The aim of this thesis has been to reconstruct paleoenvironment, paleohydrology and paleoclimate in coastal southern Mozambique, with emphasis on tracing past flooding events on the lower Limpopo River floodplain. In order to extend flood chronologies beyond periods covered by instrumental data, sediments from lakes on the floodplain were studied (Lake Lungué, Coassane Oxbow, Lake Magandane and Lake Soane). Past sea-level variations and climate changes were deduced by analyzing sediments from coastal sites north of the floodplain area (Lake Chilau, Lake Nhauhache and Macassa Bay). To achieve the established objectives, a multi-proxy approach was applied on most of the retrieved sediment cores, involving analysis of mineral magnetic parameters, grain-size and organic carbon in combination with analysis of microfossils such as diatoms and/or phytoliths. Chronologies for the constructed time-series analysis were obtained by radiocarbon dating and age-depth modelling. The synthesized data from the sampled sites on the Limpopo River floodplain suggest that the area was affected by at least 16 flooding events of variable magnitudes during the studied period. These are dated to c. AD 940, 980, 1040, 1100, 1250, 1300, 1370, 1580, 1665, 1730, 1755, 1855, 1920, 1945, 1970 and 2000. In calibrated years BP these ages correspond to 1010, 970, 910, 850, 700, 650, 580, 370, 285, 220, 195, 95, 30, and 5 cal yrs BP. The two youngest are dated to 20 and 50 years AP (After Present being 1950). Proxy data further suggest that southern Africa was subject to two periods of sea-level highstands, at c. 5000–4200 BC (6950–6150 cal yrs BP) and AD 300–950 (1650–1000 cal yrs BP). The former represents the middle part of the postglacial climatic optimum. The wettest period in the Limpopo River floodplain was reported between AD 1360 and 1560 (590 and 390 cal yrs BP) in the Lake Lungué record, while Lake Chilau experienced wet conditions between AD 1200 and 1400 (750 and 550 cal yrs BP), then returning to drier conditions that prevailed until c. AD 1600. In Lake Nhauhache, however, drier conditions prevailed from c. AD 1200–1700 (750–250 cal yrs BP), shifting towards wetter at c. AD 1900 (50 cal yrs BP). The deviating signals between records can partly be explained by Lake Lungué basin being located on the Limpopo River floodplain, responding to flooding events associated with precipitation upstream the drainage area. Therefore, wet and dry periods in floodplain lakes (e.g. Lake Lungué) are not expected to correlate with precipitation changes on a local scale, as indicated by e.g. Lake Nhauhache. This is supported by a relatively weak agreement between Lake Lungué record and other nearby records (outside the floodplain), but a better correlation with records from the upper catchment, where a more regional climate signal is provided of the southern African summer rainfall region. / Denna avhandling behandlar studier av paleomiljöer i södra Mozambique med fokus på översvämningshistorik i Limpopo-flodens nedre lopp. Tidsspannet är mitt- och sen-holocen och inkluderar även klimat och havsnivåvariationer. Följande analysmetoder har applicerats på ett flertal sedimentkärnor för att belysa ovanstående frågeställningar: mineralmagnetiska parameterar, kornstorlek, organiskt kol samt kiselmikrofossil (diatoméeter och fytoliter). Åldrar har bestämts med 14C-metoden.  