|Paper title||Automated processing of altimetry-derived river water levels at global scale - Design & first results from a new L3 processor|
|Form of presentation||Poster|
In the frame of the ESA HYDROCOASTAL project, led by Satoc Ltd, a Test Dataset (TDS) is being developed by partners starting from Altimetry L1A data products (Sentinel-3, CryoSat-2) up to L2 covering the coastal and inland water domains. Then, specific products are targeted to the monitoring of inland water: L3 for the river and lake water level estimations and L4 for the estimation of river discharge.
The TDS is a test-benchmark run focusing on selected region of interest. It serves to perform extensive validation activities in the objective to qualify and quantify the quality of the various output products (L2, L3 and L4). Outcomes of this activity will serve to validate the methods and algorithms to be adopted by the team before the project initiates the production of Globally Validated Products (GVP).
This presentation focuses on the results obtained at L3 for the estimation of river water level.
Partners involved in the production of the L2 products all have developed and implemented specific waveform retracking algorithms. Outputs from each of these retrackers, limited to those implemented over the inland water domain, are processed further on to produce L3 river water level estimates.
Results will describe the performance of each of the L2 retrackers, from the L3 "point of view". The analysis will not only focus on the vertical accuracy of the data but also on the ability of the complete L1-to-L3 chain to produce consistent water level time series, considering the effective temporal sampling as another key indicator of the data quality.
The validation activity is done over the Amazon basin and involves the systematic comparison to in situ data from the ANA (Agência Nacional de Águas, Brazil).
Eventually, the results will highlight the strength and possible weaknesses of the retracking algorithms, helping to decide which retracker is eligible to be implemented in the production of the ultimate GVP datasets.