Complex deterministic models are an important tool for studying a wide range of systems. Often though, such models are computationally too expensive to perform the many runs required. In this case one option is to build a Gaussian process emulator which acts as a surrogate, enabling fast prediction of the model output at specied input congurations. Derivative information may be available, either through the running of an appropriate adjoint model or as a result of some analysis previously performed. An emulator would likely benet from the inclusion of this derivative information. Whether further eciency is achieved, however, depends on the relation between the computational cost of obtaining the derivatives and the value of the derivative information in the emulator. In our examples we see that derivatives are more valuable in models which have shorter correlation lengths and emulators without derivatives generally tend to require twice as many model runs as the emulators with derivatives to produce a similar predictive performance. We conclude that an optimal solution is likely to be a hybrid design consisting of adjoint runs in some parts of the input space and standard model runs in others. The knowledge of the derivatives of complex models can add greatly to their utility, for example in the application of sensitivity analysis or data assimilation. One way of generating such derivatives, as suggested above, is by coding an adjoint model. Despite automatic dierentiation software, this remains a complex task and the adjoint model when written is computationally more demanding. We suggest an alternative method for generating partial derivatives of complex model output, with respect to model inputs. We propose the use of a Gaussian process emulator which, as long as the model is suitable for emulation, can be used to estimate derivatives even without any derivative information known a priori. We present encouraging results which show how an emulator of derivatives could reduce the demand for writing and running adjoint models. This is done with the use of both toy models and the climate model C-GOLDSTEIN.

No description available.

Marine diazotrophs play an important role in marine biogeochemical cycles by fixing N2 into bioavailable forms, thus sustaining oceanic productivity over broad timescales through maintenance of bioavailable nitrogen stores. However, as assessments of diazotrophic organisms are traditionally constrained to the upper ocean, the fate of diazotroph-derived new nitrogen is not clear. Many previous assessments of the fate of diazotrophs has assumed that the majority of new nitrogen produced in these organisms is recycled in the upper ocean through the microbial loop and that diazotroph contribution to export is minimal except following blooms of diazotrophic diatom associations (DDAs). In this study, a combination of light microscopy, transmission electron microscopy, and qPCR of sinking particulate material from the subtropical and tropical Atlantic Ocean and Gulf of Mexico has revealed that filamentous, heterocystous and unicellular cyanobacterial diazotrophs are present below 100 m, and provides some of the first evidence that this appears to be a widespread occurrence. Herein we identify the mechanisms by which diazotrophs are exiting the mixed layer via passive sedimentation, aggregation, and incorporation in faecal material. Diazotrophs also appear to be contributing to the export of particulate organic nitrogen with Trichodesmium composing up to 3% of PON standing stock and 1 – 17.5% of PON flux at 10 m below the mixed layer in the (sub-)tropical Atlantic Ocean. The likelihood that the subsequent remineralisation of diazotroph-derived material at depth is contributing to the N* anomaly observed in the thermocline in the North Atlantic sub-tropical gyre is also discussed. This work provides some of the first descriptions of mechanisms by which diazotrophs contribute to these anomalous nutrient distributions, such as through remineralisation of diazotroph biomass following cellular lysis. These results aid in the elucidation of the extent to which Trichodesmium and other diazotrophs are contributing to the biogeochemistry of deeper waters and provides novel insight into the cycling of fixed nitrogen in the oligotrophic ocean.

