Which images will work best to fill in the gaps in Landsat 7 ETM+ SLC-off data?
Landsat 7 SLC-off images can be filled by using other Landsat 7 scenes, using the information below.
Both primary and fill scenes should be cloud free. If cloud cover is unavoidable, please keep in mind the following considerations:
Cloud cover in the primary scene will create an extension of the "clouds" into the scan gap region and will also generate local artifacts near the edges of these features.
Cloud cover in the fill scene will create "artificial terrain" values in the scan gap region in an attempt to match the data values to those of the primary scene. Characteristics of the cloud cover may therefore create details and artificial features that do not exist on the ground.
Cloud cover may have an adverse impact on scene co-registration. The amount of impact will depend upon the amount, location, and distribution of the cloud cover that is present. Even if the particular area of interest is not cloud-covered, a larger cloud-free region surrounding the area of interest is still needed in order for the local co-registration to be accurate. If the cloud-free region around the area of interest is large enough to contain several ground control points (GCPs), the results will probably be satisfactory. The largest inaccuracies will be localized in the area(s) that have the least number of points registered.
The fill scene should be as close as possible to the anniversary date of the primary SLC-off scene or at least from the same season. Differences in seasonality can introduce a lineated appearance within the fill scans (vs. primary data scans) due to differences in sun angle. This becomes particularly apparent over relatively homogeneous areas.
There should be minimal differences in transient conditions between the primary and fill scenes (i.e. snow cover, flooding, wildfires, etc.).
The area of interest should contain low temporal variability with respect to land cover and surface features.
If you are interested in agriculture, you may want to consider choosing a fill scene from more than one year prior to the date of the primary (SLC-off) scene, since many crops a are rotated in multi-year cycles.
Landsat represents the world's longest continuously acquired collection of space-based moderate-resolution land remote sensing data. Four decades of imagery provides a unique resource for those who work in agriculture, geology, forestry, regional planning, education, mapping, and global change research. Landsat images are also invaluable for emergency response and disaster relief.
The Landsat Update is an informal communication tool, prepared periodically and distributed electronically to USGS Landsat partners, to provide information about Landsat activities and related topics of interest.