Error quantification of a high-resolution coupled hydrodynamic-ecosystem coastal-ocean model: Part 2. Chlorophyll-a, nutrients and SPM

Marine systems models are becoming increasingly complex and sophisticated, but far too little attention has been paid to model errors and the extent to which model outputs actually relate to ecosystem processes. Here we describe the application of summary error statistics to a complex 3D model (POLCOMS-ERSEM) run for the period 1988–1989 in the southern North Sea utilising information from the North Sea Project, which collected a wealth of observational data. We demonstrate that to understand model data misfit and the mechanisms creating errors, we need to use a hierarchy of techniques, including simple correlations, model bias, model efficiency, binary discriminator analysis and the distribution of model errors to assess model errors spatially and temporally. We also demonstrate that a linear cost function is an inappropriate measure of misfit. This analysis indicates that the model has some skill for all variables analysed. A summary plot of model performance indicates that model performance deteriorates as we move through the ecosystem from the physics, to the nutrients and plankton.     

A Mean Dynamic Topography of the Mediterranean Sea computed from altimetric data, in-situ measurements and a general circulation model

In the Mediterranean Sea, where the mean circulation is largely unknown and characterized by smaller scales and less intensity than in the open ocean, the interpretation of altimetric Sea Level Anomalies (SLA) is rather difficult. In the context of operational systems such as MFS (Mediterranean Forecasting System) or MERCATOR, that assimilate the altimetric information, the estimation of a realistic Mean Dynamic Topography (MDT) consistent with altimetric SLA to be used to reconstruct absolute sea level is a crucial issue. A method is developed here to estimate the required MDT combining oceanic observations as altimetric and in-situ measurements and outputs from an ocean general circulation model (OGCM).In a first step, the average over the 1993–1999 period of dynamic topography outputs from MFS OGCM provides a first guess for the computation of the MDT. Then, in a second step, drifting buoy velocities and altimetric data are combined using a synthetic method to obtain local estimates of the mean geostrophic circulation which are then used to improve the first guess through an inverse technique and map the MDT field (hereafter the Synthetic Mean Dynamic Topography or SMDT) on a 1/8° resolution grid.Many interesting current patterns and cyclonic/anticyclonic structures are visible on the SMDT obtained. The main Mediterranean coastal currents are well marked (as the Algerian Current or the Liguro–Provenço–Catalan Current). East of the Sicily channel, the Atlantic Ionian Stream divides into several main branches crossing the Ionian Sea at various latitudes before joining at 19°E into a unique Mid-Mediterranean Jet. Also, strong signatures of the main Mediterranean eddies are obtained (as for instance the Alboran gyre, the Pelops, Ierapetra, Mersa-Matruh or Shikmona anticyclones and the Cretan, Rhodes or West Cyprius cyclones). Independent in-situ measurements from Sea Campaigns NORBAL in the North Balearic Sea and the North Tyrrhenian Sea and SYMPLEX in the Sicily channel are used to validate locally the SMDT: deduced absolute altimetric dynamic topography compares well with in-situ observations. Finally, the SMDT is used to compute absolute altimetric maps in the Alboran Sea and the Algerian Current. The use of absolute altimetric signal allows to accurately follow the formation and propagation of cyclonic and anticyclonic eddies in both areas.     

Assessment of sea surface temperature observational networks in the Baltic Sea and North Sea

