Remote sensing of diffuse attenuation coefficient patterns from Landsat 8 OLI imagery of turbid inland waters: A case study of Dongting Lake

摘要

The diffuse attenuation coefficient, K_d(λ), is an important optical property. Environmental change and anthropogenic activities, however, have made it challenging to accurately assess K_d(λ) patterns in the extremely turbid inland waters. We addressed this challenge by using new Landsat 8 Operational Land Imager (OU) imagery. For the bio-optical complexity of water, we proposed an empirical band-ratio algorithm for estimating K_d(490) using our in situ measurements. Based on the acceptable performance of an OLI image-based atmospheric correction and K_d(490) validation, the algorithm was then applied to OLI images to estimate K_d(490) patterns from April 2013 to April 2016, leading to several key findings: (1) Spatial-temporal patterns of K_d(490) varied significantly in Dongting Lake. The temporal heterogeneity of K_d(490) could be explained primarily by surface-runoff changes driven by regional precipitation. The spatial heterogeneity was due to sediment resuspension, resulting from sand dredging and shipping activities; (2) K_d(490) values that were inversed at the intersection of Dongting Lake and Yangtze River were observed for the first time near the Chengliji site and resulted from the opposing temporal cycle of K_d(490) variations between Dongting Lake and the Yangtze River; (3) There was a significant positive correlation between K_d(490) and total suspended matter (TSM). This confirms that TSM often plays a principal role in the attenuation of light in extremely turbid water bodies; (4) The empirical band-ratio algorithm worked well, not only for the broader Landsat archives, but also for the narrower Sentinel-2/3 for K_d(490) estimation, which demonstrates that the algorithm could be used to quantitatively monitor multi-decade records of Landsat observations and future applications of inland water quality in turbid inland waters, such as Dongting Lake and Poyang Lake.