OF EUROPE-PROMISE
D. Prandle
Proudman Oceanographic Laboratory
Bidston Observatory,
Birkenhead, L43 7RA, UK
SUMMARY
Environmental and Financial imperatives urgently require the effective application of
oceanographic knowledge for appropriate development of the coastal zone. This knowledge
base must be synthesised within pre-operational numerical models linked to related
observational/monitoring networks constructed on an international basis. This
international (European) approach involves long term planning, to: establish and
maintain the monitoring network; train personnel in the necessary organisational and
scientific skills, and link the operational centre(s) for real-time data assimilation.
This new challenge requires rigorous evaluation of both existing models and the
associated data used for: initialising, forcing and verification.
PROMISE aims to expedite the implementation of operational oceanography and comprises European groups with wide experience in real-time modelling and monitoring. Applications are primarily in the North Sea and concentrate on the simulation of suspended sediment in the near-shore zone. This applications aims to exploit recent rapid developments in both in-situ and remote sensing techniques for measuring SPM.
The project aims are:
- to assemble high resolution comprehensive data sets
- to use these data sets to test pre-operational models
- to improve these model systems by
a. expanding their physical realism,b. improving quality and quantity of observational data andc. by further development of real-time data assimilation techniques.
The primary objective is to identify how modelling and monitoring capabilities need to be developed and combined to provide estimates of coastal - nearshore sediment exchanges. The traditional focus of coastal engineering on the response of beaches to waves has been extended to consider coastal-nearshore sediment exchange, reflecting concerns arising from acceleration in: climate change (sea level and storminess), offshore aggregate extraction, steepening of cross-shore slopes, adoption of soft-defence strategies (including beach nourishment) and recognition of large-scale interdependencies.
Such large-scale sediment exchanges cannot be measured directly - not even the resultant impact on bathymetry. Thus models are required, linking dynamical simulations of tide, surge, wave and turbulence to 'dispersion' modules representing the erosion, transport and deposition of a range of sediment types. Monitoring is required to develop the sediment algorithms, determine the accuracy of each modelling component and assess the integrated forecast of sediment transport.
The requirement for dynamical coupling of model components and subsequent evaluation against observations emphasises the need for a pre-operational modelling approach. The North Sea was adopted as a focus because of the existence of forecasting systems for: tides, surges and waves. The wave models required development for fine-scale application in near-shore shallow regions. Likewise the turbulence models required development to incorporate wave-tide-surge interactions. Extensive data sets were then required to develop and assess these various modules independently and overall. Data from national monitoring experiments off the coasts of the UK, Germany, France and Spain have been made available for this purpose.
Subsequent sections describe: the scope of these observational data sets (section 2), development in spectral wave models for propagation in shallow water (section 3), single point models of turbulence and suspended particulate matter (section 4). Section 5 considers additional efforts required to develop operational oceanographic capability in N.W. Europe, and section 6 examines the associated user requirements.
2. OBSERVATIONAL DATA SETS
Figure 1. Regional foci within North Sea simulations. |
The ZISCH and North Sea Project data, used extensively in PROMISE, illustrate the longer term value of such data sets. The similarly comprehensive coastal data sets from Holderness and Sylt Rømø are described below. PROMISE also compiled important coastal data sets from the French and Spanish coasts. Figure 1 indicates these regional foci.
a. Holderness
The Holderness Coastal Experiment (Prandle 1994) originated as a component of the UK
LOIS programme, it aimed at understanding and ultimately predicting coastal erosion.
The Holderness coast was chosen because of its rapid rate of erosion (20 m glacial till
cliffs eroding at an average rate of 1.7 m/year) and its reasonable homogeneity over a
20 km section. The requirement was for continuous monitoring of representative
conditions over a winter period providing data both for developing and verifying
numerical models and background descriptions for occasional more intensive localised
process studies. The core period chosen was from October '94 to March '95, pilot
studies were made in November-December '93 with a follow-up phase between October '95
to January '96. The range of sea-borne, in-situ and remote sensing data collected at
Holderness is indicated by the schematic shown in figure 2.
