Search the Community

Hi all, want your help. I am working on a project where I would like to have a seamless natural color imagery of entire globe at 15m resolution. I do not have the expertise to convert individual Landsat scenes to a proper color balanced mosaic. So, as an alternative I am trying to convert the already available Landsat Geocover global seamless imagery from NASA. This dataset has been created from Landsat EMT+ imagery from circa 2000 to 2003. Below is a description this product: http://www.glcf.umd.edu/library/guide/techguide_geocover.pdf The problem is this data-set uses band combination of 742, instead of normal natural color band combination of 432. This was done to minimize the haze. But 742 band combination results in purple looking cities and a florescent green. I would like to achive a proper natural color as has been used in various mapping portals such as Bing maps. When you see imagery in Bing maps at country level you are basically looking at imagery converted from Landsat Geocover imagery using band mathematics only. This document from Earthsat describes this process, but they have not given much technical detail on how they do it:http://www.earthsat.com/HTML/naturalvue/naturalvue_packet_2.pdf

I'm working on a project right now and I want to make sure that I get the radiometric corrections perfect and I've heard that thermal bands are different from 1-4. I was told that you can't just use darkest object subtraction or a cos(t) model, but that you have to have a different approach altogether. I'm going to be using LandSat8's TIRS (bands 10,11) to try to discern an estimate of surface temperature. The image that I have is relatively free of cloud cover, but I want to make this image as clean as possible so as to not accidentally calculate the Top of Atmosphere temp. Any help would be awesome!