The satellite and in situ Sea Surface Temperature (SST) observational networks in the Baltic Sea and North Sea are evaluated based on the quality of the gridded SST products generated from the networks. A multi-indicator approach is applied in the assessment. It includes evaluation of data quality, effective data coverage, field reconstruction error and model nowcast error. The results show that the best available full-coverage SST product is generated by assimilating the SST observations to obtain a yearly mean model bias of 0.07 °C and RMSE of 0.64 °C. The effective data coverage rate is 31% by using AVHRR (Advanced Very High Resolution Radiometer) data from NOAA (National Ocean and Atmosphere Administration) satellites 12, 14 and 16. The data redundancy increases rapidly with the number of infrared sensors. Using either NOAA satellite 12 or all 3 satellites makes a small difference with regard to derived effective coverage and the ocean model nowcast error. The influence of using the in situ SST observations in the SST field reconstruction is negligibly small. Instead, the major role of in situ SST observations is in calibrating the satellite observations. To study the relative importance of data quality and data coverage, an assessment is done for two satellite products: one product is based entirely on NOAA 12 data and has larger coverage but lower quality. The other product is a subset of the SAF products (derived from NOAA 14 and 16) and has lower coverage but higher quality. Based on current monitoring, modelling and assimilation technology, the results suggest that the data quality is an important factor in further improving the quality of the gridded SST products. Recommendations are made for possible further improvements of the existing SST observational networks.     

Optimal interpolation of sea surface temperature for the North Sea and Baltic Sea

Sea surface temperature fields of the North Sea and Baltic Sea have been constructed for the year 2001 using a multiplatform Optimal Interpolation scheme. The analyzed fields are constructed every 12 h on a 10 km spatial grid. The product is based upon observations from the three NOAA satellites 12, 14 and 16 together with a large amount of in situ observations. Space dependent covariance functions are estimated from the satellite observations and account for spatial and temporal lags. Several independent methods have been used to assess the error on the sea surface temperature product. Compared against independent in situ observations, the mean RMS difference for the year 2001 is 0.78 °C. The spatial distribution of the errors reveals that the Baltic Sea in general show higher errors than the North Sea. The error statistics throughout the year show a temporal variation of the errors with maximum during summer and winter. Tests with a varying number of satellite observations show that the accuracy of the satellite observations is the most important parameter in terms of reducing the errors on the interpolated sea surface temperature product.     

A numerical transport model for predicting the distributions of Cd, Cu, Ni, Pb and Zn in the southern North Sea: the sensitivity of model results to the uncertainties in the magnitudes of metal inputs

A new transport model for metals (named NOSTRADAMUS) has been developed to predict concentrations and distributions of Cd, Cu, Ni, Pb and Zn in the southern North Sea. NOSTRADAMUS is comprised of components for water, inorganic and organic suspended particulate matter transport; a primary production module contributes to the latter component. Metal exchange between dissolved (water) and total suspended particulate matter (inorganic + organic) phases is driven by distribution coefficients. Transport is based on an existent 2-D vertically integrated model, incorporating a 35 × 35 km grid. NOSTRADAMUS is largely driven by data obtained during the Natural Environment Research Council North Sea Project (NERC NSP). The sensitivity of model predictions to uncertainties in the magnitudes of metal inputs has been tested. Results are reported for a winter period (January 1989) when plankton production was low. Simulated ranges in concentrations in regions influenced by the largest inflows, i.e. the NE English coast and the Southern Bight, are similar to the ranges in the errors of the concentrations estimated at the northern and southern open sea boundaries of the model. Inclusion of uncertainties with respect to atmospheric (up to ± 54%) and riverine (± 30%) inputs makes little difference to the calculated concentrations of both dissolved and particulate fractions within the southern North Sea. When all the errors associated with the inputs are included there is good agreement between computed and observed concentrations, and that for dissolved and particulate Cd, Cu and Zn, and dissolved Ni and Pb, many of the observations fall within, or are close to, the range of values generated by the model. For particulate Pb, model simulations predict concentrations of the right order, but do not reproduce the large scatter in actual concentrations, with simulated concentrations showing a bias towards lower values compared to those observed. A factor which could have contributed to observed concentrations, and which is not included in the model, is considered to be a substantial benthic input of dissolved lead during this winter period, coupled to a rapid and extensive scavenging of the dissolved lead to particles. Significant reductions in riverine and aeolian inputs of total Cd and Cu of 70% and 50%, respectively, consistent with aims of North Sea Conferences, are predicted to lead to minor decreases (~ 10%) in water column concentrations of dissolved and particulate Cd and Cu, except near river sources, where maximum reductions of ~ 30–40% may occur.     

