Estimating water levels and volumes of lakes dated back to the 1980s using Landsat imagery and photon-counting lidar datasets

Currently, in-situ data and the time duration of altimeter data are limitations in calculating the water level and water volume of lakes and reservoirs from remotely sensed data. A novel method is proposed to estimate the temporal change in water levels and water volumes for lakes with only remotely sensed data. First, the surface profiles, including the ground and the underwater bottom, were extracted from the MABEL (Multiple Altimeter Beam Experimental Lidar) photon-counting lidar raw data via a new surface detecting algorithm. Second, the lake boundaries between land and water in different years were identified using a thresholding method based on the annual median Landsat composite. Third, water levels were calculated by matching the lidar surface profiles with the lake boundaries based on the nearby georeferenced coordinates. Finally, the water volumes in different years were estimated via the contours (i.e., lake boundaries) with different elevations. Lake Mead was selected as the study area, which is the largest reservoir in the United States in terms of water capacity. With only one day measuring lidar points in February 2012 and over 20 years of Landsat images (from 1987 to 2007), the water levels and water volumes in different years were estimated and compared with the in-situ data. Our results performed well in accordance with the in-situ measurements; the R-square of the water levels and water volumes were both over 0.99; the RMSE of the interannual variations of water levels and water volumes were 0.96 m and 0.31 km(3), respectively. The MABEL was used as a technology demonstrator for the satellite photon-counting laser altimeter and had similar data to the ICESat-2 dataset. Future ICESat-2 datasets will broaden this method to estimate water volumes for remote lakes from the 1980s, where no in-situ data are available (such as the Tibetan Plateau and polar regions with thousands of remote and wild lakes), which could not be achieved in previous studies.