Designing efficient soil survey schemes with a knowledge-based system using dynamic programming

Soil survey as a knowledge system

“…Recognition of soil survey as a cognitive process that can be modeled and used to predict soil -landscape relationships makes it possible to use 'intelligent systems' modeling to design sampling strategies (Domburg et al, 1997), to represent the process formally (as proposed below), and rule-induction programs to automate it as suggested by Lagacherie et al (1995) and further demonstrated by De Bruin et al (1999), MacMillan et al (2000, Wielemaker et al (2001), or Bui and Moran (2003).…”

Soil mapping, classification, and pedologic modeling: History and future directions

“…Acquiring the samples to provide the spatial support for developing better soil mapping models is time and resource intensive, but the more it is done, the better the maps will be (Domburg et al, 1997). Also, without fieldwork the exceptions to the current understanding of soil landscapes cannot be discovered.…”

Instance selection and classification tree analysis for large spatial datasets in digital soil mapping

“…In this study on soil class prediction we apply proportional stratified random sampling (PS) and disproportional stratified random sampling schemes (DS) as instance selection approaches (Gu et al, 2001). The soil-sampling counterparts of these two approaches are well-known in spatial soil sampling design aiming to improve estimates (Brus, 1993, Domburg et al, 1997.…”

“…Even though some methods of classical spatial soil sampling designs (Cochran, 1977;Brus, 1993;Brus and de Gruijter, 1997;Domburg et al, 1997) and instance selection (Gu et al, 2001) are based on the same theoretical concepts, the aim of instance selection is contrary to the aim of soil sampling. In soil sampling it is most important to optimize sampling schemes to derive a sample set that is as sparse as possible to save labor and lab costs.…”