The use of soil maps is often questioned by non- soil specialists, those who are not sufficiently aware of the nature and complexity of soil, and the survey procedures, that is, to map soils in an economic and affordable way. As reaction to this and also thanks to the advances in the fields of remote sensing and GIS, a new paradigm, called digital soil mapping, is emerging, where the emphasis is focused on soil attributes, assuming that these are continuously varying in space. With a glance to some of the basic definitions and concepts, concerning ‘what is a soil?” and “what is the content of a soil map?”, we believe in the concept that soils are 3-D natural bodies, and that the soils of the unmapped areas can be mapped using the scattered available soil data of the neighbouring areas. The objective of this study is to highlight the complexity of the soil and its mapping while applying some of the geopedologic-oriented ideas of digital terrain modelling to soil mapping. The attempt takes place in the Upper Pasak river basin in Petchabun, and in a smaller area (Doi Ang Khang Royal Agricultural Research Station) in the Fang district in Chiang Mai province, Thailand. Under the influence of past climate fluctuations, tectonic movements and lithologic-structural controls the Pasak valley has been formed banked by a few terraces and a series of hills, and glacis surfaces, of which the middle level appears scattered. Doi Ang Khang area with a different geologic/ geomorphologic setting (dominantly mountainous) was chosen as another case study. Various subgroups of Entisols, Mollisols, Inceptisols, Alfisols, and Ultisols (Fluventic, Arenic, Aquic, Aeric, Ultic, Ustic, Vertic,…..) occur in different surfaces, following the physiographic set up of the landscapes.

Predictive mapping in a conventional way (the ITC approach, based on image-- mainly air photo-- interpretation and on mental landscape-soils' models) was tried in the PaSak area, and validated by quantitative soil pattern analysis (eg., heterogeneity, size and shape indices). On the other hand, the automated geopedologic approach, based on parameterization of the soil forming factors (clorpt) and their integration through applying decision trees was tried out in the Doi Ang Khang area, followed up by a statistical validation. Some of the shortcomings of both the conventional ITC approach and the automated geopedologic approach are dealt with. The conclusion, in short, is that predictive mapping is recommendable, but it can only be carried out by experienced soil surveyors, as a thorough knowledge of soil geomorphology -- understanding of soil spatial composition and the relations with landforms -- is a must.

Key words: Digital soil mapping, automated Geopedology, soil geomorphology, Upper PaSak, Doi Ang Khang, Thailand