Remote sensing consists in taking images of objects from a distance. We are interested in the remote sensing of Earth from space. Remote sensors capture the incoming light from the Earth’s surface and divide it into color components like a prism would do. Traditional remote sensing divides the light into a few colors or wavelengths (about ten or less). The image from a scene (area of interest on Earth) then consists of pixels, each pixel being made of less than ten colors in various amounts or intensity. The total amount of information about the scene then consists of the number of pixels multiplied by the number of colors. The nature of the land cover can be recognized by taking ratios of color intensities and concrete, grass, trees and water can be identified that way, but the information is limited due to the limited number of colors used. In order to significantly increase the amount of information about land cover, we propose to use many more colors and do so-called hyper spectral remote sensing. Instead of less than ten colors we propose to use hundreds of colors.
Hyper spectral remote sensing is not new but has always been plagued by issues of large data volume transmission. Image (a) above shows a hypercube of data. Image (b) shows a spectrum obtained for one image pixel from many wavelength data. Horizontally, the x and y axes represent positions of pixels on the ground, vertically are the hundreds of color intensities available for each pixel. A typical hypercube may represent tens of megabytes to a gigabyte of data. If the sensor is to capture many scenes as it orbits Earth and transmit them to the ground when it flies above a ground station, it needs to be able to provide enormous amounts of information in a short time. Recently, progress has been made in compressive sensing. Rather than taking a large amount of data, then compressing them and finally beaming them to the ground, it has been shown that it is possible to collect a limited set of data both spatially and spectrally and reconstruct an image that is close to reality. This will limit the amount of information to be handled on the satellite and to be transmitted to the ground.
The HYSPIC project at Auburn University plans on placing a hyperspectral sensor on board a CubeSat using commercially available components. The HYSPIC sensor should have a ground resolution of thirty meters.
There are many engineering challenges for the avionics of a CubeSat with a HYSPIC payload. High accuracy pointing, GPS capability, and a high data rate transmitter will be required. However, compressive data collection will help in keeping the data rate manageable.