Construction of the 3D Globe Image

Rudolf Husar, rhusar@mecf.wustl.edu, Feb 24, 1998

The 3D Earth image was synthesized from four data layers: land, fires, aerosol optical thickness over the oceans and the clouds. All the data layers used in the compositing were generated from recently analyzed satellite data and each dataset was found through the Web where it is freely accessible. The brief description below lists the web data sources and the procedures for the image construction.

The Four Data Layers

The land image layer was generated by NASA-GSFC SeaWiFS Project. It is a color composite of the world for (nearly) cloud-free conditions, September 18-October 3, 1997.

Since the sensor has high spectral resolution for re, blue and green, it can act as digital camera. Initially I have tried to use NVDI to show the color and texture of the land surface but it did not look very appealing. So, Michael King suggested check out SeaWiFS. I found the in the large (3600x1800 pixel) image in the Gallery of SeaWiFS Project maintained by Gene Feldman and used it as is – no doctoring. I simply covered the oceans with the aerosol layer.

[Since December 97 I can not get my eyes off the SeaWiFS images of the world - finally we have color Earth photographs daily viewable through the Global SeaWiFS Browse Utility!]

The fire maps were obtained from the European Space Agency (ESA) Fire Atlas for January 1993 over South America and Africa. Each red dot represents a fire location detected by the AVHRR sensor.

I could have used several fire location maps published on the Web but I found this dataset as the most detailed and convenient for use.

 

The oceanic aerosol layer is based on NOAA-NESDIS AVHRR aerosol data product for the winter season, December, January and February of 1990 and 1991. Published in JGR 1997, 16889-16909.

This data layer is our own production and it is rendered as a contour of a 1x1 degree grid. The red values represent aerosol optical thickness over 0.5, the blue values are 0.1 or less.

The cloud layer is a composite of IR images from four geostationary satellites, GOES 8, GOES 9, METEOSAT and GMS 5, for Sep. 17,1997 produced by the U. Wisconsin, SSEC KidSat Project.

The U. Wisconsin routinely produces daily global composite IR images. However, this image is a special high resolution version, that was provided to me by Sanjay S. Limaye.

 

Compositing of the Four Data Layers

The land, fire and aerosol layers were simply overlaid as in a GIS system. Since the aerosol covers only the oceans, the land is visible through the empty holes over the continents. The location of fires are red dots that are drawn on top of the land. The cloud layer is placed on the top, covering the lower three layers.

Rendering the 3 Dimensional Image

The land, aerosol and fire layers are projected onto the surface of the earth sphere. The gray color of each cloud pixel was retained from the original IR image. However, I have removed some low clouds to open up some space for the land and the blue-red ocean.(is that legitimate?)

The 3D appearance of the clouds was done the hard way, (using an algorithm Kari Hoijarvi):

  1. the height of the cloud was taken to be inversely proportional to the IR temperature; (2) the cloud layer was divided into 16 height layers (3) the lowest cloud layer was drawn first at a given height (offset) from the earth’s surface. (4) each subsequent layer was painted on top of the lower layer as if they were layers of an onion. This way, higher clouds are always visible while the lower layers and the surface features my be partially covered. So, the reason the clouds appear jagged is that three are only 16 ‘ onion’ layers. Even this way, each rendering is an overnighter on our Penthium 300.