Business-process management in high-turbulence environments: the case of the offshore service vessel industry

This article focuses on business-process management in turbulent business environments, more specifically, the management of offshore service vessel (OSV) companies in environments of high volatility and complexity. Such logistics service providers must deal with various heterogeneous elements, and may face extreme situations with uncertain outcomes. This can have implications for the business processes and the business model implemented within the firm. The study reported here employed an explorative qualitative approach with in-depth case studies of firms in the OSV industry in the North Sea and the High Arctic. The results indicate that, in order to tackle the logistics challenges of turbulent environments, a firm may have to introduce entrepreneurial business models with strong links to external actors. To meet the needs of various stakeholders, three business processes are crucial for handling increased turbulence: (1) resource re-configuration adding basic resources as well as dynamic capabilities; (2) knowledge base and capacity, increasing cooperative relations with other firms and (3) added stakeholder- and knowledge-providing business processes. Results from in-depth studies in two different environmental contexts within offshore petroleum operations underscore the importance of adapting business-process management to task-environment complexity and uncertainty.     

Forecast verification of a 3D model of the Mediterranean Sea. The use of discrete wavelet transforms and EOFs in the skill assessment of spatial forecasts

The quality assessment of a nested model system of the Mediterranean Sea is realised. The model has two zooms in the Provençal Basin and in the Ligurian Sea, realised with a two-way nesting approach. The experiment lasts for nine weeks, and at each week sea surface temperature (SST) and sea level anomaly are assimilated. The quality assessment of the surface temperature is done in a spatio-temporal approach, to take into account the high complexity of the SST distribution. We focus on the multi-scale nature of oceanic processes using two powerful tools for spatio-temporal analysis, wavelets and Empirical Orthogonal Functions (EOFs). We apply two-dimensional wavelets to decompose the high-resolution model and observed SST into different spatial scales. The Ligurian Sea model results are compared to observations at each of those spatial scales, with special attention on how the assimilation affects the model behaviour. We also use EOFs to assess the similarities between the Mediterranean Sea model and the observed SST. The results show that the assimilation mainly affects the model large-scale features, whereas the small scales show little or no improvement and sometimes, even a decrease in their skill. The multiresolution analysis reveals the connection between large- and small-scale errors, and how the choice of the maximum correlation length of the assimilation scheme affects the distribution of the model error among the different spatial scales.     

Description of a flexible and extendable physical–biogeochemical model system for the water column

A modelling system for coupled physical–biogeochemical simulations in the water column is presented here. The physical model component allows for a number of different statistical turbulence closure schemes, ranging from simple algebraic closures to two-equation turbulence models with algebraic second-moment closures. The biogeochemical module consists of models which are based on a number of state variables represented by their ensemble averaged concentrations. Specific biogeochemical models may range from simple NPZ (nutrient–phytoplankton–zooplankton) to complex ecosystem models. Recently developed modified Patankar solvers for ordinary differential equations allow for stable discretisations of the production and destruction terms guaranteeing conservative and non-negative solutions. The increased stability of these new solvers over explicit solvers is demonstrated for a plankton spring bloom simulation. The model system is applied to marine ecosystem dynamics the Northern North Sea and the Central Gotland Sea. Two different biogeochemical models are applied, a conservative nitrogen-based model to the North Sea, and a more complex model including an oxygen equation to the Baltic Sea, allowing for the reproduction of chemical processes under anoxic conditions. For both applications, earlier model results obtained with slightly different model setups could be basically reproduced. It became however clear that the choice for ecosystem model parameters such as maximum phytoplankton growth rates does strongly depend on the physical model parameters (such as turbulence closure models or external forcing).     

