Our postdoctoral fellow, Dr. Ilia Parshakov, has developed an open-source Python script for the automatic selection, preprocessing, and downloading of satellite images of various spatial resolution from Google Earth Engine (GEE) for a user-specified MGRS tile. Downloaded images are suitable for the development and testing of multiscale and multi-sensor time-series image analysis techniques, such as UC‑Change. The preprocessing of 60m Landsat MSS (1972 – 1993), 30m Landsat 4 – 9 (1982 – present), 15m Landsat 7 – 9 panchromatic (1999 – present), and 10m and 20m Sentinel-2 (2015 – present) data includes the following operations:
• Cloud and cloud-shadow masking
• Snow and wildfire smoke masking
• Topographic illumination correction
• Landsat image mosaicking and tiling
• Landsat MSS images are coregistered to a reference Landsat 5 TM image in order to improve the geometric accuracy
The algorithm selects the best available images in terms of cloud cover (and image quality in case of Landsat MSS) to provide a >90% coverage for every year. In addition, the algorithm can download GEDI data (2019 – present). GEDI is a spacebourne LiDAR sensor designed to accurately measure the canopy height and biomass of Earth's forests.
More info and download instruction can be found at Time Series Data Downloader.