The Earth’s oceans host an enormous range of natural resources, over 90 billion kg of fish and shell fish are caught each year (WorldOceanReview, 2011). The oceans play an overriding role in climate regulation, removing vast quantities of carbon from the atmosphere. It is believed phytoplankton could account for more than half the earth’s oxygen production (ConsciousAlliance, 2011). Monitoring these biological and chemical characteristics offers an invaluable insight into the way our oceans work, which can then be used to generate and verify reliable models of the global ecosystem. In-situ sensors have been identified as the way forward, due to the ever changing properties of the world’s oceans in terms of chemistry and biology. The sensors will have to overcome harsh working conditions and process large quantities of data, whilst using as little power as possible. The work undertaken for this thesis aims to develop ultrasonic standing wave particle manipulation techniques for use in an oceanographic environment. Ultrasonic particle manipulation techniques are generally confined to ceramic devices, which are incompatible with oceanographic sensing on a large scale deployment. This work has bridged that gap and developed fresh approaches to ultrasonic techniques in polymer devices. In addition to this, novel manufacturing methods have been developed to improve the robustness of the devices or to make the technology more compatible with cheaper, quicker and easier manufacturing techniques. A specific problem in oceanographic sensing is biofouling – the build up of microorganisms in and around the sensor. This project has investigated the feasibility of using ultrasonic techniques to reduce this build up, in particular, the formation of biofilms within sensors. The use of ultrasonics to reduce biofouling has been investigated by others, but it generally focuses on acoustic streaming techniques which induces mixing and has high power requirements incompatible with remote sensing (Sankaranarayanan et al., 2008). This work was carried out in conjunction with the Centre for Marine Microsystems in Southampton. The centre is developing robust high performance metrology systems for use in oceanographic science. Four distinct class of sensors will all be looked at, Chemical, Physical, Nucleic acid sensors and a μ-flow cytometer (UniversityOfSouthampton, 2011). To assist in the development of oceanographic ultrasonic sensing platforms, one dimensional and two dimensional modelling was carried out in Matlab (R2014b) and ANSYS (12.0). The models were used to design new devices as well verify experimental results. In particular, ANSYS was used to investigate the mechanism behind standing wave particle manipulation devices. The modelling investigated the robustness of such a device and their suitability for scaling up and integration into an oceanographic sensor. Once the computational modelling had been carried out, devices were built using a variety of manufacturing techniques. As required, the techniques were adapted and optimised for the production of ultrasonic oceanographic sensors. The work went on to qualitatively and quantitatively analyse the affects of ultrasonic techniques in an oceanographic sensor. In particular, the formation of biofilm within a polymer sensor was analysed. Image processing software was optimised for the experiments then used to identify the effects of ultrasonic standing wave techniques. The work shows that it is possible to reduce the build up of biofilms within polymer devices over substantial time periods using ultrasonic standing wave techniques. The devices used differ from conventional ceramic devices in that less than an exact half wave was set up across the fluid channel. This means there are non zero forces acting at the polymer/fluid boundary which is beneficial for the reduction of biofilm formation. The work also identifies a mechanism for the alignment of particles within a microfluidic device through the use of surface acoustic waves, though it was not possible to verify the computational results experimentally. A novel manufacturing technique using spin coating was developed that would allow easier construction of surface wave devices. In addition to this, a new type of device was developed utilising the transparent properties of lithium niobate in a bulk acoustic wave configuration. In the process of carrying out this work an experimental method has been developed allowing the depth of particles within devices to be ascertained through long exposure images. The length of the particle streak allows the position of the particle relative to a solid/fluid boundary to be inferred, and the data can be presented in such a way to build up a picture of the device characteristics.

The inversion of compressional wave properties is presently emerging as a technique for determining the geotechnical properties of marine sediments. However, the relationships required to perform such an inversion are still under debate, with further research required to resolve the dependence of compressional wave properties on both frequency and geotechnical properties. Though the use of in situ probes provides the most promising manner of examining these relationships, previous work in this field has encountered a number of experimental difficulties. This work presents a series of well-constrained in situ transmission experiments. These were undertaken on inter-tidal sediments using a purpose built in situ device, the Sediment Probing Acoustic Detection Equipment (SPADE). Compressional wave properties were measured from 16 to 100 kHz in a range of sediment types (medium to fine sands and medium to fine silts), with several closely spaced locations examined at each general site to assess the local variability in compressional wave properties. Spreading losses, which were adjusted for sediment type, were incorporated into the data processing. Also included were a thorough error analysis and an examination of the repeatability of both the acoustic wave emitted by the source and the coupling between the probes and the sediment. The results indicate that sands possess greater group velocities, greater effective attenuation coefficients and lower quality factors than silts, while the low velocities measured in silts imply that the bulk moduli of the silt sites examined are lower than expected owing to a considerable fraction of organic matter. Significant variations were observed in compressional wave properties, which were more reliably related to variations in geotechnical properties in sands than in silts. Group velocities were observed to be independent of frequency in sands within 95 % confidence limits, with no reliable frequency-dependence being determined in silts owing to variability in the measured values. Effective attenuation coefficients were proportional to frequency within 95 % confidence limits for the majority of the sand and silt locations examined. Results indicate that compressional wave properties can be used to determine porosity, bulk density and sand fraction, while the reliable determination of mean grain diameter from compressional wave properties in inhibited by the scatter in the data. The results from this study were also used to assess the effectiveness of Biot Theory to predict the compressional wave properties of these sediment types. In sands, the Biot phase velocities agreed with measured group velocities, while Biot absorption coefficients were less than measured effective attenuation coefficients, owing to scattering or squirt flow not accounted for in the Biot Theory. In silts, Biot phase velocities are greater than measured group velocities, while Biot absorption coefficients generally agree with or are greater than measured effective attenuation coefficients. In silts, predicted velocities are greater than those measured, while absorption coefficients generally agree with or are greater than measured attenuation coefficients. The discrepancy between the measured attenuation coefficients and predicted absorption coefficients can be explained through the over-estimation of in situ porosities by the geotechnical measurement techniques adopted.

