Landfast sea ice is a recurring seasonal feature along many coastlines in the polar regions. It is characterised by a lack of horizontal motion, for at least 20 days, and its attachment to the coast or seabed. It can form as a result of restrictive geometry, such as channels or embayments, or through the grounding of thick ice ridges which add lateral stability to the ice cover. Due to its stationary and persistent nature, landfast ice fundamentally modifies the exchange of heat and momentum between the atmosphere and ocean, compared with more mobile pack ice. The current generation of sea ice models is not capable of reproducing certain aspects of landfast ice formation and breakup. In this work two landfast ice parameterisations were developed, which describe the formation and breakup of landfast ice through the grounding of thick ice ridges. The parameterisations assume the sub-grid scale distribution of ice draft and ocean depth, the two parameters important in determining the occurrence of grounded ridges. The sub-grid scale distribution of grounded ice is firstly defined by assuming that ice draft and ocean depth are independent. This parameterisation allowed ice of any thickness to occur and ground at any depth. Advancing from this the sub-grid scale distribution of the grounded ice was restricted in an effort to make it more realistic. Based on Arctic ice scour observations ice was prevented from grounding in regions where the draft thickness was much larger than the ocean depth. Both parameterisations were incorporated into a commonly used sea ice model, the Los Alamos Sea Ice Model (CICE), to which a multi-category ocean depth distribution from high resolution global bathymetry data (ETOPO1) was included. The parameterisations were tested in global standalone format (i.e. no active ocean) with realistic atmospheric forcing. Both parameterisations were found to improve the spatial distribution and the seasonal cycle of landfast ice compared to the control (i.e. no landfast ice parameterisation) in the Arctic and Antarctic. However, the grounded ridges produced by the parameterisations were very stable, and tended to become multiyear leading to the production of multiyear landfast ice, which was particularly widespread in the Antarctic. It was found that tides have a significant impact on both grounded and landfast ice. In some polar locations tides were found to increase the occurrence of landfast ice, by increasing the production of thick ridges which were able to ground. Conversely, in some regions, tides were found to decrease the occurrence of landfast ice, as strong tidal and residual currents increased the mobility of the grounded ridges and landfast ice. This thesis finishes by considering whether a sea ice model could be used to further our understanding of the physical landfast ice system. Analytically derived characteristic numbers, which describe the ability of landfast ice to form, were found to fully describe the formation of landfast ice within the sea ice model CICE during idealised 1D scenarios. For these scenarios the key parameters controlling ice motion were found to be the external forcing component, the width of the ice cover, the internal ice strength, and the thickness of the ice. However, an exact characteristic variable able to describe the occurrence of landfast ice in an idealised 2D scenario could not be found analytically, nor could it be inferred numerically, and this remains an area for further research. This thesis examines different methods of modelling landfast sea ice and provides the sea ice modelling community with a means to parametrise landfast ice formation as a result of grounded ridges without having to work at very fine resolution, as this is computationally inefficient.

Shelf seas are believed to play an important role in the oceanic export of carbon (C). The combination of enhanced primary productivity partly induced by highly energetic mixing, together with the euphotic zone being in close proximity to the seabed, make the shelf seas highly efficient region for biological C sequestration. The high productivity observed in shelf seas also has high economical significance by supporting >90% of the global fish catch. A subsurface peak in biomass, termed the subsurface chlorophyll maximum(SCM), at the base of the thermocline occurs as a result of phytoplankton being nitrate limited in the surface layer. The SCM is believed to be responsible for as much primary production as the spring bloom. The turbulent supply of nutrients across the thermocline, driven by the internal tide and wind-driven inertial oscillations, is believed to support growth at the SCM. However, although inertial oscillations are recognised as an important mixing mechanism, the turbulent flux of nitrate that they supply has not yet been estimated. In this thesis the importance of diapycnal fluxes of nutrients, particularly those generated by wind-driven inertial oscillations, is investigated in terms of sustaining productivity in the SCM, using a Lagrangian numerical model, observational and experimental approaches. The results in this study indicate that diapycnal nitrate fluxes limit new production at the SCM, and mixing generated by wind-driven inertial oscillations play a key role in supporting primary production at the SCM. In addition, this study demonstrates that the background nitrate flux, as well as transport of nitrate via dinoflagellate migration, are too low to sustain primary productivity at the SCM. Our observations demonstrate that wind-driven large, short-lived dissipation events increase the daily nitrate flux to the SCM by a factor of at least 17, supplying the SCM with ~33% to 71% of the nitrate required fornew production in shelf seas during summer. Thus it is intermittent mixing events which must supply the required nitrate for new production at the SCM. Using an experimental approach, a wind event was simulated in order to estimate the impact of mixing of water from the SCM and BML on primary and secondary production in the SCM. Nutrients were assimilated rapidly and phytoplankton production increased as expected. However, up to two thirds of the C fixed was exuded as DOC and there was a significant increase in bacterial activity. As well as supplying nutrients to the SCM, we show that diapycnal mixing may also be important in redistribution plankton, especially bacteria, which subsequently impacts both the inorganic and organic nutrient pools. Thus the sensitivity of the SCM, in terms of autotrophy and heterotrophy, to diapycnal mixing is highlighted. This study highlights the importance of short-lived events in supplying nutrients to the SCM. Concluding that these are likely undersampled and also not well represented in shelf sea models. In addition, this thesis demonstrates that there is a need to consider secondary as well as primary production to fully understand the C cycle in the shelf sea.

