5.2.2 Misleading Transient Signals
Spatial patterns in erosion rate are commonly used to inform tectonic models and to infer rock uplift rates in mountain landscapes (e.g., Adams et al., 2020; Godard et al., 2014; Kober et al., 2015; Morell et al., 2015; Safran et al., 2005; Scherler et al., 2014). However, we have shown that changes in rainfall patterns can drive long-lived and complex spatial patterns of erosion that differ from expectations for uniform changes in rainfall and thus may not be readily recognized and interpreted. We have also shown that that ongoing transient adjustment may not be obviously expressed in landscape morphology (especially for catchment-mean metrics) under some circumstances. If these caveats are not considered, subtly expressed transient spatial variations in erosion rate may be mistaken as representing quasi-steady state spatial variations in uplift rate (i.e., E = U ). At once this would give a false impression both about the spatial pattern of uplift and the importance of past climate changes on a landscape’s evolution, with direct implications for understanding connections among climate, surface processes, and tectonics. Determining whether there are circumstances in which spatial patterns of erosion and topography produced by changes in rainfall patterns that can be misleading enough to confound interpretations about factors controlling landscape evolution is critically important.
During the early transient adjustment in Case 3 (transition to a bottom-heavy rainfall pattern) there is a clear example of how such confusion may occur (Figure 10). Recall, in this case, early transient adjustment produces a concave-up knickpoint along the trunk profile but as it migrates upstream the shape evolves. This creates a broad adjustment zone. Over the first ~500 kyr, quasi-steady state adjustment proceeds ~60% upstream along the trunk, but the broad adjustment zone means most tributaries along this length experience a protracted signal of base-level changes related to trunk adjustment. Because these tributaries all also experience a net increase in rainfall, knickpoints associated with local adjustment of the trunk river (Stage 2) tend to relax as they work upstream making them more diffuse. This protracted competition between local rainfall and spatio-temporally variable rates of base-level fall, generally results in diffuse concave to broad convexo-concave adjustment zones in tributaries (e.g., Figure 8a; Movie S5). Broad adjustment zones, particularly concave-up adjustment zones, are inherently subtle and this can inhibit their recognition. This problem may be further compounded by the influence of sediment flux in natural settings (Brocard & van der Beek, 2006; Whipple & Tucker, 2002). Indeed, even in our idealized model (i.e., no sediment influence), along-stream variations in trunk and tributary local ksn variation is diffuse (Figure 10c). Based on a lack of significant knickpoints that might indicate transient adjustment and the several-fold spatial variation in erosion rate, one might reasonably interpret relationships depicted in Figure 10a reflect a quasi-steady-state landscape adjusted to a spatial gradient in uplift rate. In the absence of known surface breaking structures that might accommodate this gradient in uplift, blind structures may be inferred, with potential implications for tectonic models. The apparent viability of this interpretation is supported by the SPM if rock uplift rate is assumed to match the observed pattern of catchment averaged erosion rates in tributaries (steady-state conditions) as illustrated in Figure 10b. Figure 10b shows that the predicted steady-state upstream-averaged ksnpattern along the trunk river and mean ksn values exhibited by the tributary network is essentially identical to the transient pattern in Figure 10a. Moreover, even in detail, there are only subtle differences in the along-stream pattern of localksn between the two scenarios (Figure 10c). Thus, in this instance, ksn patterns and erosion rates that actually record a complex transient response to a change in rainfall pattern could reasonably be mistaken for a steady state landscape adjusted to a spatial gradient in uplift.
Although subtle variations in ksn values might give a misleading impression that a landscape is in quasi-steady-state, the spatial pattern of ksn-q unambiguously suggests along-stream variations in erosion rate exist along both the trunk river and tributaries. ksn-q also exhibits a coherent pattern of downstream adjustment that could readily be interpreted as a transient signal sweeping upstream through the catchment that, significantly, is inconsistent with a steady state landscape adjusted to the spatial gradient in uplift shown in Figure 10b (Figure 10c). This example shows the potential usefulness ofksn-q , both as a diagnostic tool for detecting ongoing transient adjustment to changes in rainfall patterns whereksn­ may be misleading­­­ and for resolving the relative influences of tectonics and climate.
Finally, we emphasize that our intention is not to suggest all, or any specific examples, where spatial patterns of rock uplift are inferred from erosion rates and channel steepness patterns are incorrect. Rather, our intention here is to highlight the extent to which confusion may be possible under the right circumstances, and how explicitly accounting for rainfall patterns can be a step toward addressing these challenges.