Browsing by Author "Mahongo, Shigalla Bassanda"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Analysis of met-ocean processes in relation to changes in large scale climatic systems in Tanzania and the western Indian Ocean
    (University of Dar es Salaam, 2014) Mahongo, Shigalla Bassanda
    The recent changes in world climate have highlighted the need for a proper understanding of the factors that influence changes in met-ocean processes in Tanzania and the western Indian Ocean. The region is unique in that these changes have generally not been fluctuating in consistence with the global change processes, thus posing a notable scientific challenge. This thesis addresses how these processes have been changing overtime in the region, and how they are associated with the large scale climatic phenomena. The thesis also discerns the oscillatory patterns of the met-ocean parameters, and investigates their potential for prediction. The study is implemented through wavelet and spectral analyses, multivariate linear regressions and other statistical tools, while prediction is carried out through artificial intelligence and linear time series modelling. Key findings of this study include the revelation of significant warming trends of surface air temperature, general strengthening of winds, decline in rainfall, and the presence of a quasi-22-year cycle on the low frequency timescales of winds and sea level, which mimic the Hale solar cycle. The influence of the 22-year rhythm is larger in amplitude than those due to large scale climatic systems. At the Island of Zanzibar for instance, the short sea level record of 1984-2004 largely resonated with the 22-year rhythm thus leading to a spurious negative trend. However, the relationship between solar activity and met-ocean processes is complex and varies with time and space such that this pattern cannot be generalized for other locations in the region. This thesis has also revealed that neural networks and state-of-the-art time series models are capable of making excellent predictions of sea surface temperature, thus providing potential for prediction of other met-ocean parameters. Determination of low frequency oscillations in met-ocean parameters is dependent on the length of the record, with about 100 years of data being needed for any conclusive evidence on trends in climate. Although ocean monitoring is currently possible through satellites, instrumental monitoring is still needed for calibration of the satellite sensors. Continued in-situ and satellite monitoring of the met-ocean parameters is therefore essential for a better understanding of the processes, permitting sound prediction and enabling proper planning, adaptation and resilience.