area is the area around the city, inside a circle with the cities market square as centre and a radius of 10km. Inside this study area, samples should be taken: small "test area's" of about 25 hectares (500m x 500m). The size of these area's were determined by multiplying the larges possible homogene land unit (in most cases a parcel) by five.
First idea was to calculate the sample size for the study area and then distribute that amount of test area's (25ha each) randomly inside this circular study area. Because the study area (area inside circle) is of a very heterogenous landuse/landcover however, it is possible that when the test area's are distributed over the study area, some types of LU/LC will not be covered (and thus conclusions drawn from experiments with randomly distributed test areas will not provide a sound description/result). An alternative would be to lay a grid over the study area, where every "point" in this (symmetrical) grid actually is a test area. A sort of PSU/SSU approach, with the difference that every PSU does not contain seperate SSU's, but a whole test area (of 2ha). And now, finally, the questions:
- Is this the right approach?
- If so, how do I calculate the sample size (the number of necessary test areas = the amount of 25 hectare-squares needed)?
- If so, what dimensions should that grid have (how far should every test area lie from its neighbouring test areas?
I hope somebody can give me more information on these problems, or can direct me to information sources where I can find suitable answers for these problems.
Looking forward to some reaction, many thanks in advance