Suffolk case study: Deben estuary and open coast
Composition definition and component models
The Suffolk composition comprised of a series of models linked, both at runtime (dynamically) and statically (offline). The MESO-i model was dynamically linked to 2 instances of SCAPE+, one representing Bawdsey and the other one Felixstowe, with MESO-i representing the mouth of the Deben estuary. The Deben estuary itself was modelled with ESTEEM and was linked statically, so that SCAPE+ was reading the files produced by the external ESTEEM model, which produced a time series of sediment coming into the beach system.
The 'Deben' composition comprises three models, driven by timeseries of waves, water levels, sea level rise and the (Deben) tidal prism. Central to the composition is a MESO_i model, which represents the dynamics of the inlet features around the entrance to the Deben estuary; known locally as 'the Knolls'. These features exist within the model as a set of distinct stores of sediment, which exchange material in response to inputs/demands from the neighbouring coasts, and through internal circulation. These dynamic also respond to the tidal prism of the estuary, which is represented by a timeseries input, generated by the estuary model ESTEEM.
The neighbouring coasts extending to the northeast (Bawdsey to Shingle Street) and southwest (to Cobbolds Point) are characterised in the composition with SCAPE+ models. These represent the beach, shore platform, cliffs and coast protection structures in these areas. Wave action (modulated by tide and sea level) drives sediment along the beach, and also erodes the cliff and platform, where exposed.
At the connection between the models (at every tide) SCAPE+ either delivers sediment to MESO_i, or else imposes a demand for sediment on it. MESO_i then responds by either accepting the sediment or supplying it in proportion to its internal state (i.e. if the Knolls are low in volume, they supply a small proportion of the demanded sediment). In this way sediment may pass into MESO_i from either of the SCAPE+ models, and circulate within it before passing back into a SCAPE+ model. The actual direction of transport at the model boundaries at any moment depends mainly on the angle of wave attack.
The composition requires inputs in the form of sea level change, tide levels and wave conditions (height, period and direction). Tide levels and sea level rise over the 20th Century were estimated using data from the nearby Lowestoft tide gauge. Earlier rates of sea level change were provided by Sarah Bradley, drawing on work described in Bradley et al, 2011. The composition required wave inputs at specific inshore points, and these were drawn from the outputs of Coastal Area Modelling (TOMOWAC) undertaken by the University of Manchester. The inshore data were generated through transfer functions designed to operate on offshore wave data generated by the Met Office with their WaveWatch III model (funded and provided by the Environment Agency).
Calibration of a SCAPE model normally involves setting model recession rates to match historic observations of cliff/shore platform retreat (see Walkden and Hall, 2005). For quasi-3D simulations with a dynamic beach, alongshore sediment transport rates must also be tuned. These two aspects of model behaviour (shore recession and alongshore transport) are normally linked, and so these process of calibration are interrelated; Walkden and Hall (2011) discuss this in some detail. The calibration of the SCAPE+ elements of the Deben composition was, to a degree, simplified by the disconnection between cliff recession rates and beach sediment volumes (eroded cliff/ platform material in this area is lost as mud rather than contributing to the beach). However it was made more complex and iterative by the dynamic model linkages within the composition.
Composition-scale calibration issues (which will be discussed further below) meant that only the first stages of calibration were undertaken for the SCAPE+ elements. This involved tuning the material strength parameter until cliff recession rates along the Deben frontage were broadly realistic. A precise match was not attempted, simply because of the likelihood that calibration (or even re-conceptualisation) elsewhere in the composition would subsequently undermine the initial SCAPE+ calibration.
Sediment transport was represented using the van Rijn equation (2014), which was considered to be more suitable for the coarse grained beaches in the area than the CERC equation (which SCAPE+ also represents). This equation requires the specification of a mean beach grain size, and a value of 6 mm was chosen following Burningham and French, (2015).
Calibration of the overall composition had to account for the behaviour of four models, which were built using three different tools, and which conceptualised their respective coastal systems in diverse ways. Tuning of any one model had the potential to change the other elements of the composition, and so calibration was treated as an iterative process. The composition structure suggested a sequence for this iterative process. The 'offline' linking of the ESTEEM model meant that its calibration (which is described above) could be undertaken as a largely separate process, and so this was the obvious place to start.
