Landscape evolution in the south Yilgarn Craton and Albany-Fraser Orogen, Western Australia

[摘要] Landscape evolution is the result of the interaction of climatic conditions, geological characteristics and sedimentary dynamics through time. In regolith-dominated terrains (RDTs), landscape morphologies and their stratigraphy record the 3D architecture of the overburden. The geochemical relation of the surface, the cover and the basement geology is captured by the landscape stratigraphy.Remote sensing datasets such as Digital Elevation Models (DEMs) can be used to visualise the geomorphological features of the land surface. Combining different surface geometrical features can be used to classify landscape types. Therefore, DEMs can be employed to map landscapes over large geographic areas (e.g., geological province, country or continental scale).In this study we tested the conceptual variability of landscape types within the Albany-Fraser Orogen and south Yilgarn Craton. We used supervised machine learning algorithms based on DEM data and DEM-derived products (e.g., DEM Hillshade and Flatness Map), Google Earth and Bing satellite imageries and field observations. We assessed how landscapes can be classified upon their specific surface geometric features. We generated a map showing six conceptually different landscape types.The south of Western Australia (WA) is subdivided into three main geological provinces: the Yilgarn Craton (YC), composed of Archaean (~3.0-2.6 Ga) metavolcanic and metasedimentary rock suits, granites and gneisses trending SE-NW; the Mesoproterozoic Albany-Fraser Orogen (AFO) basically comprising Archaean to Proterozoic (~2.8-1.1 Ga) metamafic rocks and (meta-) granites, orthogneisses and metagabbros in a SW-NE trend; and the Mesozoic Bremer Region (BR) made up of Jurassic and Lower Cretaceous (~200-100 Ma) sedimentary rocks. Outcrops in these regions are rare due to intense weathering and the presence of widespread transported regolith cover increasing in thickness from the coast inland.In this geological context, a total of 3000 km W-E-traverses was conducted to true the computer-generated landscape map’s geometric surface features, i.e. the six conceptually different landscape types out of machine learning algorithms with field observations. A second traverse was generated, 250 km trending N-S following a palaeovalley, and two W-E traverses (~250 km in sum) crossing the palaeovalley. Google Earth imagery and the recording of stratigraphic sequences associated with the diverse landscape types were also used.The approach implemented in this project can be generalised to assist landscape mapping in RDT by the use of DEM surface geometry. This is a quick and low-cost technique for creating a first-pass landscape type map on a large scale. However, ground truthing is always essential. Landscape delimitation using this technique will be a powerful tool to map and feature landscape types in similar contexts such as West Africa, India, Brazil and large parts of China. Landscape mapping can provide a solid reference in mineral exploration for a better understanding of geochemical dispersion of the basement geochemical signatures through cover, by linking stratigraphic units to dispersion processes and surface geometries.

[发布日期] 2018-05-11 [发布机构] CSIRO

[效力级别] Geomorphology and Regolith and Landscape Evolution [学科分类] 地球科学(综合)