Related research and discussion
Displays of 2D flat data using Google Earth are widely used for geospatially related data from satellite, aerial and ground stations. The best example of displaying 2D geospatial data is Google Earth itself. GES DISC is one of a few pioneer organizations using online geospatial data access and processing services, such as OGC WMS, WCS, and G3, to display 2D meteorological data and their analysis results in Google Earth, enhancing the value of geospatial data. In particular, OPeNDAP and GrADS protocols are used to ubiquitously provide online geospatial data to analysis and processing services, which further process data in the manner requested by the user.
In addition to the solution to visualizing vertical geospatial data in Google Earth described above, there are two other solutions for rendering a 3D orbit curtain in Google Earth: The first one is directly to process geospatial data values to produce a KML file that creates a flat curtain in Google Earth. The curtain is made from many vertical small rectangles over the surface of the Google Earth. At the highest resolution, the length of each rectangle represents the distance the satellite travels in 5 seconds. The problem with this method is that the rendering speed is very slow if the curtain resolution is as good as the one from our proposed solution, but if the rendering speed is faster, the curtain resolution and quality are not good enough for scientific research and applications. So, our approach is fast, of high resolution and high quality. The second solution is to first design the orbit model, then, project geospatial data onto the model using SketchUp’s “texture” feature. This method is used for creating the orbit of planet Saturn. The orbit of Saturn completely covers the virtual globe from Google Earth. When zooming in, the orbit of Saturn is not visible on the high-resolution surface of Google Earth. It is not suitable for displaying high-resolution geospatial data over Google Earth surface through this kind of orbit model. Moreover, the orbit of Saturn uses one generalized image stripe repeatedly as the texture of the Saturn 3D orbit model (Barnabu 2006). However, the curtain images from NASA satellites vary greatly. So, this solution is not suitable for visualizing vertical geospatial data in Google Earth.
Geens (2006) first suggested displaying vertical data in Google Earth. But the description about how to display the vertical data is rough, and the position where the image is placed in Google Earth is not accurate. No systematic and scientific procedures are related to the calculation of the position, scale, and rotation degree of a 3D data model along with the satellite orbit in Google Earth. Yamagishi et al. (2006) provided a tool to convert a seismic tomography model into KML files. The KML file used the first method discussed above. The latitude, longitude, and altitude data are provided using a seismic model for rendering the flat rectangle to represent the geospatial data in Google Earth.
Our solution makes image products at GES DISC available beyond the scientific and research communities. Thus, the virtual globe Google Earth can be used with these products for national public applications with societal benefits. Virtual globes are becoming the next generation framework for sharing data, information and knowledge, collaborating in scientific research, and visualizing education in many disciplines.
It is worthy mentioning that KML 2.2 (Beta) directly supports vertical images in Google Earth via a new tag named ‘PhotoOverlay’ (Google 2007a). It can directly and vertically display images on the surface of the Google Earth. We tried to use it to render the vertical data in Google Earth. However, the following issues were encountered:
Small-scale photographs are acceptably rendered. However, when large-scale images are displayed, one end of the image is on land, but the other end rises into the sky.
It is not easy to control the scale and rotation degree relative to satellite orbit.
Only one side of the image is viewable, no matter how you rotate the globe. But, our current Collada model can display both sides of the vertical image data.