Jupyter Notebook - Landsat-8 Cloud-Filtered Median Mosaic
- Creators
- Killough, Brian
- Others:
- Open Data Cube Community
- CEOS SEO/AMA
Description
About
This notebook uses global Landsat-8 data from Google Earth Engine which has been indexed to work with the Open Data Cube. The selected data is used to create a custom Landsat-8 cloud-filtered median mosaic for any time period and location. The mosaic can be output as a GeoTIFF product for analysis in external GIS tools. The median mosaic is quite common and reflects the "median" value of cloud-free pixels for all spectral bands in the time series.
Users should review the "Cloud Statistics" notebook for more information about the temporal and spatial samples that are used for any median mosaic. An understanding of the underlying data is important for creating a valid mosaic for further analyses. In many cases, cloud contamination will create poor or time-biased mosaics unless there are sufficient cloud-free pixels in the time series. With a careful review of the cloud coverage over a given region and time period, it is possible to improve the mosaics and avoid false outputs.
Most mosaics are created for annual or seasonal time periods. In cloudy regions, there may not be any clear (non-cloudy) pixels for the defined time window. If this is the case, those pixels will be masked from the final median mosaic product and be shown as "white" regions in the output image.
This baseline notebook runs in about 5 minutes. The default region (0.15 degrees square) and time window (one year) use about 20% of the allocated RAM memory. Selecting larger regions and time windows should be done carefully to avoid exceeding the system limits or having long run times. For example, regions near 0.5-degree square will "crash" the code and not run to completion.
Open Data Cube sandbox
You can execute this notebook in the Open Data Cube Sandbox. You can also run on your software promise environment.
For more information about the Open Data Cube Sandbox, please, check this link.