Combined mapping of soil properties using a multi-scale tree-structured spatial model

Abstract

Accurate maps of various key soil properties on fine spatial resolutions play an important role in precision modeling of agricultural systems. Recent development of alternatives to soil coring enables us to collect multiple sources of data, but data quality and spatial resolutions may differ greatly from one source to another. In this article, we use a multi-scale model for combining all data sources, despite varying resolutions and accuracies, to produce soil maps. We demonstrate that the method gives accurate results via computer simulations. Using the multi-scale method, we combine soil coring, penetrometer, and topographic data to map the depth-to-till on a 10-m resolution in an Arlington, Wisconsin field, and combine soil coring and soil electrical conductivity measurements to map field capacity on a 20-m resolution in a Waunakee, Wisconsin field. The proposed mapping technique has several advantages: (1) it is computationally fast and hence is well suited for landscape modeling; (2) it provides a means to combine more than two sources of data; and (3) it provides a way to accommodate prior knowledge of spatial dependencies associated with various data sources.