Iron is one of the most commonly studied trace metals as it exerts a significant influence on ocean productivity, carbon sequestration as well as modulating atmospheric CO2 concentrations. As iron is such a vital nutrient for biogeochemical processes it is often included as a variable in ocean biogeochemical models. In representing the iron cycle, biogeochemical models must parameterise the major processes of uptake by phytoplankton, remineralisation and scavenging. However, there is no generally accepted set of equations to represent iron dynamics and thus a variety of different parameterisations are employed across the modelling community. The thesis work focussed on the inorganic iron parameterisations with an emphasis on the scavenging formalisms which are employed in current biogeochemical models. Using an open-source numerical model (Biogeochemical Flux Model, BFM) as a background model, a more advanced inorganic iron parameterisations that simulates free iron scavenging and ligands linked to dissolved organic carbon (DOC) (from the open-source model PISCES) was included and compared to assess the implications on iron cycling and plankton community structure. The parameterisations were compared by running box models (0D) in four different regions: Southern Ocean, Equatorial Pacific, North Atlantic gyre and North-east Pacific, representing different types of iron dynamics. The free scavenging model (FePISCES) resulted in dissolved iron concentrations being two to three times greater than with the standard formulation (FeBFM), which used a simpler formalism for scavenging. Consequently, the elevated iron concentrations in FePISCES resulted in altered community compositions for phytoplankton which impacted the seasonal cycle of macronutrients and chlorophyll concentrations. Furthermore, the prognostic appreciation of ligand dynamics in FePISCES lead to a decoupling of dissolved iron from its organic species with the DOC content for a region being indirectly implicated in driving the iron system by affecting the scavenging regime. Therefore, using a different set of iron parameterisations will alter the biogeochemical behaviour of a model. The results suggest that the testing of parameterisations should be initially done within 0D models in order to assess any non-linear behaviours and ultimately embedded in 3D models to study how they interact with physics.

In this thesis I explore the effect of parameter uncertainty in ocean biogeochemical models on the calculation of carbon uptake by the ocean. The ocean currently absorbs around a quarter of the annual anthropogenic CO2 emissions to the atmosphere [Scholes et al., 2009], slowing the increase in radiative forcing associated with the increasing atmospheric CO2 concentration. Ocean biogeochemical models have been developed to study the role of the ocean ecosystem in this process. Such models consist of a greatly simplified representation of the hugely complex ocean ecosystem. This simplification requires extensive parameterisation of the biological processes that convert inorganic carbon to and from organic carbon in the ocean. The HadOCC ocean biogeochemical model is a Nutrient-Phytoplankton-Zooplankton-Detritus (NPZD) model that is used to represent the role of the ocean ecosystem in the global carbon cycle in the HadCM3 and FAMOUS GCMs. HadOCC uses twenty parameters to control the processes of biological growth, mortality, grazing and detrital sinking that control the uptake and cycling of carbon in the ocean ecosystem. These parameters represent highly complex and in some cases incompletely understood biological processes, and as a result are uncertain in value. A sensitivity analysis is performed to identify the HadOCC parameters that due to uncertainty in value have the greatest possible effect on the exchange of CO2 between the atmosphere and the ocean—the air-sea CO2 flux. These are found to be the parameters that control phytoplankton growth in the well lit surface ocean, the formation of carbonate by marine organisms and the sinking of biological detritus. The uncertainty in these parameters is found to cause changes to the air-sea CO2 flux calculated by the FAMOUS GCM. The initial effect of these changes is equivalent to the order of the error of current estimates of the net annual carbon uptake by the ocean (2.2 ± 0.3 Pg C y−1 [Gruber et al., 2009], 2.2 ± 0.5 Pg C y−1 [Denman et al., 2007]). This indicates that while the effect of ocean biogeochemical parameter uncertainty is non-negligible, it is within the bounds of the uncertainty of the total (inorganic and organic) ocean carbon system, and is considerably less than the uncertainty in the carbon uptake of the terrestrial biosphere [Houghton, 2007]. However, as the ocean plays a crucial role in the global carbon cycle and the regulation of the Earth’s climate, further understanding and better modelling of the role of the ocean ecosystem in the global carbon cycle and its reaction to anthropogenic climate forcing remains important.