Analyserna visar att mineralmagnetiska parametrar är lämpliga för att identifiera översvämningar, speciellt magnetisk susceptibilitet och SIRM. Dessa parametrar visar på ett sediments magnetiska kornstorlerkar och koncentrationen av finkorniga magnetiska mineral. Även kornstorleksanalyser fungerade som ett bra komplement till dessa metoder. Sammantaget har ett minimum av 16 översvämningar registrerats i Limpopo/flodens nedre lopp under de senaste 1100 åren. Särskilt stora översvämningar har daterats till AD 1250 (700 kal år BP), AD 1370 (580 kal år BP), AD 1580 (370 kal år BP), AD 1855 (95 kal år BP), AD 1920 (30 kal år BP), AD 1970 (20 kal år AP) och AD 2000 (50 kal år AP), där AP betyder ”after present”, vilket är 1950. Mindre kraftiga översvämningar har daterats till AD 940 (1010 kal år BP), AD 980 (970 kal år BP), AD 1040 (910 kal år BP), AD 1100 (850 kal år BP), AD 1300 (650 kal år BP), AD 1665 (285 kal år BP), AD 1730 (220 kal år BP), AD 1755 (195 kal år BP) och AD 1945 (5 kal år BP). Skriftliga källor visar på att åtminstone åtta översvämningar har ägt rum under de senaste 62 åren. Under samma period har denna undersökning registrerat två händelser, vilket indikerar att endast ett fåtal har registrerats med ovan nämnda metoder. Det är sannolikt att sedimentationsmönstret under en översvämning är komplicerat och inte heller likartad från gång till gång. Detta innebär att ett flertal sedimentkärnor behöver analyseras och dateras väl. Ytterligare en försvårande faktor är flodens mycket aktiva meandringsaktivitet. Analys av diatoméer har visats vara värdefulla när det gäller översvämningshistorik, klimatförändringar och havsytevariationer. Sedimenten i den något till flodslättens nedre lopp perifert belägna sjön Lungué indikerar fuktiga förhållanden mellan AD 1360 och 1560 (590 till 390 kal år BP). Kombinerade fytolit- och diatoméanalyser av sedimenten i sjön Chilau indikerar fuktiga förhållanden från AD 1200 till 1400 (750 till 550 kal år BP), varefter torrare förhållanden rådde till ca AD 1600 (350 kal år BP). Diatoméanalys av sedimenten i sjön Nhauhache indikerar generellt sett torrare förhållanden mellan AD 1200 och AD 1700 (750 till 250 kal år BP) då klimatet blev fuktigare. De fuktiga och torra förhållanden som dokumenterats i sjön Lungué och översvämningshistoriken korrelerar inte helt med lokala nederbördsförhållanden. Däremot stämmer erhållna data bättre med förhållanden i den övre delen av Limpopoflodens dräneringsområde. Även sjöar och fyllda meanderbågar på själva flodslätten verkar reagera tydligare på översvämningar orsakade av nederbörd i den övre delen av flodloppet. Undersökningen indikerar två faser av en förhöjd havsyta längs kusten i södra Mozambique. En äldre fas är daterad till ca 5000–4200 f.Kr. (6950–6150 kal år BP), vilken representerar den mellersta delen av det postglaciala värmeoptimet. Denna tolkning stöds även av andra undersökningar. En yngre fas har daterats till AD 300–950 (1650–1000 kal år BP), vilken korresponderar till en period med något förhöjda temperaturer i ett globalt perspektiv. / O presente projecto teve como objectivo a reconstrução do paleoambiente, paleohidrologia e paleoclima da costa Sul de Moçambique, com ênfase na identificação de vestígios de eventos de cheias que tenham, no passado, afectado a planície de inundação do Rio Limpopo, com vista a fazer a extensão das datações de cheias para o período anterior aos instrumentos de medição. Deste modo, foram estudados sedimentos recuperados através de testemunhos de sondagem em lagos dentro da planície de inundação (Lago Magandane, Lago Lungué, Lago Soane e o canal abandonado Coassane). As variações do nível do mar e mudanças climáticas do passado foram deduzidas a partir da análise de sedimentos de áreas costeiras a norte da planície de inundação (Lago Chilau, Lago Nhauhache e a Baía de Macassa). Para o alcance destes objectivos, os sedimentos dos testemunhos de sondagem recuperados foram submetidos a várias análises laboratoriais que incluem propriedades de minerais magnéticos, granulometria, teor de matéria