Sample planning for quantifying and mapping magnetic susceptibility, clay content, and base saturation using auxiliary information.

Author: Daniel de Bortoli Teixeira, José Marques Júnior, Diego Silva Siqueira, Vinicius Vasconcelos, Osmar A. Carvalho Júnior, Eder Souza Martins, Gener Tadeu Pereira

Keywords: Sampling density, Geology, Geomorphology, Soil class, Geostatistics, Simple kriging with varying local mean



There is a great global demand for detailed soil property description; therefore, an ideal site-specific sampling has become indispensable to meet this demand. This study aimed to assess the implications of incorporating geological, geomorphological, and pedological information in reducing the required sampling density for magnetic susceptibility (MS), clay content (CC), and base saturation (BS) characterizations. The study area is located in Guatapará-SP (Brazil) and has 870 ha. A total of 371 samples were collected at a depth of 0–0.25 m for assessing magnetic susceptibility (MS), clay content, and base saturation (BS). A density of one sample was considered every 2.6, 3, 4, 5, 6, 7, 8, 9, 11, and 14 ha. The incorporation of secondary information in a geostatistical framework was performed by means of simple kriging with varying local means. Accuracy assessment of the spatial estimates at each sampling density, with and without incorporating secondary information, was performed by using external validation. For MS, geology and geomorphology information were responsible for about 45% and 44% spatial continuity, respectively. As for CC, these results were higher, being of 54% (geology) and 53% (geomorphology). Conversely, no spatial variability was detected for these properties by using pedological information. For BS, there was no relationship between secondary information and its spatial continuity. Incorporating geological and geomorphological information to MS data enabled a reduction in the number of samples required of 37% and 44%, respectively, in order to represent its spatial pattern. Likewise, this information provides a 35% reduction in the required sampling density for CC. However, secondary information was no helpful in decreasing sampling density for BS. In brief, incorporating pre-existing information can ensure a high quality of estimates and decrease the number of samples required for a detailed description for both MS and CC. Estimates of spatial patterns with geological and geomorphological information for modeling of soil properties might have a greater potential of use for environmental model composition.