Scientific visualization is used to graphically represent scientific data in order to provide a better understanding of crucial scientific results. It is widely used in different disciplines, including engineering, medicine, chemistry, education, economics, geographical systems. Visualization is often employed in scientific portals, which are used to analyze and monitor data produced by jobs run through the portal. This visualization can be used for the real-time dynamic monitoring of computations in progress, and can also provide better insight into individual scientific jobs or post processing procedures. In this thesis, we present our approach to visualization in scientific portals built using the Cyberinfrastructure Web Application Framework (CyberWeb) and its services. The objective of this work is to develop dynamic visualization functionality in the CyberWeb which will allow clients to monitor jobs both during the job and after its completion. The Cyberweb Visualization service (CWViz) enables users to analyze intermittent results of jobs running on various remote hosts. CWViz provides a clear understanding of each job's progress for the user by creating real-time and post-processing visualization plots. CWViz includes features such as the ability to create a variety of plot types, like 2D linear, gradient, and 3D surface or mesh plots; animation and movies; a dynamic web-based portal client; and a command line interface. The system uses the gnuplot library and gnuplot.py package to create the various plots. It also enables users to customize these plots by changing the different plot parameters via a web interface or client API. These plots are created dynamically, and all requests sent by the user are handled in run-time, with proper error handling. This visualization service tool in CyberWeb gives the user a very clear cut understanding of how jobs run, as well as their progression; thereby helping to control them more efficiently.