Fuzzy continuous classification and spatial interpolation in conventional soil survey for soil mapping of the lower Piave plain

Soil maps on a reconnaissance or semi-detailed scale are frequently used as the basic level of information for regional land use assessment. However, they are impractical for suitability predictions concerning quality crops, which require an enlargement of details that can be guaranteed only by detailed soil survey and mapping. The change of scale increases the importance of field observations and introduces in the soil-landscape model the fuzziness related to surveyorʹs subjective judgement. The present study explored the applicability of the fuzzy c-means approach to conventional soil surveys for detailed land use assessment with the purpose of taking into consideration the fuzziness of data collected in areas where soil type transitions are not easily observable on the surface. The investigation was done using data from 1888 auger borings of a 42 000-ha survey carried out in the lower part of the fluvial plain east of Venice on the 1:25 000 map scale. The morphological quantitative attributes of the master horizons A, Bw, Bg, Bk and C were first reduced in number so as to remove attribute redundancy. They were then submitted to fuzzy c-means, testing cluster validity with both internal and external criterion measures. The best partition was obtained with four classes, which were consistent with the main soil-forming processes of the area not only in terms of class centre features, but also of spatial variability and distribution pattern of the class memberships. The investigated approach might represent part of a complex analytical sequence capable of using both the spatial distribution of fuzzy continuous classes and the landform analysis to formulate a spatially detailed territorial model representing the basic level of information for a locally oriented land use assessment.