orgânica e de microfósseis (diatomáceas e fitólitos). Para o estabelecimento do modelo cronológico foram usadas datações pelo método 14C feitas em matéria orgânica, assim como conchas de gastrópodes e bivalves. A combinação dos resultados obtidos para os pontos de amostragem localizados dentro da planície de inundação do Rio Limpopo, sugerem que esta área foi afectada por um mínimo de 16 eventos de cheias de variada magnitude nos últimos 1100 anos. As cheias de alta magnitude tiveram lugar nos anos 1250, 1370, 1580, 1855, 1920, 1970 e 2000. Por seu turno, as cheias de magnitude moderada ocorrerram nos anos 940, 980, 1040, 1100, 1300, 1665, 1730, 1755 e 1945.. Os anos a negrito são indicativos de eventos de cheias de alta magnitude. O número total de eventos identificados no presente estudo é um mínimo que poderá ter afectado a planície de inundação, tendo em consideração que fontes escritas reportam a ocorrência de pelo menos oito eventos nos últimos 62 anos, tendo o presente estudo revelado somente dois. Isto indica que apenas eventos de cheias intensos podem ser revelados pelos métodos aplicados nesta investigação. Adicionalmente, este estudo mostra a necessidade de aumentar o número de pontos de amostragem para permitir a identificação de mais eventos de cheias ocorridos no passado, devido à diferenciada resposta sedimentológica e geomagnética às cheias, a qual se encontra directamente relacionada à distância do rio meandrante. As análises de microfósseis de diatomáceas permitiram reconstruir o paleoclima e as influências do nível do mar nas áreas de estudo deste projecto. Os registros de diatomáceas do Lago Lungué indicam um período húmido entre AD 1360 e 1560. No Lago Chilau, dados de diatomáceas e fitólitos sugerem condições de humidade entre AD 1200 e 1400. Por seu turno, no Lago Nhauhache, condições de seca dominam entre AD 1200 e 1700. As condições de seca e humidade documentadas no Lago Lungué e os eventos de cheias revelados pelos sedimentos analisados da planície de inundação do Rio Limpopo não mostram total correlação com períodos de baixa e alta precipitação a nível local, mas apresentam boa correlação com evidências a montante na área de drenagem do Rio Limpopo. Em geral, lagos activos ou preenchidos por sedimentos mostrarão ser mais susceptíveis a cheias com origem a montante do que a nível local. As investigações conduzidas permitiram identificar dois períodos de alto nível do mar na região sul de Moçambique. A fase mais antiga teve lugar há cerca de 5000–4200 BC (6950–6150 anos do calendário BP), representando a fase mais recente do pós-glaciar climático óptimo (postglacial climatic optimum, PCA). A fase recente é datada de AD 300–950 (1650–1000 anos do calendário BP), correspondendo a um período de temperaturas relativamente altas a nível global. / Climate and Environmental Research

Geomorphic features have been one of the major tools for sea-level change studies. The present work shows an example of sea-level change study on karst terrain in the tropics. Sea-level notches as well as flank margin caves were identified in the research area and their elevation measured. The time of formation of the sea-level indicators was constrained by lithology study and dating methods such as facies comparison and U-Th dating. Denudation and uplift were also studied for the same purpose. From this study it can be concluded that sea-level stands within the glacial cycle can cause the formation of flank margin caves and that the position of these sea-level stands can be determined. The research area was estimated to have cumulatively uplifted ~22 m in the past 125 ka years (~0.18 mm/yr) while the surface has been denuded some 8 m in the same span of time (~0.064 mm/yr).

