A Consistent Radar Altimetry Dataset for Major World Rivers: Extraction Methods and Preliminary Data Products

Abstract:

Our group has made several efforts to develop the systematics for processing multiple satellite mission inland altimetry data with the purpose of creating a pre-SWOT climate data record of world’s rivers greater than 900m in width. The project is a component of a NASA MEaSUREs (Making Earth System Data Records for Use in Research Environments) project undertaken by UCLA, Princeton U., NASA/GSFC and Ohio State Univ. The first method developed allows for the identification of measurements that represent the target river through height filtering and is based on USGS flow data from 105 gauges on rivers with watersheds over 20,000 km². Proximal topographic variations led to some contamination of the radar returns. We were able to identify them using the previously mentioned height filter, and correlated their frequency with near-river topographic indices. Significant efforts have also been made to detect river ice using only radar backscatter. Over 631 Landsat images were processed and given an ice cover designation then compared with measured backscatter profiles; demonstrating that isolating a one- to-one relationship between ice and backscatter will be challenging. An additional focus of the group has been automation of detecting altimeter/river intersections as well as the creation of “virtual stations” or masks for data extraction at those locations. Using RivWidth parameters to generate polygons and a raster proximity based intersection detection methods have both shown promising results for automation of this process. This project will soon be producing validated climate data records in the form of geocentric river height changes, both in terms of scale of the study area and access to previously unmonitored regions. Once established, these methods will also be applicable to the study of future satellite cycles. Preliminary river height change data products have been produced for the Mississippi, St Lawrence, Yukon, Mackenzie, and part of the Ganges-Brahmaputra-Meghan Rivers. We have found that 43.9% of virtual stations in North America produce viable altimetry river height time series with at least 50% of cycles represented.