5. Discussion
The central themes we have explored so far are how the spatial rainfall pattern influences the channel profile morphology, and how temporal changes in rainfall pattern affect erosion rates and profile morphology during periods of transient adjustment. We have detailed the expected response to along-stream variations in erosional efficiency caused by spatial rainfall gradients according to the SPM and have shown how the transient response to a change in rainfall pattern is fundamentally different from a spatially uniform change in rainfall. A change in rainfall pattern will always result in spatially variable changes of erosion rates that also change with time during the transient response. In some circumstances a given location may, over time, experience both elevated and reduced erosion rates (and channel steepness values) relative to equilibrium in response to a single change in rainfall. It is important to keep in mind, however, that the nature of transient response depends strongly on the initial conditions at the time of the change in rainfall pattern. Therefore, there is a complex relationship between the transient response at any given location or time and both the change in mean rainfall and the final rainfall pattern. In the following discussion, we focus on highlighting some implications for the different expectations that follow from changes in rainfall pattern and discussing examples where conventional expectations based on spatially uniform changes in rainfall can potentially lead researchers astray. Where possible, we attempt to identify additional information or strategies that may be leveraged by future studies.
5.1 Revising Expectations for Erosional and Morphological Responses to Changing Climate 5.1.1 Relative Nature of Erosional Response
To this point, our choice of a steady state initial condition with spatially uniform rainfall has been convenient, as have been the terms top-heavy and bottom-heavy to describe typical orographic rainfall patterns. While idealized, this provides an intuitive starting point for understanding how more complicated – but almost certainly more realistic – climate change scenarios might play out. Recall, according to the SPM, transient climate-driven changes in erosion rate are dictated by a relative change in discharge. Where discharge is increased, erosion rates increase in response and river gradient declines toward a new equilibrium steepness; thus, a river subjected to an increase in discharge can be considered locally, if transiently, oversteepened relative to equilibrium, and vice versa. As we have shown, because discharge generally accumulates non-linearly downstream within a river basin, a change in rainfall pattern can create circumstances where the relative change in discharge inverts along the river length – at position xsc – producing a complex transient response (Figures 2 & 3). This implies that the river is simultaneously oversteepened and understeepened on either side of positionxsc . These transient states dictate whether erosion rates initially increase or decrease following the change in rainfall, respectively, not whether the new rainfall pattern is itself top-heavy or bottom-heavy, and the positions of these transient states shift throughout adjustment.
The nature of landscape response to relative changes in discharge implies, for instance, that relaxation of a bottom-heavy rainfall gradient can cause a complex transient response resembling a change from uniform to top-heavy rainfall patterns. That is, a weaker bottom-heavy gradient is relatively top-heavy compared to an extreme bottom-heavy gradient; similarly, a gentler top-heavy gradient is relatively bottom-heavy compared to an extreme top-heavy gradient, and vice versa (Figure 7). Thus, for example, in the case of a change in climate that causes an extreme bottom-heavy rainfall gradient to become less bottom-heavy and results in a complex transient response (e.g., Figure 7a), rainfall and erosion rate are expected to increase in the headwaters of the catchment and decrease near the outlet as seen for Case 4 (uniform to top-heavy). This response is not consistent with expectations for any uniform increase or decrease in rainfall, even if such a shift accurately reflects the change in mean rainfall. Therefore, neither the final rainfall pattern alone (i.e., modern observed pattern) nor accurate inference about the relative change in mean rainfall (wetter or drier) necessarily allow a robust prediction of changes in erosion rate within a catchment following a change in climate where rainfall patterns have changed significantly.
Interestingly, changes in climate do not need to involve extreme changes in rainfall patterns (e.g., reversal from top-heavy to bottom-heavy), or to occur over short timescales to drive complex transient responses. Indeed, even subtle changes in rainfall pattern potentially driven by minor, commonly occurring variations temperature and atmospheric conditions (e.g., Mutz et al., 2018; Roe et al., 2003; Siler & Roe, 2014), may induce complex responses and significantly, if temporarily, alter the spatial pattern of erosion in a catchment (Figures 3 & 7). Indeed, such climate changes may have occurred in the Peruvian Andes and eastern-central Himalaya in the transition from Pliocene to Pleistocene climates, the latter represented by Last Glacial Maximum conditions (LGM; Figure 7c & 7d). Even if rivers in each of these ranges were in a transient state during Pliocene time, any adjustment toward equilibrium with the Pliocene rainfall pattern that occurred would then be in disequilibrium with the Pleistocene (LGM) rainfall pattern, and would have driven a complex response.
As transient adjustments proceed relatively more rapidly where rainfall is more concentrated (i.e., erosional efficiency is higher), changes in rainfall pattern have the potential to produce spatially distinct effects different from what would be expected from considering uniform changes in mean climate. Transient adjustments may therefore be relatively enhanced or underdeveloped in different locations within the same catchment, or adjustment to quasi-equilibrium may be essentially complete in some locations while others reflect only an incipient response to the climate change. We noted an example of this behavior in Case 4, where low-elevation dry tributary catchments preserve transient conditions the longest, contrasting with the notion that headwater catchments should be the last to equilibrate. Similarly segregated conditions occur in Case 3, where adjustment to quasi-equilibrium is essentially complete in wet low-elevation catchments long before the migrating trunk knickpoint even reaches drier high-elevation catchments. Because such complex, climate change-driven landscape adjustments are not reasonably captured by a conceptual framework based on spatially uniform changes in rainfall (e.g., compare Cases 3 & 4 to Cases 1 & 2), apparent inconsistencies between expectations and observations have the potential to give a false impression about the primary forcing(s) controlling erosion rates.
Additionally, if large-scale changes in rainfall patterns like we model develop incrementally over long timescales (e.g., millions of years), they could still result in complex transient responses. Greenhouse-icehouse transitions and orogenic growth are among many geologically significant events that may cause temporally distinct, sustained, and dramatic changes to climate and/or circulation patterns where complex responses could arise (Mutz et al., 2018; Poulsen et al., 2010; Roe et al., 2003; Zachos et al., 2001), If, for example, the bottom-heavy gradient in Case 3 instead develops over several million years, regardless any added complexity to the general trajectory of this change in rainfall pattern, the result is that it supports a 30% increase fluvial relief despite also increasing total rainfall by ~80%, and channel steepness patterns fundamentally change as the catchment adjusts. Gradual changes in rainfall patterns cause morphological adjustments to become more diffuse, and induce relatively smaller transient changes in erosion rate than abrupt changes. However, the spatial pattern of erosion is still significantly affected (i.e., in excess of a factor or two from steady state) so long as the timescale over which the rainfall pattern evolves does not far exceed that of the catchment adjustment timescale, which may be several million years for large river basins (Roe et al., 2003; Whipple, 2001). As such, the general characteristics of the classes of transient behavior following changes in rainfall pattern toward relativelybottom-heavy or top-heavy conditions remain intact, even for long-term transient responses.