Basin Relief and Hypsometry Indicate Rapid Erosion in the Badlands of the Active Fold-Thrust Belt of Southwestern Taiwan

The badlands of southwestern Taiwan lie within an active fold-and-thrust belt, where surface geology mainly consists of a thick Plio-Pleistocene mudstone formation. In the absence of fluvial markers of river incision, we investigate relief and hypsometry within the badlands as potential proxies for long-term (0.1-1 ka) tectonic uplift. In parallel, hypsometric curves allow us to assess the balance between tectonics and erosion.

We selected three badland sites located at different structural positions, with decadal uplift rates from 12 to 55 mm/yr (continuous GNSS, levelling or traverse measurements), yet with similar lithology and climate. The badlands are of calanchi type, with unvegetated slopes and sharp ridges, low relief (<50 m) and short basin length (<200 m). Topographic datasets include high-resolution (2.5 to 10 cm) UAV-derived Digital Surface Models (DSM) at all sites and a 1-m LiDAR Digital Earth Model (DEM) at one site. Drainage network, divides and geomorphic metrics (basin relief BR, hypsometric integral HI and hypsometric curve) were extracted using ArcGIS and Matlab TopoToolbox, for 12 to 27 basins over areas of 5000 to 34000 m² from site to site.

At the fastest-uplift site, the 10-cm-resolution DSM and 1-m DEM lead to similar average BR (35 ± 7 m and 37 ± 7 m, respectively) and HI (0.49 ± 0.05 and 0.46 ± 0.05), although results for individual basins differ significantly for 20% of the 24 basins. Hence, even though crestlines and gullies are commonly narrower than 1 m, the 1-m LiDAR DEM mainly provides representative values for the investigated metrics. Results obtained from UAV DSMs at the three sites show no clear influence from decadal uplift or structural position. With increasing uplift of 12, 23, and 55 mm/yr, we respectively obtained average BR and HI of 27 ± 8 m and 0.50 ± 0.03, 23 ± 6 m and 0.48 ± 0.07, and 35 ± 7 m and 0.49 ± 0.05. Hypsometric curves fluctuate around a S shape at all sites, indicating a transitional stage with sustained uplift and erosion. A notable difference in the field is the thinner crestines and larger amount of clasts transiting along the hillslopes at the fastest-uplift site, indicating a larger production of clasts than run-off can transport. We interpret these results as erosion rates exceeding the already rapid uplift rates. This would be facilitated by the low erodibility of the mudstone formation. Indeed, a regional analysis based on a 20 m DEM shows that mean and maximum values of local slope and relief are lower in the mudstone domain than in siltstone and sandstone domains, in spite of active anticlines and larger decadal uplift being located in the mudstone domain. Ongoing complementary works on basin-wide denudation rates in several-km-long river basins draining the mudstone domains led to contrastingly low denudation rates of 0.8 mm/yr, indicating that badlands denudation either represents a different timescale or that other processes dominate denudation at the larger spatial scale.