Figure 2. Schematic of regional observation exercise (Holderness). |
b. Sylt Rømø Bight (North German Bight)
The Sylt-Rømø Bight, located at the border between Denmark and Germany,
is a semi-closed Wadden Sea lagoon of approximately 1000 km2. Exchange of
water and material proceeds through one deep tidal gully. Tidal elevation and currents
are controlled by the water level of the German Bight. The development of waves depends
on a complex interaction of local winds, tidal elevation, tidal current directions and
the local topography.
In the last years, the Sylt-Rømø Bight was subject to intense interdisciplinary research on the exchange of material and energy between the different compartments of the ecosystem. Considering the sediment budget a substantial sediment loss over the last hundred years with a 50 percent loss of the intertidal areas was identified. In one hypothesis these effects were attributed to the increase of intense storm events with parallel reduction of retention areas due to coastal protection measures.
Over several months periods during 1996 and 1997 the sediment dynamics during stormy and calm weather periods were continuously recorded at selected locations of the main gullies and one tidal flat position. These measurements which were organised and financed by the GKSS Research Centre Geesthacht, were completed by two two-week ship cruises to gather profiles along the tidal gullies. Observations of the microturbulent structures over a tidal cycles (the PROMICSOS campaign) enhance the measured time series. The requirement was a consistent data set comprising meteorological, hydrodynamic (including waves) and suspended sediment parameters to develop and test a numerical coupled sediment transport model and its individual atmospheric, wave, current and suspended sediment dynamics components. Four three week periods containing significant wind events were selected as benchmark tests for flat water models. These data are carefully documented and accessible via the World Wide Web.
In the framework of this extensive measuring campaign a joint PROMISE - MICSOS (PROMIX) experiment was conducted. This experiment was aimed at deriving a suitable data set of both the turbulent kinetic energy (TKE) and its dissipation rate. The PROMIX experiment covered two tidal cycles and consisted of highly resolving ADCP measurements and microstructure measurements of the dissipation rate of TKE. Data of both the TKE and its dissipation rate are needed to validate the 1D turbulence-SPM-wave interaction model. Jointly cooperating two PROMISE partners (IfM and GKSS) were directly involved in the PROMIX experiment, in which the IfM was responsible for the experiment. The microstructure measurements were carried out by Meerestechnik-Elektronik GmbH (Trappenkamp, Germany).
3. WAVE MODELLING
The wave group within PROMISE has coalesced to build on earlier collaboration in WASA,
ECAWOM, SCAWVEX and the WISE sub-group of WAM. The main goal of the first year
activities for wave modelling was to achieve a common implementation of the
third-generation spectral wave model WAM-Cycle 4 - suitable for dissemination. The
goals of the second and third year are to extend this 'deep water' version for high
spatial resolution application in shallow water and dynamic coupling to a hydrodynamic
model to satisfy the requirements for sediment transport modelling.
The model intercomparisons on the North Sea scale (POL Report No. 47) have clearly
established the (deep water) characteristics to which model applications are sensitive.
Extensive tests were conducted on the intercomparison of the different implementations;
on the comparison against buoy measurements (figure 3) and with ERS-1 and
TOPEX/POSEIDON satellite altimeter data. WAM Cycle 4 has been successfully applied to
hindcast wind waves in the North Sea. Statistical analysis of buoy or satellite data
and model results indicated that the current North Sea implementation is of comparable
quality as other implementations cited in literature. The intercomparison exercise did
not reveal any significant differences in model results between the POL and the
KULeuven/MUMM implementations, although boundary conditions and grid resolution were
quite different. Small differences in bathymetry are also not very important. The wind
fields provided by meteorological offices (UKMO / DNMI) seemed to be of comparable and
good quality.
Figure 3. Intercomparison of wave models. |
The last phase consisting of the dynamical linking of tide/surge models and wave models has started. A flexible and robust framework, limiting the changes needed to individual model components while keeping the overhead down, is being built and is undergoing testing.