Towards a holistic digital twin solution for real-time monitoring of aquaculture net cage systems

Digital twins and relevant concepts are being applied in a wide variety of ways, and they are of most use when an actual real-world physical system or process (a physical twin) is changing over time and when measurement data correlated with this change can be captured. In this work, a digital twin model was implemented for real-time monitoring of aquaculture net cage systems, which is notoriously challenging because of several difficult-to-measure properties, such as forces on and deformation of the flexible netting structures, waves and flow field alterations around the cage and complex stiffness behaviour of the mooring elements made by fibre ropes. These properties were set to be adaptable according to the resultant outputs, such as cage responses and mooring loads that were continuously compared with the measurement data obtained from remote monitoring sensors. In this way, real-time sensor data were assimilated into the numerical simulation model for representing the actual net cage system. No dedicated sensors were used for fish monitoring, but the fish behavioural responses to current, wave and cage deformation were modelled according to relevant field observational data. A wireless sensor network has also been tested for the digital twin implementation, which was found to be suitable for practical uses in fish farms.     

A two-layer model of the Rhine plume

The outflow of Rhine water into the shallow Southern Bight of the North Sea leads to almost discontinuous vertical density distributions and sharp frontal structures around the river mouth. Strong tidal motion, wind and baroclinic effects have large influence on the dynamics and dispersion of river water. A three-dimensional tidal model, including advective and diffusive transport of salinity, is used in the two-layer mode for simulation of Rhine water outflow to quantify the interaction of the different processes and the effect on dispersion and mixing of river water. Layer depths are adjusted in a way that no advective transports between upper and lower layer take place in case of sufficiently stable stratification. In case of weak or no stratification the upper layer depth is fixed, and advective transports between layers are computed. Model results show frontal eddy development and (limited) growing internal waves due to baroclinic instability. Comparisons with observational data are presented.     

Mooring cable tracking using active vision for a biomimetic autonomous underwater vehicle

This study presents a localization system using visual information for guidance and navigation of a biomimetic autonomous underwater vehicle (BAUV). The BAUV tracks a mooring cable using two cameras and sonar. Sonar has good accuracy in detecting longitudinal distances but is poor in detecting lateral distances. Since a stereo image has quantization errors, for the cameras, measurement errors in lateral directions are less than those in the optic-axis direction. An extended Kalman filter was employed to combine observational information derived from the cameras and sonar of the mooring cable position with the navigation data of the BAUV. This work demonstrates, using water tank experiments, the effectiveness of the proposed tracking technique in decreasing uncertainty in position estimations of the BAUV and mooring cable.     

An empirical model of changing global competition in the shipbuilding industry

We developed an empirical model to explain the changing global competition among major countries involved in the shipbuilding industry over the last three decades. In order to develop future strategies for not only shipbuilding companies but shipping companies and governments, it is necessary to understand the past logic of their competitive strategies. Using the real strategic behaviours and decision rules of shipbuilding companies derived through a literature survey and interviews, we developed a model using the Cournot oligopoly with Choquet Expected Utility (CEU) theory. Three national players, Japan, Korea and China, are supposed to be either optimistic or pessimistic, which allows us to incorporate the different attitude of the players into the demand uncertainty for our model. Competitive reaction functions are derived and compared with the estimation results via multiple regression with annual shipbuilding tonnage data. Our results provide strong evidence that the major players are bound to use simple rules for competitive quantity decisions, but change the rules completely when they become ineffective. In addition, the competitive behaviour of the players depends on both the degree of demand uncertainty and the different attitudes of the players towards uncertainty.     

Management of coastal eutrophication: Integration of field data, ecosystem-scale simulations and screening models

