Evaluation and Analysis of Uncertainty on Lake Elevation Measurement Performance of GEDI in the U.S. and Canada

GEDI (Global Ecosystem Dynamics Investigation) data was analyzed by orbit and by each lake to understand waveform characteristics and define filtering criteria for improved water surface elevation measurement. The L2A data were filtered using guidance provided in the L2A Algorithm Theoretical Basis Document (ATBD). Further adjustment and filtering were applied to identify outliers and improve the accuracy of GEDI elevation measurement. The daily, weekly, and monthly time series of lake elevation were compared at each lake for each of the eight GEDI beams. For seven lakes in North America, GEDI elevation estimates exhibited an overall good agreement with in-situ water levels from local gauges with a mean elevation bias of 0.23 m and Pearson correlation coefficient of 0.70. The comparison of GEDI elevations to those from the HYDROWEB database had a mean elevation bias of 0.12 m and correlation coefficient of 0.61. Over the seven lakes, the bias between GEDI elevations and in-situ data ranged from -0.10 m to +0.53 m with a correlation coefficient ranging from 0.52 to 0.84. The bias between GEDI elevations and satellite-based data ranged from -0.31 m to +0.37 m with a correlation coefficient ranging from 0.38 to 0.82. This work emphasizes the feasibility of GEDI data for accurately calculating lake water levels and has the potential to further our knowledge of lakes' hydrological significance, particularly in the data limited area in the world.