Within the technical framework of WAM cy-4 a wave model using turbulent diffusion as a key dissipation mechanism was developed in the last years (Günther and Rosenthal, 1992). The model was designed for to nonuniform and instationary systems (Schneggenburger et al. 1997). The model forced by the local measured winds and computed water level and current fields every 15 minutes was successfully used to hindcast a storm period in the Sylt-Rømø-Bight. The strong influence of the local current, water depth and changes of the effective fetch on the wave development could be demonstrated. The agreement with measurement collected during the Sylt-Rømø-Campaign was remarkable (Schneggenburger at al. 1998).
4. SINGLE POINT MODELS OF TURBULENCE AND SUSPENDED PARTICULATE MATTER (SPM)
The parallel EC initiative (EU MAST Workshop, Bergen 8-10.08/96) focusing on
rationalisation of turbulence modelling contributed significantly to the associated
goals within PROMISE. The report (POL Report No. 48) outlines progress in YEAR 1 in
relation to tidal applications. Extension to include the effects of surface waves was
completed in year 2. Meanwhile applications of these models both directly to the
regional data sets and incorporation into wider-area models is progressing as planned.
One-dimensional generic turbulence-SPM-wave interaction models were derived in order to investigate the role of turbulence in relation to SPM and density stratification on a microscale in space and time. The results are planned to improve existing multi-dimensional SPM-models on a pre-operational level.
Two versions of a one-dimensional turbulence-SPM-wave interaction model were established: Version (A) considers cohesive SPM (i.e. flocs), version (B) focuses on non-cohesive material (i.e. sand). Starting point was the k-e model as described by Baumert & Radach (1994) and Burchard & Baumert (1992).
To derive the generic turbulence-SPM-wave interaction model (version A), the k-e model has been considerably improved and extended by the inclusion of a transport module for SPM, which includes sinking and diffusion processes. At present, only a single floc size is considered. However, further development will lead to a minimum of three size classes. As version A includes only cohesive materials, the processes of erosion and sedimentation are not implemented i.e. there is no exchange with the bed in version A of the model. In order to be able to resolve the fluffy layer close to the bottom, a non-uniform spatial grid was introduced.
Version B of the model considers non-cohesive SPM and hence, in addition to the
advection and diffusion processes, also includes resuspension and sedimentation
processes, after Sheng & Villaret (1989). It is also necessary to allow the grid to
adapt to the time dependent variation in water depth (moveable bed), thus also changing
the model's domain of integration.
Figure 4. Model simulation of turbulent dissipation rate. |
Implementing the wind wave boundary condition we followed the work of Grant and Madsen (1979) in the form applied by Davies and Lawrence (1995). This approach relates the apparent bottom roughness felt by the current to the combined action of wind waves and current. Depending on the water depth the combined action of wind waves and current lead to an increased bottom friction and subsequently influences the bottom processes regarding the SPM dynamics.
A first test of the model was conducted for the site of the FLEX experiment (1976), from which the required meteorological forcing data and measured temperature profiles for validation of the model results were available. First results showed that the characteristic features of the vertical temperature gradient are well reproduced. Furthermore, turbulence microstructure parameters showed reasonable values.
Further model runs against suitable PROMISE data sets will test the model and related processes in more detail (figure 4, Simpson et al. 1996).
5. OPERATIONAL MODELLING
A clear conclusion regarding the observational data sets is the inadequacy (accuracy
and resolution) of the bathymetric data and the wind data. Moreover questions of
copyright/proprietary rights to both these frequently arise.
The development of generic modules and the ready availability of general purpose model codes have removed much of the mystique that traditionally surrounded marine modelling centres. Present day concerns concentrate on how best to optimise the model formulations and associated forcing data for the required applications. Thence "Rationalisation" often implies accessing the most appropriate: algorithms/numerical solutions, bathymetry and related model set-up data, initial, coastal and open-sea (ocean-ocean) boundary conditions, surface forcing, assimilation and verification data. Whilst the long-term solution may be fully integrated atmospheric-ocean-shelf-sea-coastal-estuarine-hydrological operational models, shorter term expedients need to be developed.
