• Mashup Visualization with Google Earth and GIS
  • 1.1. Introduction to KML and Google Earth
  • 1.2. GIS and Google Earth
    • 1.2.1. Raster Surfaces and the LandSerf GIS
    • 1.2.2. Surface Analysis
    • 1.2.3. GeoVisualization Mashup
  • 1.3. Glossary
  • 1.4. Bibliography

1.2. GIS and Google Earth

In this session we will perform more sophisticated analysis of our data to address some of the issues raised in the previous session. Remember, the goal of the geovisualization is to gain some insight into two large spatial data sets and the relationships between them, namely :

• a large database of photographs with time stamps and point-based georeferencing
• information about the population recorded by the US Census Bureau

We will consider some alternative models of these data that will attempt to overcome some of the limitations of the previous approaches and reveal further structure in the data.

Limitations of the vector modelling of our photographic and population data using points and areas include :

• the size of the full photographic dataset (over 1.7M entries) slows Google Earth and can create cluttered visual output that is difficult to visualize. (Note : we only considered a subset of the most interesting photographs at any one time in the first session of this tutorial for these reasons.)
• the distribution of photographic sources is greatly influenced by the distribution of people (we cannot easily distinguish the two distributions)
• the population is summarised over discrete spatial units (e.g. states, counties, tracts). These artificial units result in arbitrary changes at boundaries and other artifacts resulting from the method of data modelling rather than the phenomenon being modelled
• opportunities for analysis of point (photographs) and area (population) data are limited.

We attempted to overcome some of these limitations by sampling the dataset (by space, time and theme) and using appropriate cartography for its visual representation. However, we can go further by using an alternative representation of the data that is more suited to modelling large and continuous datasets. This transformation can be undertaken in a Geographic Information System. We can use KML to incorporate the results into Google Earth for visual comparison with our vector data.

Learning Outcomes

On completing this session you will be able to do the following :

• derive and visualize continuous surfaces in a GIS
• use Chi statistic surfaces to compare pairs of datasets
• use KML to incorporate output from a GIS into Google Earth
• use visualization to compare and contrast representations of geographic data

Update: 14th October 2008 (eLML) - Contact - Print (PDF) - ©