Submitted by Haroudo Xavier Filho (haroudo.xavierfo@ufpe.br) on 2016-07-01T12:51:52Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) TESE_Gaspar_FL_2015.pdf: 3846981 bytes, checksum: 5c71e8dea7a17180048b9ef119b20236 (MD5) / Made available in DSpace on 2016-07-01T12:51:52Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) TESE_Gaspar_FL_2015.pdf: 3846981 bytes, checksum: 5c71e8dea7a17180048b9ef119b20236 (MD5) Previous issue date: 2015-10-28 / CAPES / A plataforma continental Pernambucana é caracterizada pela oligotrofia e estabilidade térmica ao longo do ano influenciada pela corrente Norte do Brasil, que traz águas quentes e ricas em CO2 para a plataforma continental do nordeste brasileiro. Dentre os principais rios translitorâneos de Pernambuco estão os Rios Capibaribe e Jaboatão que recebem os efluentes domésticos do Recife e região metropolitana, o que causa alterações nos ciclos naturais do carbono. Desta maneira, foram avaliadas a variabilidade da pCO2 nos estuários do Capibaribe e de Barra de Jangadas assim como na plataforma continental interna do Estado de Pernambuco. Para análise nos estuários foram realizadas coletas durante a maré baixa, bimestralmente de Novembro de 2010 a Setembro de 2011. O estuário do Capibaribe apresentou os valores mais elevados de alcalinidade total (AT) e carbono inorgânico total (TCO2), com diferença significativa entre os dois rios. As médias anuais de alcalinidade encontradas foram 1649 ± 390 μmol kg-1 no Capibaribe e 1557±315 μmol kg-1 em Barra de Jangadas. Em relação a pCO2, os estuários apresentaram supersaturação de CO2 em relação a atmosfera durante todo o ano, com médias de 3317 ± 2034 μatm no Capibaribe e 6018 ± 4589 μatm em Barra de Jangadas. Estes valores durante o período chuvoso variaram entre os estuários, com diminuição na pCO2 no Capibaribe e aumentos de até 300% dos valores de pCO2 em Barra de Jangadas. A fim de se obter uma melhor avaliação da distribuição espacial da pCO2 na plataforma interna, foi construído um equipamento para a medição contínua e direta da pCO2 na água como alternativa de baixo-custo em relação aos fabricados importados. A primeira campanha realizada aconteceu em Dezembro de 2014, partindo do porto do Recife em direção ao estuário de Barra de Jangadas. Apesar de o período seco ser o de maior produtividade primária a região apresentou-se oligotrófica, com picos de chl-a (8.4 mg m3) próximo ao Capibaribe. Apenas uma pequena área apresentou subsaturação de CO2 em relação à atmosfera, com o valor mínimo registrado de 376,6 μatm. Foi identificada uma elevada fugacidade de CO2 (fCO2) na área da plataforma interna, com média de 474,33 ± 66,57 μatm, resultando em um fluxo médio para a atmosfera de 8,5 ± 6,82 mmol C m-2d-1. A partir dos resultados obtidos foi proposto um modelo para predizer a fCO2 a partir de valores de salinidade e temperatura para a plataforma interna durante o verão. Uma análise espacial e sazonal mais abrangente dos parâmetros do sistema carbonato na plataforma interna foi realizada utilizando-se dados de AT e TCO2 em áreas com e sem influência de rios entre os anos de 2013 e 2014. As médias da pCO2 e do pH na plataforma interna foram de 449 ± 45 μatm e 8,00 ± 0,03 respectivamente. Os parâmetros analisados foram influenciados pela distância da costa, o que resultou na elaboração de modelos para a predição de AT, TCO2, pCO2, pH, ΩCa e ΩAr a partir de dados de temperatura, salinidade e longitude. A região apresenta elevada alcalinidade inclusive na área influenciada por rios, onde foi encontrada a menor média de AT 2358 ± 28 μmol kg-1. A região apresenta, em curto prazo, baixa vulnerabilidade ao processo de acidificação oceânica. Os valores encontrados de pCO2 e fluxos de CO2 na plataforma continental de Pernambuco estão acima dos estimados nas médias globais para plataformas continentais abertas baseadas em modelos matemáticos. A variabilidade na pCO2 encontrada está ligada à própria característica da água tropical do Atlântico Sul, que cobre a rasa plataforma pernambucana com águas quentes. Onde a baixa solubilidade do CO2 devido à alta temperatura, associada à oligotrofia e aportes de matéria orgânica fazem com que a região da plataforma seja fonte de CO2 para atmosfera durante todo o ano. / The continental shelf of Pernambuco is characterized by oligotrophy and thermal stability throughout the year, influenced by the North Brazil current that brings warm water, rich in CO2 to the continental shelf of northeastern Brazil. The Capibaribe and Jaboatão rivers are one of the major rivers of Pernambuco. They receive a large input of domestic effluents from Recife and its metropolitan area, which causes changes in the natural carbon cycle. Here we evaluated the pCO2 variability in these two estuaries and in the inner shelf area of Pernambuco. To evaluate the role of the estuaries, samples were collected bimonthly during low tide, on a seasonal cycle from November 2010 to September 2011. The Capibaribe estuary presented the highest values of total alkalinity (TA) and total inorganic carbon (TCO2), with a significant difference between the two