Principal Investigator: Fritz Policelli, NASA's Goddard Space Flight Center
Streamflow is a critical measurement for closing the water balance of catchment areas and basins, a component of one of the “Most Important” objectives in the most recent National Academies Earth Science Decadal Survey. Yet streamflow measurements from in-situ gauges are not available in much of the world.
While there are a number of algorithms designed for automated retrieval of surface water extent from synthetic aperture radar (SAR) data and algorithms designed for automated retrieval of stream width from surface water extent maps, there are currently no operational systems readily available to the global community of researchers and applications users to obtain SAR-based surface water maps and corresponding stream width information for a region and time period of interest.
This partnership between NASA, NASA's Alaska Satellite Facility Distributed Active Archive Center (ASF DAAC), and the German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt e.V.) will implement such a system to create training data for machine learning using the European Space Agency's (ESA) Sentinel-1 C-band SAR data from ASF DAAC's growing cloud-based SAR data archive and new Sentinel-1 data as it is received. The project will implement, test, and operationalize a system to derive effective stream width data using data from ESA's Sentinel-1 C-band radar satellite constellation, archive the data produced, and distribute the data for free and open use to train machine learning models relating to stream flow and effective stream width.
Researchers armed with these models for reaches of interest will be able to estimate stream flow using effective stream width data derived from sources including Sentinel- 1, Sentinel-2, and Landsat satellites. These stream flow estimates will be highly complementary to stream flow estimates provided by NASA's Surface Water and Ocean Topography (SWOT) Mission (planned for launch in September 2021), which will provide streamflow estimates for a global set of predefined reaches.
However, while the SWOT coverage is limited to reaches with a width greater than 100 meters and the revisit time is only twice in 21 days, these data could potentially be used in models to provide stream flow estimates at a frequency corresponding to the revisit time of the Sentinel-1 constellation of six days or less at a spatial resolution of approximately 10 m.
The team will also produce, archive, and distribute an intermediate product, surface water extent maps, for free and open use. These data can be used by researchers to derive their own effective stream width measurements for training machine learning models, or for a number of other important research and applications purposes.
The team will evaluate the accuracy of the products and validate the utility of the products for training machine learning algorithms relating effective stream width and streamflow for a representative sample of data. Synthetic surface water maps and effective stream width data will be developed and evaluated.
The operational system will be implemented at ASF DAAC, which is funded by NASA to archive and distribute Sentinel-1 data under an agreement between the U.S. State Department and the European Commission. The system will take advantage of ASF DAAC's existing Level-2 processing platforms such as HyP3 (Hybrid Pluggable Processing Pipeline) for operational processing, which will allow the generation of higher-level data products. As HyP3 is a cloud-based platform that is co-located with ASF DAAC's Sentinel-1 archive, the platform will enable easy implementation of new algorithms and exercising of these algorithms at scale.