The subject of the thesis is the study of the Outgoing Longwave Radiation (OLR) in different timescale over the African and Atlantic region. We have used geostationary satellite data (GERB) to examine some of the shorter timescale radiative variability by utilising data with much higher time resolution than in previous work.

The work begins with Fourier analysis of the OLR data to identify the most significant modes of variability within the OLR in short time scales, and observe such modes in spatial terms. These modes (semi-diurnal, diurnal, 2-3 days, 4-7 days) are then independently tested using various methods, to attribute them to the appropriate climate processes.

The diurnal cycle has been decomposed using Empirical Orthogonal Function (EOF) technique, and we have managed to separate various diurnal components according to their variance contributions in the dataset. We have found that various conditions can alter the variance contribution, the magnitude and the time phase of the diurnal cycle and its components.

We have attempted to investigate the diurnal variability over different scenes (surface types and clear/cloudy sky) using Principal Component Analysis(PCA). A time delay (2 hours) has been found between the peaks of the components over different surface types.

We have developed a simple radiation model using incident solar radiation (ISR) from GERB, surface properties and a single layer atmosphere. The results were analysed using PCA and we have found a delay of 1 hour in our model. We have investigated the possible causes for the discrepancies between the model components and the OLR data components. We have then conducted regional tests to attempt to test the possible causes.