Where have the geomorphologists gone?
- May 2
- 4 min read
An academic competition asked the question “How do we get better at long-term coastal (morphodynamic) modelling?” My response was to express the opinion we’ve gone too far down one pathway, while substantially neglecting several areas of science, notably geomorphology.
This opinion is built from watching the transition of science from the 1980s through to early 2000s, when the attempt to move ‘short-term’ modelling into ‘longer-term’ concepts was most heavily open to discussion. Most people who purport to have ‘solved’ this simply demonstrate cases where the short-term approach should simply work a bit better. The dialogue had in 2013 as part of ICCOAST (organised by Robert Nicholls & leading to the paper by French et al. on appropriate complexity modelling) did nothing to convince me that anything had changed meaningfully since the 1990s… At a similar time, a gathering organized by Bruce Thom in Adelaide focused on an unspoken question “How do we keep geomorphology relevant?”
At the time, a couple of critical transitions were occurring in coastal modelling:
Computing power began to meaningfully support bed-updating models. This naturally prompted the ‘morphological factor’ concept, where modelled change could be magnified by a factor to allow for time scale.
There was substantial model convergence and unification, particularly through directly interfacing modules. For many commercial models, provision of diverse applicability met the inertia of model development, to match different numerical schemes and spatial frameworks. A consequence was that many model packages only contained a limited number of algorithms, often using simplifications to provide numerical efficiency.
The resultant ‘seamless’ modelling substantially reduced how much morphology was used to guide model selection, and model ‘validation’ increasingly moved towards comparison of short-term observations against hydrodynamic module outputs. Morphodynamic modelling outputs, if observations of change were available, was often open to scaling, or based on bulk behaviour, limiting evaluation of how well different active processes are resolved. At the unfortunately common worst, this resulted in propagation and amplification of initial condition responses.
Many of these limitations can be intrinsically addressed through good modelling practice, where each module is carefully assessed. Over the last two decades, this has become increasingly plausible, as model development focus has gradually shifted from module interaction towards wider applicability. Simultaneously, increased computing power has supported higher resolution model grids, which has reduced issues related to discrete numerical representation.
However, despite continuing computing advances and, more hopefully, improving modelling practices, there are fundamental limitations of numerical models to represent long-term morphodynamics. Many hints towards pathways for improvement exist in geomorphic studies, particularly those before withering of the relationship between geomorphology and coastal modelling.
One ‘obvious’ aspect is identification of features where morphodynamic feedback or compensation can offset drivers to response (e.g. beach cusps are a widely observed example). These will distort practices such as morphologic factoring, and often this turns sediment transport behaviour into a function of sediment supply, rather than forcing conditions. Models tend to work ‘best’ where there ae fewer feedback / compensation mechanisms. It is partly for this reason that simple models often given just as meaningful outcomes as a higher resolution representation.
This is partly echoed by model focus on drivers, rather than form. We typically treat sediment transport as a function of drivers, with form as a response. However, there are clear cases that form contributes as much to transport as the driver: i.e. once a body assumes a form, it will continue to support transport (while it has supply). This behaviour is most clearly illustrated on sand spits and migratory bars.
Although sedimentary form, commonly described by slopes or bedforms, is well known to influence sediment transport, it is rarely incorporated into modelling. This is partly because most sediment transport algorithms were based on horizontal bed testing, but also because of challenges representing bed slopes from grid points, especially using triangular grids.
The capacity for sediment storage to occur (or not) is often critical for longer-term dynamics, especially if interactions with dunes & estuaries are influential. This is rarely well represented in coastal modelling, partly because storage behaviour has limited response to drivers. Observations in reef sheltered or headland-controlled coasts demonstrate the importance of short-term storage due to sheltering, as well as sediment transport observations following tropical cyclone impacts, where there are acute shifts in ‘storage’ locations.
There are many sub-scale features or processes (time or space) that can influence sedimentary response to drivers. Examples include bed forms, bed surface winnowing, sand waves or migratory bed forms. Representation can be diverse, and require non-standard application of models, typically developed through supplementary evaluation of a modelling system.
One of the primary examples of temporal sub-scale behaviour is evaluation of groyne or headland bypassing. Although switching between concepts of net and gross transport may help getting closer to matching real behaviour, even then it is a very scale dependent proposition. A second example is provided by sand waves, which typically develop during an acute supply phase and subsequently act as discrete sand masses, often with feature scale and time scales of change much faster than applicable for wider coastal dynamics.
Overall, significant advances have been made how we use numerical modelling systems to evaluate long-term coastal change. However, there are parts we will consistently get wrong, or find extremely difficult to simulate, until we expand our practices to consider:
Sediment supply pathways and variability.
Bathymetric form and bedforms.
Migratory bed features.
Sedimentary storage.
The scientific field of geomorphology, which has become increasingly abandoned in numerical modelling practice, offers some pathways towards improvement!
An example of a site where sediment dynamics are crucial, but it's extremely difficult to simulate using numerical modelling.
Comments