Le niveau de la mer est une des variables climatiques essentielles dont la variabilité résulte de nombreuses interactions complexes entre toutes les composantes du système climatique sur une large gamme d'échelles spatiales et temporelles. Au cours du XXème siècle, les mesures marégraphiques ont permis d'estimer la hausse du niveau de la mer global entre 1,6 mm/an et 1,8 mm/an. Depuis 1993, les observations faites par les satellites altimétriques indiquent une hausse du niveau de la mer plus rapide de 3,3 mm/an. Grâce à leur couverture quasi-globale, elles révèlent aussi une forte variabilité du niveau de la mer à l'échelle régionale, parfois plusieurs fois supérieure à la moyenne globale du niveau de la mer. Compte tenu de l'impact très négatif de l'augmentation du niveau de la mer pour la société, sa surveillance, la compréhension de ses causes ainsi que sa prévision sont désormais considérées comme des priorités scientifiques et sociétales majeures. Dans cette thèse, nous validons d'abord les variations du niveau de la mer mesurées par la nouvelle mission d'altimétrie satellitaire, SARAL-AltiKa, en comparant les mesures avec celles de Jason- 2 et des marégraphes. Un autre volet de cette première partie de thèse a consisté à estimer les parts respectives des facteurs responsables des variations du niveau de la mer depuis 2003 en utilisant des observations issues de l'altimétrie satellitaire (missions altimétrique Jason-1, Jason-2 et Envisat), de la mission GRACE, et des profils de température et salinité de l'océan par les flotteurs Argo. Une attention particulière est portée à la contribution de l'océan profond non 'vue' par Argo. Nous montrons que les incertitudes dues aux approches du traitement des données et aux erreurs systématiques des différents systèmes d'observation nous empêchent encore d'obtenir des résultats précis sur cette contribution. Dans la deuxième partie de la thèse, en utilisant les données de reconstruction du niveau de la mer dans le passé, nous étudions la variabilité régionale du niveau de la mer et estimons sa hausse totale (composante régionale plus moyenne globale) de 1950 à 2009 dans trois régions vulnérables: l'océan Indien, la mer de Chine méridionale et la mer des Caraïbes. Pour les sites où l'on dispose de mesures du mouvement de la croûte terrestre par GPS, nous évaluons la hausse locale du niveau de la mer relatif (hausse du niveau de la mer totale plus mouvement de la croûte locale) depuis 1950. En comparant les résultats de ces trois régions avec une étude précédente sur le Pacifique tropical, nous constatons que le Pacifique tropical présente la plus forte amplitude des variations du niveau de la mer sur la période d'étude. Dans la dernière partie de la thèse, nous nous concentrons par conséquent sur le Pacifique tropical. Nous analysons les rôles respectifs de la dynamique océanique, des modes de variabilité interne du climat et du forçage anthropique sur les structures de la variabilité régionale du niveau de la mer du Pacifique tropical depuis 1993. Nous montrons qu'une partie importante de la variabilité régionale du niveau de la mer du Pacifique tropical peut être expliquée par le mouvement vertical de la thermocline en réponse à l'action du vent. En tentant de séparer le signal correspondant au mode de variabilité interne du climat de celui de la hausse régionale du niveau de la mer dans le Pacifique tropical, nous montrons également que le signal résiduel restant (c'est-à-dire le signal total moins le signal de variabilité interne) ne correspond probablement pas à l'empreinte externe du forçage anthropique. / Sea level is an integrated climate parameter that involves interactions of all components of the climate system (oceans, ice sheets, glaciers, atmosphere, and land water reservoirs) on a wide range of spatial and temporal scales. Over the 20th century, tide gauge records indicate a rise in global sea level between 1.6mm/yr and 1.8 mm/yr. Since 1993, sea level variations have been measured precisely by satellite altimetry. They indicate a faster sea level rise of 3.3 mm/yr over 1993-2015. Owing to their global coverage, they also reveal a strong regional sea level variability that sometimes is several times greater than the global mean sea level rise. Considering the highly negative impact of sea level rise for society, monitoring sea level change and understanding its causes are henceforth high priorities. In this thesis, we first validate the sea level variations measured by the new satellite altimetry mission, SARAL-AltiKa by comparing the measurements with Jason-2 and tide gauge records. We then attempt to close the global mean sea level budget since 2003 and estimate the deep ocean contribution by making use of observational data from satellite altimetry, Argo profiles and GRACE mission. We show that uncertainties due to data processing approaches and systematic errors of different observing systems still prevent us from obtaining accurate results. In the second part of the thesis, by making use of past sea level reconstruction, we study the