Late Quaternary loess landscape evolution on an active tectonic margin, Charwell Basin, South Island, New Zealand

Loess deposits constitute an important archive of aeolian deposition reflecting wider patterns of glacial atmospheric circulation, and more localised interactions between riverine source areas, loess trapping efficiency and geomorphic controls on erosion rate. Conceptual models have been formulated to explain the coeval evolution of loess mantles and associated landscapes (loess landscape models) but none apply to areas of tectonically induced base-level lowering. This study uses an age sequence of alluvial fill terraces in the Charwell Basin, north-eastern South Island New Zealand, which straddles the transpressive Hope Fault, to investigate geomorphic controls on loess landscape evolution in an active tectonic region. We hypothesize that the more evolved drainage networks on older terraces will more effectively propagate base-level lowering by way of a greater areal proportion of steep and convex hillslopes and a smaller proportion of non-eroding interfluves. Eventually, as the proportion of interfluves diminishes and hillslope convexity increases, terraces shift from being net loess accumulators to areas of net loess erosion. We investigate the nature of erosion and the geomorphic thresholds associated with this transition. Morphometric analysis of alluvial terraces and terrace remnants of increasing age demonstrated geomorphic evolution through time, with a decrease in extent of original planar terrace tread morphology and an increase in frequency of steeper slopes and convexo-concave land elements. The number of loess sheets and the thickness of loess increased across the three youngest terraces. The next oldest (ca. 150 ka) terrace remnant had the greatest maximum number of loess sheets (3) and loess thickness (8 m) but the loess mantle was highly variable. A detailed loess stratigraphic analysis and the morphometric analysis place this terrace in a transition between dominantly planar, uniformly loess-mantled landforms and loess-free ridge and valley terrain exemplified by the oldest terrace remnant. Variations in thickness and preservation of loess sheets demonstrated spatially and temporally variable erosion during loess accumulation. To test our hypothesis of loess persistence we calculated critical steady-state hillslope curvatures from a soil transport model, calibrated for the study area, above which the uppermost loess sheet (L1, max. thickness 1.8 m) should be completely eroded. We compared loess distribution mapped in the field to values of slope curvature calculated from topographic surveys and found two of three critical curvature values had acceptable predictive ability. Where predictions failed this is probably due to transient responses related to active steam incision. Because all critical curvature values predicted presence of loess on the oldest terrace remnant where there is none we conclude that important factors other than morphometric ones are important in determining loess distribution in loess landscapes in active tectonic regions. These may include internal changes to regolith affecting erodibility, or vegetation or topographic interaction with wind patterns affecting loess trapping. (C) 2009 Elsevier B.V. All rights reserved.