Satellite data have fundamentally changed how we perceive and understand the world we live in. The amount of data produced is rapidly increasing and classical computing resources and processing tools are no longer sufficient. These massive amounts of data require huge storage in addition to advanced computing capacity in order to allow users to benefit from the derived datasets. Cloud computing has provided the required storage and computing capacity on a scalable level to add or remove resources according to requirements. Simultaneously, recent advances in data processing techniques such as Deep Learning (DL) have paved the way to integrated solutions for Earth Observation (EO) big data understanding. By bringing together a unique combination of computing capacity, ultra fast data storage and advanced data processing techniques, our ability to derive useful insights will be revolutionised. This thesis focuses on harnessing cloud computing and deep learning capabilities to enhance spatial resolution of satellite imagery data. Particularly, the thesis highlights cloud computing architectures to accommodate satellite image processing services. A conceptual data model was developed to enable utilising cloud computing resources with EO big data in near real-time. Moreover, a state-of-the-art super resolution algorithm (SRCNN) was adapted and tested in detail to be exploited with the satellite image domain. In addition, a novel fusion-based data augmentation approach was developed to boost super resolution accuracy. To evaluate the super resolution accuracy with a real-life application, landcover classification was adopted to assess the accuracy between super resolved Landsat-8 data and crowd-source data collected using the Google Earth interface. The accuracy achieved opens a wide field of research with deep learning and data augmentation in the satellite image super resolution domain.