Abstract
Satellite altimetry is an important contributor for measuring the water level of continental water bodies. The technique has been applied for almost three decades. In this period the data quality has increased and the applications have evolved from the study of a few large lakes and rivers, to near global applications at various scales. Products from current satellite altimetry missions should be validated to continuously improve the measurements. Sentinel-3A has been operating since 2016 and is the first mission operating in synthetic aperture radar (SAR) mode globally. Here we evaluate its performance in capturing lake level variations based on a physical and an empirical retracker provided in the official level 2 product. The validation is performed for more than 100 lakes in the United States and Canada where the altimetry based water levels are compared with in situ data. As validation measures we consider the root mean squared error, the Pearson correlation, and the percentage of outliers. For the US sites the median of the RMSE value is 25 cm and 19 cm and the median of the Pearson correlations are 0.86 and 0.93 for the physical and empirical retracker, respectively. The percentage of outliers (median) is 11% for both retrackers. The validations measures are slightly poorer for the Canadian sites; the median RMSE is approximately 5 cm larger, the Pearson correlation 0.1 lower, and the percentage of outliers 5% larger. The poorer performance for the Canadian sites is mainly related to the presence of lake ice in the winter period where the surface elevations are not able to map the surface correctly. The validation measures improve considerably when evaluated for summer data only. For both areas we show that the reconstruction of the water level variations based on the empirical retracker is significantly better compared to that of the physical retracker in terms of the RMSE and the Pearson correlation.
Highlights
Satellite altimetry has been used for inland water applications for nearly three decades proving its value in many studies [1,2,3,4,5,6]
The results of the validation in terms of a bias corrected root mean squared error (RMSE), the Pearson correlation, and the number of outliers is performed at 50 virtual stations distributed over 40 lakes in the United States (US) and 172 virtual stations in Canada distributed over 98 lakes are shown in Figures 2 and 3
We have evaluated the performance of Sentinel-3A in capturing lake level variation based on a comparison of more than 100 lakes with in situ data
Summary
Satellite altimetry has been used for inland water applications for nearly three decades proving its value in many studies [1,2,3,4,5,6]. Sentinel-3A has been operating for approximately four years and some studies have already evaluated the data quality for inland water applications. Reference [14] evaluated Sentinel-3A water levels based on both standard and non standard retrackers at 26 virtual stations at Chinese rivers and demonstrated RMSE values between 12 and 90 cm. Reference [21] did several evaluations to quantify the performance of Sentinel-3A and 3B for inland water for example, demonstrated an accuracy of 1.5 cm over lake Issyk-Kyl and showed the consistency of the water surface height between the Sentinel-3A and 3B over Lake Lagoda in Russia. The validation is performed against in situ data from more than 100 lakes located in the United States (US) and Canada with an area between 50 and 7000 km2 In this investigation we focus on the overall trend among the considered lakes to provide a general performance estimate for Sentinel-3A. For selected lakes we investigate the reasons for apparent large deviations between the Sentinel-3A and in situ water levels and discuss the importance and need for alternative quality control when in situ data is not available
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
