In a nut shell: Latin Hypercube if cost and access are not restrictive. Otherwise conditional Latin Hypercube (CLH) and/or Flexible Latin Hypercube. Both allow for the selection of sites based on the proximity to roads or access points and provide alternative points when the ones selected are not accessible or are not suitable.

Generally LHC assures that for each category/class there are enough sampling points to represent the distribution of the population of all possible points within each category.

The caveat to this is that one has to be certain that all the categories are represented and that is tricky because it assumes that all the possible categories are indeed included in some way by either prior knowledge or some other means of analysis that still used some assumptions. Vicious circle, as you are trying to sample all the possible categories but we don’t know for sure what this categories are unless you conduct field reconnaissance, otherwise you depend on what you know and the sampling becomes self-fulfilling. Besides the number of categories/classes “depends” (don’t like to use this word but no choice) on purpose of map, scale and process in general.

Of course there are other sampling methods that can be implemented with tweaking.

Here is a great article from Budy and Alex on conditional latin Hypercube.

Hi Zohreh

There are many ways and approaches. Here are just one way ...: To divide an agriculture field in subareas for agriculture purposes I use geoelectric measurements (EC values ). This I analyzed by using a cluster analysis (K-Mean method). The classification of the EC-values in cluster classes form the subareas. These subareas are targeted sought/ select and pedological mapped on the traditionally way with an auger.

A correlation between EC value and soil type/ grain size distribution is given. Also a relationship between yield and soil type / grain size distribution exists.

Maybe the mentioned Literature will help you.

Best regards from Osnabrück, Germany

Article Geoelectric Measurements combined with traditional Field Map...

Conference Paper Part Field Management: Comparison of EC-value, soil texture,...

Conference Paper Use of the BOs-1EP for the low-sample estimation of the spat...