True Color LandSat Images
LandSat 7 images are now pansharpened to 14.25 meter resolution. The images available here have been processed from large, single band tiles into one-degree blocks by the Geo-Archaeological Information Applications (GAIA) Lab at Arizona State University. Imagery from the countries shown below can be downloaded here. Other regions can be made to order--email shsavage@asu.edu for details.
Detailed metadata for the images is available from the USGS Earth Resources Observation and Science Center website.
Download True Color LandSat Images
Select One of These Links and Scroll Down for the Map
- Afghanistan (including parts of surrounding countries)
- Egypt (including parts of Israel, the Palestinian Territories, Libya, and Saudi Arabia)
- Greece (including parts of surrounding countries)
- Iran (including parts of surrounding countries)
- Iraq (including parts of Jordan, Syria, Turkey, Iran, Kuwait and Saudi Arabia)
- Italy (including southern Corsica)
- Libya (including parts of surrounding countries)
- Morocco (including parts of Algeria and Western Sahara)
- Saudi Arabia (including parts of surrounding countries)
- Southern Levant (including Sinai, Israel, Palestine Territories, Jordan, Lebanon, Cyprus, and parts of Syria, Iraq, and Saudi Arabia)
- Syria and Lebanon (including parts of Israel, Jordan, Iraq, and Turkey)
- Tunesia (including parts of surrounding countries)
- Turkey (including parts of surrounding countries)
LandSat 7 Images
NASA has posted imagery derived from Orthorectified LandSat Enhanced Thematic Mapper (ETM+) Compressed Mosaics, at a pixel size of 14.25 meters. The images date from 2000 +/- 3 years instead of 1990 +/- 3 years.
The Jet propulsion Laboratory and NASA have another web site where mosaics of "unprocessed" Orthorectified LandSat Enhanced Thematic Mapper (ETM+) scenes are available as individual bands. This site can be found at http://stp.jpl.nasa.gov/WMS_GM.html, but experience has shown that it's down more often than it's up.
The datasets available there were compiled through NASA's Commercial Remote Sensing Program; this program is a cooperative effort between NASA and the commercial remote sensing community to provide users with access to quality-screened, high-resolution satellite images with global coverage over the Earth's land masses. The resulting Landsat Enhanced Thematic Mapper Plus (ETM+) data set contains orthorectified ETM+ sensor data from the Landsat-7 satellite. Only the ETM+ images most closely meeting a specific set of criteria, including acquisition date, cloud percentage, data quality parameters, and best available phenology, were considered for the collection.
JPL/NASA Data Set Characteristics
Approximately 8,500 LandSat ETM+ images were selected for the NASA/JPL project. These scenes were acquired between 1999 and 2001, and have been orthorectified so as to tightly co-register with earlier circa 1990 global LandSat TM coverage. NOTE: In both the circa 1990 and circa 2000 data sets, the temporal dates of individual scenes may vary between 1-3 years of the 1990/2000 baseline.) These images were either cloud-free or contained minimal cloud cover. In addition, only ETM+ images with a high quality ranking in regards to the possible presence of errors such as missing scans or saturated bands were selected. It was important to collect data during periods of peak greenness so NASA adopted an approach for image selection that was based upon a data set containing global 1-kilometer advanced very high-resolution radiometer Normalized Difference Vegetation Index data. It was sometimes difficult for NASA to obtain clear images during time frames of peak greenness, resulting in some image substitution from other times of the year.
JPL/NASA Processing Procedures
LandSat ETM+ data are orthorectified using existing orthorectified TM data as
horizontal control. Pixel cross-correlation is used to pass the control
from the earlier orthorectified TM data to the LandSat ETM+. A patented LandSat
large block triangulation procedure is used to perform a photogrammetric
adjustment of each individual LandSat ETM+ scene; data are then placed into a
UTM / WGS84 map projection. All bands are individually re-sampled, using a
nearest neighbor algorithm. The result is a final image map product with a
Root Mean Square Error (RMSE) of better than 50 meters in positional accuracy.
 |
 |
| False color image using Bands 7, 4 and 2 (MrSid server). | "True color" produced by combining Bands 1, 2 and 3. |
 |
 |
| Adjusted color levels in PhotoShop. | Pan sharpened and color balanced in PhotoShop. |
Processing Done at the GAIA Lab
The first image shown above is from the NASA MrSid image server, LandSat 5, delivered in the normal false color scheme. These images used the following spectral bands:
- Band 7 (mid-infrared light) is displayed as red.
- Band 4 (near-infrared light) is displayed as green.
- Band 2 (visible green light) is displayed as blue.
An approximation of "true color" can be created by combining the first three bands in either LandSat 5 or 7. The upper right image shows the result from a portion of a LandSat 7 tile; Band 1 is the blue part of the spectrum, band 2 is green, and band 3 is red. The upper right image was processed from these raw LandSat bands.
Color Balancing the Images
Simply combining the first three bands doesn't really produce true color. The images come out looking more like Mars than Earth, as the upper right picture shown above illustrates. Further processing is required for better appearance. The "Auto Level" command in Adobe PhotoShop (or the "Level" command with a saved .alv file) can be run against the image, producing results similar to the lower left image shown above. It's a lot of work to produce true color images, and running PhotoShop on three bands, where each band is 812 mb is clearly out of the question. Since the images have to be cut into smaller tiles anyway, we cut them up first, then did the color correction.
No single color filter, whether it is the Auto level or a custom color filter, works for every tile--there is no "one size fits all." To create the color adjusted .jpg images that are included, The GAIA Lab applied a variety of color adjustments, depending on the tile. If you have PhotoShop or similar image processing software you can make further adjustments on the color balanced .jpg files, or start from scratch by loading the .tiff file, then working from it.
Pansharpening the Images
The images created by blending the first three bands from the LandSat scenes are at 28.5 meter resolution. They have been pansharpened to 14.25 meter resolution by overlaying the panchromatic band on top of the three color bands and assigning the color value based on the three color bands, and the luminosity factor based on the value from the panchromatic band. Since the panchromatic band is recorded at 14.25 meter resolution, the result is a "true color" pansharpened image. This has the effect of making the image files a lot larger--especially the .jpg files, which don't compress very much in the zipping process. So the tiles you download will be larger than the ones that included non-pansharpened images. The lower right image shown above is an example of the final product.
Further Adjustments
Since the images are derived many "granules" (satellite photograph frames) taken at different times, there are significant differences in the amount of haze in different parts of the original tiles, especially over water. I used the color fill function in Photoshop to create a uniform dark blue for oceans and seas.