It was reasonable to address the calibration of the SCAPE+ elements next, as these largely determined the flux of sediment at the boundaries of MESO_i. As noted above, this involves tuning rates of both alongshore sediment transport and also recession of consolidated cliff/ platform. Both of these aspects had an impact on sediment fluxes at the boundaries of MESO_i; cliff/ platform recession at East Lane influenced the ease with which sediment could pass that location.
Calibration of MESO_i was undertaken last. The objective was for the composition to replicate the general cyclicity of the growth and breaching of the updrift shoal of the Deben inlet, the period of which has been relatively well-established. In the model this is influenced by: (1) the influx of sediment from the SCAPE+ models, (2) the Deben tidal prism (captured by ESTEEM) and also (3) an internal MESO_i diffusion coefficient, which changes the rate of flux between the MESO_i elements. The value of this diffusion coefficient was select to provide (approximately) the right average period between MESO_i breaching events.
Scenarios adopted for the Deben composition
The table above records the scenarios under which the composition was run. A core set of 24 were chosen to explore the consequences of variations in coastal/ estuarial defence and climate change. These were run under the condition that the bed of the estuary was fixed, although the tidal prism could change with estuary management and sea level rise. Four additional simulations (numbers 25 to 28) were also run, under which morphological changes in the estuary were allowed.
The wave climate scenarios were based on recent Environment Agency guidance (https://www.gov.uk/guidance/flood-risk-assessments-climate-change-allowances) and comprised:
- Wave direction: rotation of the wave climate by ten degrees, both clockwise and anticlockwise, from 1990 onwards; and
- Wave height: growth in wave height by 5% in 1990 and 10% in 2056.
The following three sea level rise scenarios were adopted:
- Low: representing continuation of the historic rate of sea level rise;
- Medium: following the Environment Agency guidance for southeast England and
- High: equivalent to the H++ scenario provided by the Environment Agency in Adapting to Climate Change: Advice for Flood and Coastal Erosion Risk Management Authorities
Alternative coast protection policies were represented through the construction/ removal of seawall structures along both the Felixstowe and Deben SCAPE+ models. A 'Natural Coast' scenario was intended to allow exploration of the dynamics of the system in the absence of structures. Under the 'Defended Coast' scenario structures were represented along the whole Felixstowe frontage (from 1950), at Bawdsey Manor (from 1950) and at East Lane (from 2000). Under both the 'Partial Removal' and 'Removal' scenarios the structures at Bawdsey Manor and East Lane were assumed to be removed in 2050. It may be noted that the last two scenarios were identical at the coast, but this was not the case within the estuary.
Probabilistic methods and results
The input files used to define these scenarios and the resulting output files can be found here.
Evaluation for end-users
As part of the Environment Agency project "Embedding iCOASST into practice", HR Wallingford have undertaken an independent evaluation of the usability of the iCOASST models. Each of the models have been downloaded from this website, compiled and run using (i) the documentation and (ii) the site-specific data sets on which they have been developed (also provided on this website). The outcome of this evaluation for this model can be found here and should be referenced by anyone interested in using or developing the model further.
Bradley, S.L., Milne, G.A., Shennan, I. and Edwards, R., 2011. An improved glacial isostatic adjustment model for the British Isles. Journal of Quaternary Science, 26(5), pp.541-552.
Burningham, H and French, J, 2015. Large-scale spatial variability in the contemporary coastal sand and gravel resource, Suffolk, eastern UK. In Proceedings Coastal Sediments.
van Rijn, L.C., 2014. A simple general expression for longshore transport of sand, gravel and shingle. Coastal Engineering,90, pp.23-39.
Walkden, M.J.A. and Hall, J.W., 2005. A predictive mesoscale model of the erosion and profile development of soft rock shores. Coastal Engineering,52(6), pp.535-563.
Walkden, M.J. and Hall, J.W., 2011. A mesoscale predictive model of the evolution and management of a soft-rock coast. Journal of Coastal Research, 27(3), pp.529-543.