CoAuthors

Steve Tuozzolo (Ohio State University School of Earth Sciences, United States); Michael Durand (Ohio State University Main Campus, United States); Tamlin Pavelsky (University of North Carolina Department of Geological Sciences, United States); George Allen (University of North Carolina Department of Geological Sciences, United States); Stephane Calmant (IRD/LEGOS, France); Yuchan Yi (Ohio State University School of Earth Sciences, United States); Yuanyuan Jia (Ohio State University School of Earth Sciences, United States); Qi Guo (Ohio State University School of Earth Sciences, United States); C.K. Shum (Ohio State University School of Earth Sciences, United States)

Here we present a new radar altimetry dataset GRRATS (Global River Radar Altimetry Time Series) extracted over global ocean-draining rivers wider than 900 m. The dataset includes 909 time series from 39 rivers. A new method of filtering VS height time series is presented where, DEM based heights were used to establish limits for ice1 retracked Jason2 and Envisat heights. While GRRATS is following in the footsteps of several predecessors, it has something to offer to the decadal climatological record. It features the most comprehensive set of RA time series currently available for North America, as well as Eurasia. Consistent methodologies for flagging ice cover are presented and employed throughout. DEM heights used in height filtering were retained and can be used as river height profiles. All VS have been assigned a letter grade A-F to aid end users in selection of data that will be most helpful. Grades a were based on fit statistics when available and qualitative assessment otherwise. Due to the inclusiveness of the dataset, not all VS were able to undergo validation (415 of 909), but those that were demonstrate that confidence in the data product is warranted. Validation was accomplished using records from 45 in situ gauges from 12 rivers. Meta-analysis was performed to compare each gauge with each VS by relative height. 69.7% had positive Nash Sutcliff Efficiency (NES) values, and the median NSE value was 0.69. The mean standard deviation of error (STDE) was 1.28 m, which is promising, as Envisat and Jason-2 are accurate to around .28 m.