rivers. Both estuaries showed seasonal variation in TCO2 and TA values. The annual TA averages were 1649 ± 390 μmol kg-1 in the Capibaribe and 1557 ± 315 μmol kg-1 at Barra de Jangadas. Regarding the pCO2, the estuaries were saturated in CO2 throughout the year, with average for the dry season of 3317 ± 2034 μatm in the Capibaribe and 6018 ± 4589 μatm in Barra de Jangadas. These values varied seasonally between estuaries, with a pCO2 decrease during the rainy season in the Capibaribe, and an increase of up to 300% in the Barra de Jangadas estuary. In order to obtain a better evaluation of the spatial distribution of pCO2 over the shelf waters, it was developed an equipment for the continuous and direct measurement of sea surface pCO2 as low-cost alternative to the commercially manufactured ones. The first cruise took place in December 2014, starting from the port of Recife towards the Barra de Jangadas estuary. Even though the dry period has shown the highest primary productivity, the region presented itself as oligotrophic, with peaks of chl-a (8.4 mg m3) under influence of the Capibaribe plume. Only a small area showed CO2 subsaturation to the atmosphere, with the minimum value of 376.6 μatm. In general, a high CO2 fugacity (fCO2) was identified in the entire area of the inner shelf, with average of 474.33 ± 66.57 μatm, resulting in an mean flux to the atmosphere of 8.5 ± 6.82 mmol C m-2 d-1. It was proposed a model to predict the fCO2 using salinity and temperature values for inner the shelf during the summer. A more comprehensive spatial and seasonal distribution of the carbonate parameters was conducted by using values of TA TCO2 in areas subjected and non-subjected to riverine inputs during 2013 and 2014. The area presents a relative stability to what concerns the pCO2 and pH distribution, with overall average values of 449 ± 45 μatm and 8.00 ± 0,03 respectively. In general, the parameters analyzed were influenced by the distance from the coast, resulting in the development of prediction models for TA, TCO2, pCO2, pH, ΩCa and ΩAr using temperature, salinity and longitude data. The area has a low vulnerability to ocean acidification process, the TA is high even in the areas under riverine influence, where the lowest average values was registered 2358 ± 28 μmol kg-1. The values of pCO2 and CO2 fluxes found on the continental shelf of Pernambuco are above those expected on global averages estimated by mathematical models for open continental shelf. The variability in pCO2 found is linked to the own characteristic of the South Atlantic tropical water, which covers the shallow Pernambuco platform with warm waters. The low solubility of CO2 due to high temperature, combined with oligotrophy and the transport of organic matter to be respired outside the estuaries, turns the continental shelf into a source of CO2 to the atmosphere throughout the year.

This research aims to integrate mathematical water quality models with Bayesian inference techniques for obtaining effective model calibration and rigorous assessment of the uncertainty underlying model predictions. The first part of my work combines a Bayesian calibration framework with a complex biogeochemical model to reproduce oligo-, meso- and eutrophic lake conditions. The model accurately describes the observed patterns and also provides realistic estimates of predictive uncertainty for water quality variables. The Bayesian estimations are also used for appraising the exceedance frequency and confidence of compliance of different water quality criteria. The second part introduces a Bayesian hierarchical framework (BHF) for calibrating eutrophication models at multiple systems (or sites of the same system). The models calibrated under the BHF provided accurate system representations for all the scenarios examined. The BHF allows overcoming problems of insufficient local data by “borrowing strength” from well-studied sites. Both frameworks can facilitate environmental management decisions.

This research aims to integrate mathematical water quality models with Bayesian inference techniques for obtaining effective model calibration and rigorous assessment of the uncertainty underlying model predictions. The first part of my work combines a Bayesian calibration framework with a complex biogeochemical model to reproduce oligo-, meso- and eutrophic lake conditions. The model accurately describes the observed patterns and also provides realistic estimates of predictive uncertainty for water quality variables. The Bayesian estimations are also used for appraising the exceedance frequency and confidence of compliance of different water quality criteria. The second part introduces a Bayesian hierarchical framework (BHF) for calibrating eutrophication models at multiple systems (or sites of the same system). The models calibrated under the BHF provided accurate system representations for all the scenarios examined. The BHF allows overcoming problems of insufficient local data by “borrowing strength” from well-studied sites. Both frameworks can facilitate environmental management decisions.