A hybrid approach for eutrophication assessment in estuarine and coastal ecosystems is presented. The ASSETS screening model (http://www.eutro.org) classifies eutrophication status into five classes: High (better), Good, Moderate, Poor and Bad (worse). This model was applied to a dataset from a shallow coastal barrier island system in southwest Europe (Ria Formosa), with a resulting score of Good. A detailed dynamic model was developed for this ecosystem, and the outputs were used to drive the screening model. Four scenarios were run on the research model: pristine, standard (simulates present loading), half and double the current nutrient loading. The Ria Formosa has a short water residence time and eutrophication symptoms are not apparent in the water column. However, benthic symptoms are expressed as excessive macroalgal growth and strong dissolved oxygen fluctuations in the tide pools. The standard simulation results showed an ASSETS grade identical to the field data application. The application of the screening model to the other scenario outputs showed the responsiveness of ASSETS to changes in pressure, state and response, scoring a grade of High under pristine conditions, Good for half the standard scenario and Moderate for double the present loadings. The use of this hybrid approach allows managers to test the outcome of measures against a set of well-defined metrics for the evaluation of state. It additionally provides a way of testing and improving the pressure component of ASSETS. Sensitivity analysis revealed that sub-sampling the output of the research model at a monthly scale, typical for the acquisition of field data, may significantly affect the outcome of the screening model, by overlooking extreme events such as occasional night-time anoxia in tide pools.     

Assimilation of sea surface temperature in the POL Coastal Ocean Modelling System

A one-dimensional scheme is used to assimilate satellite Sea Surface Temperature data into the Proudman Oceanographic Laboratory Coastal Ocean Modelling System, set up in the Irish Sea with a fine resolution ( 1.8 km). The capabilities of the assimilation scheme are investigated using two different sets of satellite data, of lower and similar resolution to that of the model respectively. Comparison of results with independent data show that assimilation improves the modelled Sea Surface Temperature, but does not address model representation of the temperature vertical structure. It is concluded that for the Irish Sea and at the scales resolved by the model, the assimilation problem cannot be approached in a one-dimensional framework. It is also pointed out that forecast error needs to account explicitly for errors in the representation of the vertical structure of the thermal field.Three-dimensional methods that are suited for coastal systems are then suggested.     

Importance of ocean circulation in ecological modeling: An example from the North Sea

There is an increasing number of ecological models for the North Sea around. Skogen and Moll (2000) [Skogen, M.D., Moll, A. 2000. Interannual variability of the North Sea primary production: comparison from two model studies. Continental Shelf Research 20 (2), 129–151] compared the interannual variability of the North Sea primary production using two state-of-the-art ecological models, NORWECOM and ECOHAM1. Their conclusion was that the two models agreed on an annual mean primary production, its variability and the timing and size of the peak production. On the other hand, there was a low (even negative dependent of area) correlation in the production in different years between the two models.In the present work, these conclusions are brought further. To try to better understand the observed differences between the two models, the two ecological models are run in an identical physical setting. With such a set-up also the interannual variability between the two models is in agreement, and it is concluded that the single most important factor for a reliable modeling of phytoplankton and nutrient distributions and transports within the North Sea is a proper physical model.     

Environmental factors which affect growth of Japanese common squid, Todarodes pacificus, analyzed by a bioenergetics model coupled with a lower trophic ecosystem model

Bioenergetics model is applied to Japanese common squid, Todarodes pacificus. The temporal change of wet weight of common squid, which migrates in the Sea of Japan, is simulated. The time dependent horizontal distribution of prey is calculated a priori by 3-D coupled physical–biological model. The biological model NEMURO (North Pacific Ecosystem Model for Understanding Regional Oceanography) is used to simulate the lower-trophic ecosystem including three kinds of zooplankton biomass two of which is used as prey of common squid. A bioenergetics model reproduced appropriate growth curve of common squid, migrating in the North Pacific and the Sea of Japan. The results show that the wet weight of common squid in the northern Sea of Japan is heavier than that migrating in the central Sea of Japan, because prey density of the northern Sea of Japan is higher than that of the central Sea of Japan. We also investigate the wet weight anomaly for a global warming scenario. In this case, wet weight of common squid decreases because water temperature exceeds the optimum temperature for common squid. This result indicates that migration route and spawning area of common squid might change with global warming.     

