"adapting" Landsat7 tile view to a location
This is going to sound really ignorant. I miss the vocabulary, but I bet you'll understand when you see the picture attached.
This picture is a small piece of a GPS track laid on top of a geotiff.
The geotiff is a landsat7 tile downloaded from http://glovis.usgs.gov
I'm thrilled to see how they match perfectly (I cropped to show the road tied seamlessly from the GPS to the Satellite).
The world is not flat and I understand why the landsat tile shows only a (+- 15km) central band of solid picture and then starts to form widening strips of black between strips of satellite picture, but I would like to change this to suit my area of interest.
How doi we make an area on the outside edge of the scene as if it was part of the central band (some way to "re-center" the tiff so the "stripes" disappear over my are oaf interest).
I know this sounds terrible.. any help much welcome.
I've never understood exactly how these gaps are formed. But they are real gaps, missing data. There's often a gap mask supplied with the other channels. You can use this mask to hide the black, so that another image can be made to show through. So the short answer is, you need another image (or more than one), with gaps in different places, to supply the missing detail.
I think what you want to do is remove the "collar" from the images, assuming that you have a collection of overlapping images with the black border. See http://www.globalmapper.com/support/faq.htm#12 for instructions on doing this.
Another option is to just use the File->Download Online Imagery menu command to pull in seamless Landsat data for your area of interest.
Let me know if I can be of further assistance.
Global Mapper Support
(Mike I'm not sure you can have looked at the thumbnail posted. These are gaps not a collar.)
They do look like gaps, but it would be possible to have each strip as a single image with black edges, although it is likely just gaps in this case.
Global Mapper Support
Ooops.. I thought it was going to be easy.
The picture I submitted is a cropped screen capture of a slide from a scene (a "tile") i downloaded from http://glovis.usgs.gov.
With their interface, you can select a location a review the scenes available for that place. It downloads in big compressed chunks and once you expand it, you get a collection on slides (each slide records the surface reflection at a given wave length).
All this is pretty bulky, so I don't want to clog you with a full slide, but I attach here another crop that shows how these slides are delivered: towards the left of this view, you can see a band of seamless image (about 15km wide) that runs north-north-east to south-south-west - that's right below the satellite's path. Moving away from that path, we get the widening "gaps".
I have the feeling that if we printed this thing and cut out all the gaps and glued the bands side by side, we would get something to fit big globe at scale. I thought this software (if correctly instructed) would know how to do that and then make things flat again with a seamless view in my area of interest.
Sorry again for my lack of vocabulary.
actually, tjhb is right
sorry, looking closer at it, I think that what I just posted is wrong.
Like tjhb said "they are real gaps, missing data".
The "readme" that came with the file says
* GAP MASKS
All Landsat 7 ETM+ SLC-off imagery processed on or after December 11,
2008, will include gap mask files. (Please note the difference between
acquisition date and processing date.) The gap mask files are bit mask
files showing the locations of the image gaps (areas that fall between
ETM+ scans). One tarred and gzip-compressed gap mask file is provided
for each band in GeoTIFF format.
The readme also has phone numbers for support. I'll post what I learn after I get to speak with someone there.
Filter out SLC-off images in your search at GloVis. (SLC-on images have no gaps.)
Or combine two or more SLC-off images in such a way that they fill the gaps in each other. (Using their gap masks to make the gaps invisible.) Of course doing this you also have to worry about matching colour and tonal range between images.
Landsat 7 suffered a failure years ago in the SLC. More info is here:
Thank you very much Frank2 for providing that explanation. I'm really happy to finally understand why. Makes perfect sense. Not some arcane technical "feature" but a bit of broken kit.