Complex InSAR Coherence Processing

The production of the first global coherence and backscatter dataset from one year of Sentinel-1 Synthetic Aperture Radar data with an input volume of approximately 1 Petabyte involved the strategic synergy of Gamma Remote Sensing and Earth Big Data software. 

The complex process, culminated in the creation of approximately 7 million cloud optimized GeoTIFF (COG) data organized into 25,000 1x1 degree tiles and was achieved with remarkable efficiency and speed, and as such at a very affordable cost.

The software suites were combined for calibration, burst preprocessing, and transformation of SAR data into coherence and backscatter products. Processing needed to be performed on a path-by-path basis. Robust image matching to reference bursts ensured the accuracy and reliability of the processed data and was the foundation for the subsequent steps in coherence decay modeling. The path processing was staged in 178 job queues. Only after the initial SLC import and burst extraction was completed with data processed for an entire path, did coherence estimation for all image pairs with 6-, 12-, 18-, 24-, 36-, 48-days repeat intervals commence. The complex workflow was staged within Earth Big Data's software which allowed for optimized matching of compute resources and costs for each processing step (see diagram below).