Over the past decade within the renewable energy sector a strong research and development focus has resulted in the growth of an embryonic tidal stream energy industry. Previous assessments of the tidal stream resource appear to have neglected shallow tidal flows. This resource located in water depths of 10-30m is significant because it is generally more accessible for energy extraction than deeper offshore tidal sites and hence a good location for first generation tidal stream arrays or fences. The close proximity to shore may lead to improvements in construction feasibility and economic prospects. The objective of this project is to investigate several aspects concerning the exploitation of shallow tidal flows for energy extraction. Fundamental to this project is the importance of developing research alongside and in conjunction with industrial shallow water prototype projects. The key objectives are: (1) The development and understanding of the use of artificial flow constraint structures in the form of specifically-shaped foundations (herein described as “rampfoundations”) that constrain the flow leading to an increase in the magnitude and quality of power from marine current energy convertors (MCEC) operating in shallow tidal flows. (2) The investigation of seabed and free-surface proximity effects on the downstream wake structure of a MCEC. (3) Commercial shallow water device optimisation; utilising project results to aid with the design and development of full-scale commercial demonstrators. Through theoretical and scaled experimental modelling, and commercial collaboration the project has concluded ramp foundations could be utilised to locally increase tidal flow velocities and increase MCEC output across a tidal cycle in shallow flows. Predicted power benefits are in the region of 5-22% depending on lateral and vertical ramp channel blockage ratios. The ramp width or overall array width must therefore be tuned to the channel width to maximise power benefits. Rampfoundations will thus only be technically viable in relatively narrow channels or ideally in MCEC arrays or tidal fences. Results have shown that the downstream wake length is dependent on and varies with the vertical flow constraint and it is critical that the downstream array spacing of MCECs are tuned to the local flow depth. An optimum device height to flow depth ratio to minimise wake length has been identified. It is hoped that this ramp-foundation concept and the relationship between boundary proximity and wake length will continue to help with the development of a niche shallow tidal energy market

Coastal flooding is a natural hazard causing devastation to many regions throughout the world, induced by the coincidence of high spring tides, large storm surges and waves. To reduce the risk posed by coastal inundation, warning systems have been developed to enable preparations to an expected threat. Although current operational predictions provide invaluable warnings, uncertainty in model formulations and input datasets, can lead to errors in forecasts. In order to provide coastal managers with the best possible information with which to make decisions, recent research has begun to focus on the movement from deterministic to probabilistic forecasting, which aims to explicitly account for uncertainty in the system. This research described the implementation of a regional tide-surge-wave model for the Solent-Southampton Water estuarine system, a region that is likely to experience increased risk of coastal flooding in the coming century. The accuracy of the model predictions were examined relative to in-situ measurements and those obtained from independent systems. Using the model, sources of error were examined and their effects upon the model predictions quantified, with particular reference made to the spatial variability throughout the region. In light of recent research, a probabilistic modelling approach, utilising a Monte Carlo technique used to provide a forecast capable of representing the uncertainty in the system, within a suitable time-frame for real-time flood forecasting that included an hourly Kalman filter data assimilation update. The findings presented in this thesis will be of interest to coastal modellers working in complex estuarine environments where the influences of tide-surge-wave interactions upon model predictions are uncertain. Furthermore, the application of a computationally efficient model, presented here, will provide a useful comparison with traditional physically-based systems to those wishing to quantify uncertainty in regions where computational resources are low