Marine primary productivity by phytoplankton drives the conversion of CO2 into organic carbon, thus supporting key marine ecosystem services such as biogeochemical nutrient cycles and fisheries. Therefore, the ability to accurately quantify and predict the environmental influences that govern this process is of paramount importance. Through surveying and experimental approaches, I investigated the variability of phytoplankton physiology and biomass across a range of light and nutrient environments. I found that growth and physiological responses differ in geographically distinct regimes of nutrient limitation across the North Sea. Variable nutrient stoichiometry across the region could therefore lead to alternative growth and productivity rates that reflect the populations suited to the different conditions. Fast repetition rate fluorometry (FRRf) measurements of photosystem II (PSII) characteristics during a spring bloom were shown to change with hydrography and community composition. Thus, providing further support that the physiological state of PSII can be used as an indicator of bloom status and community composition. FRRf-derived productivity parameters were also measured and shown to change in response to changes in taxonomy. Through geostatistical approaches the spatial distribution of phytoplankton was characterised during a spring bloom. I show both in situ and satellite sampling approaches possess the ability to capture mesoscale variability in phytoplankton distribution, but ocean colour estimates lose accuracy in highly heterogenous conditions. This thesis provides a step towards capturing the extent of spatial and temporal variability in phytoplankton stocks and rates in temperate shelf seas, in part, by providing a better understanding of the strengths and limitations of the use of fluorescence-based measurements.

No description available.

The effect of drought on freshwater ecosystems is a growing environmental concern. This study aimed to determine the effect of the supra-seasonal 2010-2012 drought on the macroinvertebrate communities and functional feeding groups (FFG) of chalk streams. Three rivers were sampled during and after the drought, three sites were analysed on each river, a perennial control site, a dewatering site (shallow during drought) and a rewetted site (dry during drought). All wetted sites were sampled for macroinvertebrates, algae, velocity and depth monthly. In addition an in-situ experiment investigated grazing rates in dewatering, rewetted and perennial sites and a mesocosm experiment investigated the effect of reduced water velocity and temporary drying on grazing function and macroinvertebrate mortality. Macroinvertebrate data were analysed for structural and functional changes in communities and algal data were analysed for changes of biomass both during drought and recovery. Both found rapid recovery post-drought with high incidence of resilience in the macroinvertebrate communities. Functional recovery was determined by investigating grazing rates in dewatering, rewetted and perennial sites. Algal and macroinvertebrate biomass, and grazing function recovered within a month of rewetting. Despite a rapid recovery of overall biomass and grazing function, FFG results showed delayed effects, and four months following drought differences were still present. Mesocosm experiments showed that macroinvertebrate mortality increased due to dewatering for some species, however slower velocity had little effect. Thus, this study showed that ecological function can return despite changes to macroinvertebrate community structure, indicating that functional redundancy promotes the resilience of chalk stream communities to drought.