patterns of the regional sea level variability and estimate climate related (global mean plus regional component) sea level change over 1950-2009 at three vulnerable regions: Indian Ocean, South China and Caribbean Sea. For the sites where vertical crustal motion monitoring is available, we compute the total relative sea level (i.e. total sea level rise plus the local vertical crustal motion) since 1950. On comparing the results from these three regions with already existing results in tropical Pacific, we find that tropical Pacific displays the highest magnitude of sea level variations. In the last part of the thesis, we therefore focus on the tropical Pacific and analyze the respective roles of ocean dynamic processes, internal climate modes and external anthropogenic forcing on tropical Pacific sea level spatial trend patterns since 1993. Building up on the relationship between thermocline and sea level in the tropical region, we show that most of the observed sea level spatial trend pattern in the tropical Pacific can be explained by the wind driven vertical thermocline movement. By performing detection and attribution study on sea level spatial trend patterns in the tropical Pacific and attempting to eliminate signal corresponding to the main internal climate mode, we further show that the remaining residual sea level trend pattern does not correspond to externally forced anthropogenic sea level signal. In addition, we also suggest that satellite altimetry measurement may not still be accurate enough to detect the anthropogenic signal in the 20 year tropical Pacific sea level trends.

The port site of Berenike Troglodytica - located on the Egyptian Red Sea coast - served the spice and incense routes that linked the Mediterranean World (specifically the Roman Empire) to India, Southern Arabia and East Africa. In the Greco-Roman period the site was at the cutting edge of what was then the embryonic global economy, ideally situated as a key node connecting Indian Ocean and Mediterranean trade for almost 800 years. It is now located in an arid, marginal, hostile environment but the situation must have been very different 2300 years ago, at the time of its founding. At the time of elephant-hunting trips during the Hellenistic period before the inception of its important role in the global markets of the day in the Roman period Berenike would have to have looked much different to what we can now imagine. What was it like then, when the first prospectors visited this location at the time of Ptolemy II? Why this particular place, and this particular landscape setting seemed such a propitious location for the siting of an important new harbour? Given the importance of the port over almost a millennium it is perhaps surprising that very little is known about the different factors impacting on the foundation, evolution, heyday and subsequent decline of the city; or the size, shape, and capacity of its harbour. The intention of this research is to address this shortfall in our knowledge, to examine the drivers behind the rise and fall of this port city, and to explore the extent to which the dynamics of the physical landscape were integral to this story. Using an innovative Earth Science approach, changes in the archaeological 'coastscape' have been reconstructed and correlated with periods of occupation and abandonment of the port, shedding light on the nature, degree and directionality of human-environment interactions at the site. This work has revealed profound changes in the configuration of the coastal landscape and environment (including the sea level) during the lifespan of Berenike, highlighting the ability of people to exploit changes in their immediate environment, and demonstrating that, ultimately, the decline of the port was partly due to these landscape dynamics. To further explore these themes the landscape reconstructions have been supplemented by semi-quantitative analyses of a suite of variables likely to influence the initial siting of new ports of trade. These have shown that although the site of Berenike was ideal in terms of its coastal landscape potential, possessing a natural sheltered bay and lagoon system, the choice of location was not solely influenced by its environmental conditions. Additionally, a detailed review of vessels that plied Red Sea and Indian Ocean routes is presented here in order to better understand the design and functioning of Berenike's harbour. This serves the purpose of identifying unifying features that provide more detail about the size and draught of vessels and the potential capacity of the harbour basin. By using this multi-scalar approach it has been possible to reconstruct the 'coastscape' of the site through the key periods of its occupancy and those phases immediately before and after its operation. This has wide-ranging implications for researchers studying ancient ports along this trade network as a larger database will tease out more details about how influential the landscape was in the initial siting of the port and its subsequent use and decline.