Survey of zone tessellation techniques for defined parameters in precision agriculture

Precision agriculture is a trending topic since the last few years. Use of technology and advances in science helps in getting a better yield of crops in a given agricultural area. Novel ground based sensors along with aerial and satellite imagery provide large geo-referenced data sets with high spatial resolution. Management zone delineation is a key task of precision agriculture. From a data-mining point of view this comes down to variant of spatial clustering which has a constraint of keeping the resulting clusters spatially mostly contiguous. Existing spatial clustering algorithms are used to delineate into so-called management zones. However none of the standard clustering algorithms would allow tackling the task on given data set. We discuss a few algorithms for a given data set which returns mostly contiguous tessellation and which can be parameterized easily by human expert. Various attributes obtained by applying different algorithms will be compared. Results on available multi-variate data sets and subsets will be presented. We conclude which is the best algorithm for the given data set.