The report (POL Report 49) on remote sensing/assimilation recognises a number of salient features of such expedients. The most useful remote sensing techniques should be:
- for sediments: optical data, including AVHRR and Landsat/SPOT. Improved quality should become available with the launching this year of SeaWIFS on SeaSTAR
- for beach processes and coastal erosion, SAR can be useful for monitoring, in addition to the traditional optical sensors
- for winds, scatterometer for global-scale; SAR images for fine-scale (1 km resolution) have poor temporal coverage
- for waves, altimeter for wave height. SAR can be used for swell in exposed areas such as the Bay of Biscay
- for temperature fronts, thermal infrared images (AVHRR/ATSR)
- for current fronts, SAR gives qualitative information, but with poor temporal coverage.
Figure 5. Examples of different model grids. |
There is generally a trade-off between the spatial resolution of satellite data and its available temporal coverage. Continuous, nearly global coverage is afforded by geostationary meteorological satellites, and coverage several times daily is available for NOAA AVHRR. These optical and infra-red data are only available under cloud-free conditions, which is a significant disadvantage for mid-latitude locations which are subject to frequent poor weather. Passive microwave and radar sensors can penetrate cloud cover, but the former sensors generally have rather too poor resolution for covering North Sea scales. The detail available in SAR images is good, but the coverage is only every few days at present, though there is a possibility of improvement in data coverage with the use of the recently launched RADARSAT.
When interpreting remote sensing data it is important to consider that satellites only observe the Earth's surface. In order to achieve three-dimensional marine information and forecasts, remote sensing data together with in-situ data must be assimilated in numerical models. There has, over the last decade, been a rapid development of various data assimilation methods which can be used with ocean hydrodynamic and ecosystem models. At present, however, none of these methods are used operationally, at least partly because the necessary observational data are either insufficiently accurate or do not provide sufficiently good coverage. In a fully operational system the access time for the most recent observations also becomes important. Real time analyses and predictions from the European weather services must be used to ensure a proper forcing of the model and to make it possible to generate realistic predictions of the marine system.
A further aspect of PROMISE addresses optimising the modelled North Sea hydrodynamics, given known or assumed uncertainties in field data and model parameters and input (ten Brummelhuis et al. 1997). Similarly, the sensitivity of modelled SPM behaviour to such uncertainties and assumptions is analysed.
Extensive rationalisation exercises have been undertaken to illustrate the range of variability in model performance as functions of differences in: formulation (figure 5), data prescription and model operation. Likewise the adequacy of related databases (meteorological, geological, bathymetric) has been assessed. The task of quantifying the sensitivity of net estimates of sediment transport to the full range of 'impact' variables is ongoing - the aim being to prioritorise the requirements necessary to extend predictability.
Significant problems are to get the various models to communicate with each other at suitable space and time intervals and to distribute the model results in a way suitable for end users. For example, it is necessary to produce some kind of parameterisation of the results of detailed turbulence models in order for them to be applied in hydrodynamic/sediment transport models at a North Sea scale. It is also necessary to obtain input/forcing data and data for model verification on space and time scales suitable for the models. This applies no less to remote sensing data than to field observations.
This project illustrates the challenge of integrating the expertise of specialists in modelling, monitoring, remote sensing and data assimilation to provide a state-of-the-art assessment of the accuracy and limitations of estimating sediment fluxes. The parallel applications of a suite of models to a range of coastal sites provides broad insights into the underlying problems and the scope for extending predictability.
6. DISSEMINATION AND EXPLOITATION
The challenge of dissemination to the external user community was
addressed by convening a 'User-Group' meeting which included sessions on: existing
operational services, planned future developments and developments in monitoring,
technology and assimilation. This communication of research activities was followed
by accounts of the related requirements of users (including flood defence, port
operators, coastal management, ecologists, oil and gas and naval). The attendance by
60 participants from ten European countries ensured enthusiastic
participation.
Figure 6. Operational oceanography. |
The singular facet of this meeting was the pragmatic focus engendered by the participation of users, model operators, scientists and monitoring experts. Exciting opportunities are presented by the rapid advances in: computational power, monitoring technology and systems, scientific understanding and numerical methods. However, investment and the associated progress will depend on demonstrable benefits to end users. The pace will be dictated by our ability to collaborate in maximising the potential of past investments as well as careful planning for the future. Initiatives are needed to develop structured research, development and evaluation programmes to parallel the GOOS plans for the period 2000-2005. There is a well-defined need for great accuracy, resolution and reliability of existing services. The ultimate goal is a fusion of environmental data and knowledge utilising fully the communications and computational capacities to optimise sustainable development of our marine environment (figure 6).