In this study fluorescent sand tracer techniques were used to investigate sediment movement on the foreshore at two sites on the south Lincolnshire coast. Working over one tidal cycle, low water to low water tracer release at different points across the beach revealed a complexity of sand movement in this strongly tidal environment. It was found that sand moved both at different rates and in contrasting dominant directions on different parts of the beach over the same tidal cycle. At Gibraltar Point there was confirmation from the tracer experiments of the importance of tidal generated currents on sand movements on the lower foreshore. Tests conducted to study the patterns and rates of movement of different grain sizes produced inconclusive results largely due to low recovery rates for the tracer. However, from the evidence available, it appeared that sorting of sediment was taking place only in the sense that finer material was moved away from the release point at a faster rate than coarse material and not because different sized grains were moved in different directions. In a consideration of models for the prediction of longshore sand transport two of the most commonly used models were tested using data collected during the tracer experiments. The results confirmed the success of the two models, one based on the wave power equation and the other on an energetics approach, and coefficients produced were in close agreement with those obtained by Komar (1969) in a previous study. Finally, field measurements of a series of variables such as wave height, wave period and longshore current velocity were combined with measures of sand movement and direction in a linear multiple regression analysis to study the associations present and to produce a simple predictive model of sand movement. Using both stepwise and combinatorial methods of regression it was found that 80% of the variation of amount of longshore sand movement could be accounted for by the 'best' equation. Wave height alone explained 61.3% of the variation but beach slope, water temperature and longshore current velocity were also of importance. 59% of the variation of average distance moved by sand grains normal to the shore was explained by beach slope, whilst wave period was seen to be the major factor in determining the direction of sand movement onshore or offshore. A set of equations was produced which together with longshore current flow direction, could be used to predict the average position of tracer on the beach face after one tidal cycle. At the same time individual equations could be used to model specific aspects of sand movement.

The region where deep open-ocean convection occurs in the Greenland Sea corresponds to that where a sea ice winter feature, the Odden, usually forms. The role of sea ice in modifying the surface waters to overturn to depth is evaluated through the combination of in siu measurements, satellite imagery, meteorological measurements and drifting buoy data. Results suggest local meteorological and oceanographic conditions govern the ice conditions over the region. The high ambient wave energy precludes the formation of ice beyond the frazil-pancake stage; the changing surface pressure field, due to passing storm systems, influences the daily shape and extent of the Odden and enables pancake ice to expel brine at an increased rate. Finally, the analysis of drifting buoy data reveal that the ice is in free drift. t These characteristics suggests the Odden may be regarded as a large scale latent heat polynya, with the predominately northerly winds blowing newly formed sea-ice constantly southward such that it melts in a different area from that of its formation. This salt separation process whereby the majority of brine is deposited where the ice was formed, and a smaller amount being released, through brine drainage, as the ice drifts with the prevailing wind has important consequences for the spatial and temporal distribution of the salt flux and hence surrounding hydrography. This is clearly demonstrated through the development of a salt flux model, which involves brine drainage and drift. A simple one-dimensional mixed layer model, driven by results of the salt flux model, predicts a strong density enhancement and deepening of the mixed layer over time. It is therefore envisaged that the formation of sea ice, brine drainage and drift are fundamental in eroding the pycnocline between the surface waters and those below. Sea ice should therefore be viewed as a preconditioning activity to deep overturning of the waters of the central Greenland Sea.

The Argo dataset is used to obtain estimates of the heat storage and heat divergence with the aim of the assessing the usefulness of the Argo array for investigating the North Atlantic heat budget. The accuracy of the Argo-based mixed layer heat storage varies significantly throughout the North Atlantic. Errors are smallest, around 10-20 Wm-2 on monthly timescales for 10° x 10° boxes, reducing to 5-10 Wm-2 on seasonal scales in the subtropics and eastern basin. Heat storage errors over a fixed 300 m layer are higher, but typically remain below 20 Wm-2 on seasonal timescales away from the western boundary. The heat budget is closed (using net heat fluxes from the NCEP climatology and NOC reanalysis) within the estimated error throughout the subtropical and eastern North Atlantic, indicating the value of the Argo dataset in studies of this nature. However, within the western boundary the heat budget residual typically exceeds 50 Wm-2, with the heat storage overestimated or the heating from the net heat flux and/or advective and diffusive divergence underestimated. Assuming that heat storage error estimates are accurate and considering results in the literature regarding the bias in net heat flux products, it is likely that heating from divergence is underestimated. The heating contribution from this term may be large on scales that cannot be resolved using Argo. In the eastern and subtropical North Atlantic, the errors in the Argo-based heat budget terms are smaller than the uncertainty in the net heat flux products and can thus be used to provide insight into which atmospheric dataset (the NCEP reanalysis or the NOC climatology) may be more accurate. The NOC net heat flux is more accurate than that from NCEP throughout the year in the subtropics and during the first half of the year in the eastern mid-latitudes. The errors in the mixed layer heat storage are smaller than the interannual variability in this term. Thus Argo can be used to investigate variability on this scale. While the current Argo dataset is on the short side for studies of this nature, continued funding of the array is expected to provide more insightful results.

