topic Re: Why is resolution so bad on 30 m satellite images? in Data Management Questions

Why is resolution so bad on 30 m satellite images?

grahamlangley-edwards — Thu, 08 Mar 2012 14:11:16 GMT

Hello, I have some TM 4-5 satellite images that I downloaded from the GLOVIS website. The resolution is supposed to be 30m, but the images becomes very blurry and pixelated as soon as I zoom in a little. I have built pyramids (level 4, Bilinear) and calculated statistics. I tried resampling the cellsize, but that didn't work. Under Properties > Source the cellsize is listed as 1,1. I am trying to georeference these images, and need to zoom in to identify landmarks. Any help or suggestions are greatly appreciated.

Thanks,

Graham

Re: Why is resolution so bad on 30 m satellite images?

JeffreySwain — Thu, 08 Mar 2012 15:21:29 GMT

Any 30 Meter raster imagery would for the most part be considered very pixelated. Satellite imagery like LandSat comes in 30 x 30 meter pixels. Depending on the scale that the imagery is viewed at, it could be very pixelated. The pyramid generation and changing the display method on the layer properties may alter the look of the raster, but at the core, 30 x 30 would be coarse. I would not consider trying to georference such imagery unless you have some identifiable object that occupies 900 square meters (30 * 30) or the size of one pixel. The imagery you are using would be best used in landcover analysis where you could identify the pixels based on finer resolution imagery.

Consider the following link and open the section on pixels. When you see the rasters presented there, it will explain the expectations.

[ATTACH=CONFIG]12530[/ATTACH]

Re: Why is resolution so bad on 30 m satellite images?

grahamlangley-edwards — Thu, 08 Mar 2012 17:05:34 GMT

Hi Jeff, thank you for your reply. I have indeed been tasked with using these images for a landcover analysis, but it seems that I need to georeference them before I can really do anything. For example, I need to mosaic the images together, which I cannot do until the images are georeferenced. Is there any way I can improve the resolution a little bit before attempting to georeference? Additionally, I have to orthorectify these images. However, the images do not come with Rational Polynomial Coefficients, and from what I understand I must have RPC values in order to orthorectify. Is that correct? Or is there a way to orthorectify without using them, or perhaps to calculate them? Again, thanks for your help!

Regards,

Graham

Re: Why is resolution so bad on 30 m satellite images?

MelitaKennedy — Thu, 08 Mar 2012 18:51:20 GMT

I saw a USGS website that stated that the GloVis imagery was all orthorectified.

http://www.usgs.gov/newsroom/article.asp?ID=1425

Are you sure that these images aren't? Can you find other images for the same area that are already orthorectified?

Melita

Re: Why is resolution so bad on 30 m satellite images?

JeffreySwain — Fri, 09 Mar 2012 12:48:50 GMT

You will need the RPCs to orthorectify, but I echo what Melita said and that it should be projected already. I would be sure that on the way to you that they have not been altered.