REFERENCES
Baumert, H. and G. Radach (1992)
Hysteresis of turbulent kinetic energy in
nonrotational tidal flows: a model study. J. Geophys. Res. Vol. 97, 3669-3677.
ten Brummelhuis, P.G.J., Gerritsen, H. and Van der Kaaij, T (1997)
Sensitivity
analysis and calibration of the hydrodynamic PROMISE model using adjoint modelling
techniques. Delft Hydraulics Research Report Z2025/Z2047.
Burchard, H. and H. Baumert (1995)
On the performance of a mixed-layer model
based on the k-e turbulence closure. J. Geophys. Res. Vol. 100, C5, 8523-8540.
Craig, P. and Banner, M. (1994)
Modelling wave-enhanced turbulence in the ocean
surface layer. Journal of Physical Oceanography, 24 (12), 2546-2559.
Grant, W.D. and Madsen, O.S. (1979)
Combined wave and current interaction with
a rough bottom. Journal of Geophysical Research, 84 (C4), 1797-1808.
Davies, A.M. and Lawrence, J. (1995)
Modelling the effect of wave-current
interaction on the three dimensional wind driven circulation of the eastern Irish Sea.
Journal of Physical Oceanography, 25 (1), 29-45.
Jenkins, A.D., G , Evensen., E. Bjørgo., H. Espedal., E, Korsbakken., K. Kloster.,
T. Hamre., V. Jensen., P. Samuel., D. Durand., M. Gómez Lahoz., J. Carlos Carretero.,
C de Valk., K. Mastenbroek., R.J. Vos., H. Gerritsen. and P.G.J ten Brummelhuis.
Remote Sensing Applied to Operational Models - North Sea. POL Report No. 49, 1997.
Luo, W., Flather, R. and Monbaliu, J. (1998)
Modelling effect of mean current
on bottom friction dissipation in a third generation model (WAM), abstract submitted to
ICCE-conference, Copenhagen.
Monbaliu, J., R. Flather, J.C. Carretero and H. Gerritsen:
"Wave modelling in
the Promise Project", Oceanology 98, Brighton, 1998.
Monbaliu, J., M.Y. Zhang., K. De Backer., J. Hargreaves., W. Luo., R. Flather.,
J.C. Carretero., M. Gomez Lahoz., I. Lozano., M Stawarz., H. Günther., H. Rosenthal.
and J. Ozer.
WAM Model Intercomparisons - North Sea. POL Report No. 47, 1997.
Nöhren, I., K, Duwe. and H, Baumert.
PROMISE Literature Review on Turbulence,
Wave and Suspended Matter Transport Modelling. POL Report No. 50, 1997.
Prandle, D.
The Holderness Coastal Experiment '93-'96. POL Report No. 44. 1996.
Regener, M., B. Szilagyi. and H. Baumert.
The Generic 1-D turbulence-SPM
model. POL Report No. 48, 1997.
Sheng, Y.P., Villaret, C. (1989)
Modelling the effect of suspended sediment
stratification on bottom exchange processes. J. Geoph. Res., Vol. 94, No C10,
14.429 - 14.444.
Schneggenburger, C., Günther, H. and Rosenthal, W. (1997)
Shallow water wave
modelling with non-linear dissipation. Dt. Hydrogr. Z., 49 (2/3), in press.
Schneggenburger, C., Günther, H. and Rosenthal, W. (1988)
Shallow water wave
modelling with nonlinear dissipation: Application to small scale tidal systems.
In: Proceedings of the 5th International Workshop on Wave Hindcasting and Forecasting,
January 27-30, 1998, Melbourne, Florida, USA.
Simpson, J.H., Crawford, W.R., Rippeth, T.P., Campbell, A.R. and Josepth, V.S.C.
(1996)
The vertical structure of turbulent dissipation in shelf seas. J. Physical
Oceanography. 26(8), 1579-1590.
Last updated: 23rd August 1998. Please send comments to A.Lane@pol.ac.uk