SeaWiFS, and the other new generation of ocean colour sensors, must be backed by comprehensive calibration and validation programmes if they are to achieve their aims of accurate, synoptic, global measurement of oceanic phytoplankton biomass and production. In situ measurements of normalised water-leaving radiance {L^) made simultaneously with satellite measurements can complement on-orbit methods of tracking changes in the calibration of the satellite radiometer and allow end-to-end vicarious validation of the whole remote sensing method, including new atmospheric correction methods. A moored, optical databuoy was developed to measure L^ in the western English Channel for extended periods. Tests indicate that the buoy is capable of measuring spectral incident irradiance with less than 10% error and water-leaving radiance with less than 20% error; the errors are reduced by averaging the data. There were 24 match-ups with good quality SeaWiFS data at the buoy site during the 10 months of deployment between May 1997 and September 1998. The RMS differences between the buoy and SeaWiFS measurements ofLw varied from 49.5% at 555 nm to 101.5% at 412 nm. This leads to the diffuse attenuation at 490 nm [^490)] being systematically overestimated by the SeaWiFS algorithm by between 50 and 70%, with an RMS difference of 68.6%. Near surface chlorophyll-a concentrations (Ca) differ from those based on the buoy optical measurements by between -52% (SeaWiFS estimate lower than buoy estimate) to +123%, with an RMS difference of 57%. The RMS differences in measurements of Lwn could be reduced to less than 13% by increasing the calibration coefficients of SeaWiFS by between 3.7% and 0.25% in the visible bands and by tuning the extrapolation of aerosol radiances from the near infra red to the visible wavelengths. This study also found that remotely sensed measurements of A^490)an^ Ca may be biased because of undersampling.

This work focuses on the effects of the large Mediterranean seagrass Posidonia oceanica on coastal waves and wave-induced flows, which has significant implications for coastal protection. Investigations were made on both a natural shallow Posidonia oceanica bed and, in controlled conditions of full-scale Posidonia mimics under regular and irregular waves. In the field, waves and currents were monitored during low energy conditions and a Mistral wind event. Data were collected on the distribution of Posidonia patches, density and canopy height, as well as bed sediment type and bathymetry. In the flume, measurements were made of water surface elevation along the flume and oscillatory flows at 3 locations and 4 elevations, under several wave conditions, water depths and for 2 canopy densities. The mimics were designed carefully to recreate the hydraulic behaviour of Posidonia plants under waves. Field results indicate that shallow Posidonia meadows are effective at reducing wave energy under low wave energy conditions and small wave amplitudes. The flume experiments confirm this trend. Under both regular and irregular waves, drag coefficients decrease with increasing Reynolds vegetation numbers; wave dissipation factors decrease with wave orbital amplitude. Under spectral waves, most wave energy dissipation occurs at the peak spectral frequency and it is largest for the least energetic wave spectra. At high wave Reynolds numbers, the canopyinduced hydraulic roughness (r) appears to be a function of the canopy element density only, and the empirical formula of Nielsen (1992) is successfully applied. However, more work is required in low energy conditions to examine the range of validity of the formula. In natural conditions under small amplitude waves, attenuation of wave-induced flows is negligible in the upper canopy; flume experiments confirm this trend. The typical flow intensification at the canopy top, measured for other seagrasses, occurs only for tests with the largest wave amplitudes, whilst, under smaller waves, flow intensification is located within the upper part of the canopy. In the lower canopy, flows are always reduced and flows decelerate exponentially with increasing orbital amplitude. This is a novel observation in flexible canopies. The artificial canopy, like the natural Posidonia bed, enhances flow asymmetries at the canopy top, especially under waves with large wave orbital amplitudes. This is thought to be a mechanism to enhance shoreward drift. Turbulence in the artificial canopy, under regular waves, peaks at the canopy top, as occurs under unidirectional flows and for other seagrass beds exposed to waves. Vertical turbulent exchanges are enhanced at the edge of the seagrass patch and are larger for lower submergence ratios (the ratio of canopy height to water depth). A reduction in submergence ratio in the flume, also causes increased shear stresses at the top of the canopy, lower wave height decay and reduced oscillatory flow attenuation in the lower part of the canopy. The denser canopy, in the conditions tested, increases relative roughness (r/A), wave attenuation, in canopy oscillatory flow reduction and turbulent kinetic energy at the top of the canopy. Oscillatory flows characterised by small orbital amplitudes can penetrate further into the canopy than larger orbital velocities, inducing a larger drag, thus increasing wave dissipation, as proposed for rigid canopies (corals). This is manifested as a thinner canopy boundary layer under small orbital amplitude waves than the large amplitude waves. A conceptual model is proposed to summarise these findings. Under storm conditions Posidonia is believed to be less efficient at reducing wave energy, however it remains effective at reducing sediment transport locally and, by inducing a preferential shoreward drift, at preventing sand dispersal offshore