Understanding Seasonality in Water Clarity in 10,000+ Lakes Across the Contiguous United States Using Remote Sensing

Abstract:

Water clarity regulates many properties of aquatic ecosystems including the depth of primary productivity and thermal stratification, making it a primary characteristic of interest in many water quality research and monitoring programs. While water clarity is often characterized as a relatively static feature, it can vary with the time of year and in response to disturbance events. Understanding seasonality in water clarity through space as well as how seasonality changes through time is critical to understanding how light influences the phenology of aquatic ecosystems. However, the majority of lakes do not have a long record of seasonally-defined water clarity records, and those that do are often biased towards large lakes used for long-term research, water supply, or state monitoring programs. Additionally, there are large spatial monitoring gaps across the United States. Remote sensing offers an alternative to traditional sampling methods to acquire a long-term, unbiased record of water clarity. Here we use a database consisting of 35 years of ~280,000 matchups between field observations and Landsat overpasses to train a machine learning model to estimate water clarity using Landsat imagery. We assess the phenology of water clarity and the degree to which seasonality has changed in >10,000 US lakes from 1984-2020, and find that land cover and land use classes are associated with different patterns in water clarity seasonality. Our results highlight regional differences in water clarity across the United States and demonstrate the importance of understanding water clarity seasonality as a primary feature of aquatic ecosystems.