Satellite and in-situ radiometric measurements of sea surface temperature (SST) together with conventional SST and meteorological parameters are used to provide a description of the ocean surface skin temperature deviation (skin temperature - bulk temperature, AT) for a transect made across the Atlantic ocean from 50°N 00°W to 23°S 35°W during September and October 1992. Methods of in-situ SST measurement are discussed and the errors associated with each technique are given. The principles of infra red radiometry are explained. The differences between the calibration strategies used to determine SST using infra-red radiometers from both in-situ and satellite platforms are reviewed and the errors associated with each technique are given. Differences between published in-situ infra red SST data indicate that there may be a bias in these data as a consequence of the calibration strategy adopted. The need for an inter calibration of in-situ infra red radiometer systems used for the validaion of satellite SST is highlighted. Satellite SST algorithms are discussed and the principles of atmospheric correction are explained. The difference between the radiometric 'skin' temperature of the ocean and the conventional 'bulk' temperature at depth is defined. A review of current observations of AT is given. Several theoretical treatments of AT are reviewed. The definitions of the surface fluxes of heat and momentum are given. A description of the collection of data and an analysis of the calibration of the infra-red radiometer used to measure the skin temperature is presented. Data have been processed to obtain AT and the surface fluxes of heat and momentum nave been evaluated according to the bulk aerodynamic formulae. The relationships between AT and the measurements made are presented for the entire data set and for day and night time observations separately. Four time series of observed data are presented and the local conditions during the time of measurement are used to discuss AT. AT has a mean value of 0.39°C ±0.3°C and is shown to be a persistent feature of the Atlantic ocean. Correlation analyses reveal the skin and bulk temperature fields to be correlated at length scales > 155 km. Night time correlations are consistently higher than the day time at all length scales. For this reason it is recommended that satellite validation data are only collected during the night. High sea states are shown to affect both in-situ and satellite observations of SST biasing these data warm. The regional nature of AT is presented which is related to the dominant atmosphere-ocean conditions for each region. AT is shown to be greatest at the higher latitudes and weak in the tropical regions. Several parameterisations of AT are used to obtain estimates of AT using the data collected. These are found to be inadequate to predict AT at small temporal scales. A regional dependence of AT is found in these parameterisations. The coefficient A, of the Saunders (1969) parameterisation has been evaluated and is shown to have a regional dependence on the local atmosphere ocean conditions. The coefficient Ci and Ci of the Hasse (1971) parameterisation have been evaluated using the data collected. These are Ci=4.74 and C2=1.22. A comparison between the Along Track Scanning Radiometer Average SST is presented. Satellite - in-situ bulk AT has been obtained and shown to be comparable to that observed in-situ. This comparison highlights the need to make skin SST validation measurements rather than bulk SST measurements. The ATSR ASST data are shown to return a SST accurate to better than 0.3°C.

In-situ sensors are crucially important for understanding the physico-chemical processes that occur in natural water environments. Manual sampling with laboratory analysis cannot provide the temporal and spatial resolution required to characterize marine and fresh water ecosystems, and this approach is both expensive and time consuming, and may also be affected by artefacts during handling and storage. In-situ sensors minimize these drawbacks and provide a tool to obtain long-term data banks which will allow a more synoptic interpretation of the biogeochemical cycles of key elements in water systems. The trace metals iron and manganese are examples of key elements that shape the biogeochemistry of aquatic systems. Processes influenced by them include phytoplankton growth, deep-sea vent chemistry and redox equilibria in environments with strong oxygen concentration gradients. This thesis describes the development, optimisation and application in environment of two sequential prototypes of a Lab-On-A-Chip microfluidic autonomous analyser for the in-situ determination iron and manganese in aquatic environments. A first prototype (Prototype 1,P1) of the device existed at the beginning of this project. It was labtested and deployed at depth in the Lucky Strike Vents Field (Mid Atlantic Ridge) for the determination of Fe(II). An operative fault during the deployment triggered a trouble shooting process which highlighted some weak points in the device. Those weaknesses were addressed and solved in a second version of the device (Prototype 2, P2) whose novel feature was the in-line mixing by diffusion of reagents and samples. Total Fe, Fe(II) and Mn could be measured with a frequency of up to 12 and 6 samples per hour respectively, with limits of detection of 35 nM and 27 nM for Total Fe and Fe(II) and 28 nM for Mn. The robustness and reliability of P2 was tested in the laboratory and in the environment in both marine (Baltic Sea) and fresh (Beaulieu River) waters. The results of these deployments are presented and directions for further developments of the technology are proposed.