Variability of mesozooplankton spatial distribution in the North Aegean Sea, as influenced by the Black Sea waters outflow

The North Aegean Sea constitutes an important region of the Mediterranean Sea since in its eastern part the mesotrophic, low salinity and relatively cold water from the Black Sea (outflowing from the Dardanelles strait) meets the oligotrophic, warm and very saline water of Levantine origin, thus forming a thermohaline front. Mesozooplankton samples were collected at discrete layers according to the hydrology of the upper 100 m, during May 1997 and September 1998. In May highest biomass and abundance values (up to 66.82 mg m^− 3 and 14,157 ind m^− 3) were detected in the 10–20 m layer (within the halocline) of the stations positioned close to the Dardanelles strait. The front moved slightly southwards in September, characterized by high biomass and abundance values within the halocline layer. The areas moderately or non influenced by Black Sea water revealed lower standing stock values than the frontal area in both cruises and maxima were detected in the uppermost low salinity layer. Samples collected at the stations and/or layers more influenced by Black Sea water were distinguished from those collected at layers and/or stations more affected by Levantine waters in both periods. In May the former samples were characterized by the copepods Acartia clausi, Centropages typicus, Paracalanus parvus. The abundance of the above species decreased gradually with increasing salinity, in the horizontal and/or in the vertical dimension, with a parallel increase of the copepods Oithona plumifera, Oithona copepodites, Oncaea media, Ctenocalanus vanus, Farranula rostrata. During September the frontal area as well as that covered by the modified Black Sea water, were highly dominated by the cladoceran Penilia avirostris and doliolids. For both seasons, MDS plots, issued from the combination of mesozooplankton and water-type data, revealed the gradual differentiation of zooplankton composition from the frontal area towards the area covered by Levantine water, following the spreading and mixing of the Black sea water. The observed temporal and spatial variability in the distribution pattern of mesozooplankton standing stock and species composition seems to depend considerably on the variability of circulation and frontal flows.     

Using OMP analysis to observe temporal variability in water mass distribution

Previous optimum multiparameter analyses have investigated spatial distribution of water masses and reported successful results. This study tests the method further with application to a time series to examine temporal variability in water mass distribution. Observations collected from the Sargasso Sea are utilised, mainly to detect well-documented property changes that have occurred in Labrador Sea Water, but also possible fluctuations in water mass contributions as a whole.The results show much structure and variability in the contributions of Central Water and upper deep waters in the depth range 900–2000 m. It is proposed that the passage of warm and cold core eddies is indicated by these changes in relative contributions of water masses, although quantitative comparison of satellite altimetry data with the model results is not conclusive.Significant irregularities in the distribution of Labrador Sea Water are also revealed by the model. Through-flow of modified (since the early 1990s) Labrador Sea Water is denoted in the results by an apparent absence from mid-1995 to early 1998. This suggests a transit time of 5–6 years between the Labrador Sea and Bermuda, which is consistent with observational findings described in the literature.     

Interactions of wind and density driven currents in North Sea ROFIs—a model study

Three different versions of a baroclinic three-dimensional circulation model of the North Sea are used to obtain information on the wind and density interactions in the North Sea ROFIs (Regions Of Freshwater Influence): the standard version with fully prognostic treatment of salinity and temperature is compared to a barotropic model run on the same grid on the one hand and to an also fully prognostic model run on a four times coarser grid on the other hand. In order to gain knowledge on the wind and density interactions, two opposing wind directions are chosen for investigation, namely a time of strong north wind, 21st–28th April 1982, and a time of strong southwest wind, 22nd–24th May 1982. In the April case the effect of the salinity gradients on the border of the ROFIs of Rhine, Weser, Ems and Elbe, i.e. along the continental shore, is shown to lead to a clear enhancement of the mean surface currents. In May this result is partly disguised by the additional effect of the thermocline in the deeper parts of the North Sea, i.e. in the classical shelf sea regime region. Nevertheless, the same pattern of enhanced mean surface currents along the coast is detected and is of the same order of magnitude as in the April case. It is thus concluded that although the circulation in the North Sea is reversed by the wind, the density induced component of the general circulation is modified only slightly.