Currently, researchers at Sandia National Laboratories are creating software that is designed to determine the source of a toxic release given sensor readings of the toxin concentration at fixed locations in the building. One of the most important concerns in solving such problems is computation time since even a crude approximation to the source, if found in a timely manner, will give emergency personnel the chance to take appropriate actions to contain the substance. The manner in which the toxin spreads depends on the air flow within the building. Due to the turbulence in the air flow, it is necessary to calculate the flow field on a fine mesh. Unfortunately, using a fine mesh for every calculation in this problem may result in prohibitively long computation times when other features are incorporated into the model. The goal of this thesis is to reduce the computation time required by the software mentioned above by applying two different mesh coarsening strategies after the flow field is computed. The first of these strategies is to use a uniformly coarse mesh and the second is to use our knowledge of the air flow in the building to construct an adaptive mesh. The objective of the latter strategy is to use a fine mesh only in areas where it is absolutely necessary, i.e., in areas where there is a great change in the flow field.

The thesis deals with the creation of a three-dimensional virtual view with the use of the new WebGL standard in combination with O3D javascript framework. The result is freely available web application which generates a virtual tour of sent series of photographs. Afterwards, the photographs are processed by the server, which on the basis of found corresponding points creates 3D scene. The application is based on a freeware application Bundler [40] which is used for identifying 3D points in uploaded photographs and related cameras (the places from which the photographs were taken). Bundler outputs are processed by Jetty web server with the use of Java script. At this point, all data required for the creation of a view are gathered. The display of the final result is then realized via HTML5, CSS, WebGL and Javascript using O3D.