This thesis uses a novel, combined mineralogical, geochemical (including stable S iso- topes), and microbiological approach to semi-quantitatively determine Scotia Sea sedi- ment formation processes. The factors that control the localisation of chemosynthetic, microbial consortia in metalliferous sediment beneath Southern Ocean vent fields is investigated along with the impact of hydrothermal venting on sediment composition. Circum-Antarctic ridges represent nearly 40 % of the Earth’s ∼58,000 km ridge crest, but remain severely understudied. In the austral summer of 2009–2010, the Royal Research Ship James Cook expedition JC42 explored the northernmost (E2) and southernmost (E9) bare-rock segments of the East Scotia Ridge, and the sedimented Kemp Caldera (a southern feature of the South Sandwich Arc), and collected >20 co-registered vent fluid, chimney sulfide and hydrothermally-influenced sediment samples using the ISIS remotely-operated vehicle. The hydrothermal materials from E9 and the Kemp Caldera are the focus of this thesis. E9 sediment composition is controlled by the simple mix- ing of >90 % local basalt that is affected by subduction-related and enriched mantle components, <10 % particulate plume fallout, which is dominated by an Fe-, Cu-, Zn-, Ba-, and Pb-rich, near-plume phase, and <1 % collapsed chimney material. The ma- jor, minor, trace, and rare earth element sediment content at E9 is largely determined by proximity to active venting. The thin sediment cover throughout E9, indicates an early stage of sediment formation and the recent onset of venting at this site. Kemp Caldera sediment components include 55–60 % phreatomagmatic shards of local basalt that were most likely deposited by a recent, volcanic event, 30–45 % crystalline elemental S derived from the magmatic disproportionation of SO2 (identified by a δ34S signature of +4.8 ‰ to +5.9 ‰), and 0–10 % buoyant plume particles (rich in P, K, Mn, Fe, and the rare earth elements). Biogeochemical Fe, Mn, and S cycling is investigated in two very different sediment systems of the Kemp Caldera: Toxic Castle and Tubeworm Field. Toxic Castle sediments are compiled from the episodic deposition of magmatic and hydrothermal components, while pore fluid composition is strongly influenced by diffuse, upwelling hydrothermal fluid. The original magmatic-hydrothermal signature is diagenetically altered in the solid phase Tubeworm Field sediments, likely initiated by dissimilatory sulfate reduction. Pore fluid Fe and Mn redox zonation in the surface sediments at Tubeworm Field is typical of biogeochemical cycling in stratified marine sediments. Microbial cell counts (identified by 4’,6-diamidino-2-phenylindole staining and fluorescent in situ hybridisation microscopy) are relatively consistent across the ma- jor Tubeworm Field redox boundaries, although there is a significant downcore shift in the microbial community structure. A dominant presence of δ-, E-, and γ-proteobacteria (which host known S metabolisers), confirms active, microbial S cycling in the deeper Tubeworm Field sediments. Sediment descriptions from modern hydrothermal (partic- ularly back-arc basin-associated) systems are relatively scarce, in comparison to those of vent fluid and chimney material, which is surprising given the potential economic im- portance of hydrothermally-derived, metalliferous, rare earth-enriched sediments. This thesis increases our knowledge of sediment formation processes and subsequent biogeo- chemical cycling, in a range of back-arc-associated bare-rock and sedimented systems, along poorly-surveyed circum-Antarctic ridges, and accentuates the requirement for continued, interdisciplinary hydrothermal surveys of the global, submarine ridge system. We must fully understand the complex interaction of geological, chemical, and biologi- cal components that constitute the complete hydrothermal system, before we allow the commercial exploitation of unique ecosystems that have forever changed our perception of life in the deep sea.