For decades researchers have been studying forest soils and summarizing findings in the form of soil surveys with thematic soil maps depicting soil associations, broad polygons representing groups of individual soil types. With growing availability of high-resolution spatial data, it has become possible to model and map how individual soil properties vary, both spatially and with depth, across the landscape at high resolution. This dissertation demonstrates how this can be accomplished for the Province of New Brunswick (NB), Canada by way of digital soil mapping (DSM) based on (i) existing soil information and related data sets, (ii) principles of soil formation as dictated by location-specific changes in topography, surficial geology, and climate. For this purpose, existing elevation data sets were fused via error reduction procedures to generate a comprehensive province-wide digital elevation model (DEM) at 10m resolution. The resulting DEM was then used to delineate a variety of data sets detailing spatial variations in topography, hydrology, and climate. Various sources of spatial geology depictions were combined by way of similarities in classifications resulting in re-delineations of landform and lithological attributes. In combination, the data layers generated were used to determine how specific soil properties (n = 12,058) vary, both spatially and with increasing depth, across the province at 10m resolution. These determinations were made possible by way of machine-based random forest regression modelling. This dissertation provides details in terms of how (i) a province-wide soil database was generated from existing soil survey reports, (ii) how missing soil data were substituted through the process of pedotransfer function development and analysis, (iii) how the province-wide DEM layers were fused, and (iv) how the DSM procedure was formulated and executed. The soil properties selected for modelling and mapping purposes refer to soil depth, drainage, bulk density, texture, coarse fragment content, and soil organic matter content. In turn, these properties, in combination with spatial data sets (topography, geology, and climate), can be used to model and map other soil variables such as, e.g., pH, soil water retention at field capacity and permanent wilting point, and cation exchange capacity.
Zircon - This is a contributing Drupal Theme Design by WeebPal.
