Combining lake and watershed characteristics with Landsat TM data for remote estimation of regional lake clarity

Abstract

Water clarity is a reliable indicator of lake productivity and an ideal metric of regional water quality. Clarity is an indicator of other water quality variables including chlorophyll-a, total phosphorus and trophic status; however, unlike these metrics, clarity can be accurately and efficiently estimated remotely on a regional scale. Remote sensing is useful in regions containing a large number of lakes that are cost prohibitive to monitor regularly using traditional field methods. Field-assessed lakes generally are easily accessible and may represent a spatially irregular, non-random sample of a region. We developed a remote monitoring program for Maine lakes > 8 ha (1511 lakes) to supplement existing field monitoring programs. We combined Landsat 5 Thematic Mapper (TM) and Landsat 7 Enhanced Thematic Mapper Plus (ETM+) brightness values for TM bands 1 (blue) and 3 (red) to estimate water clarity (secchi disk depth) during 1990–2010. Although similar procedures have been applied to Minnesota and Wisconsin lakes, neither state incorporates physical lake variables or watershed characteristics that potentially affect clarity into their models. Average lake depth consistently improved model fitness, and the proportion of wetland area in lake watersheds also explained variability in clarity in some cases. Nine regression models predicted water clarity (R 2 = 0.69–0.90) during 1990–2010, with separate models for eastern (TM path 11; four models) and western Maine (TM path 12; five models that captured differences in topography and landscape disturbance. Average absolute difference between model-estimated and observed secchi depth ranged 0.65–1.03 m. Eutrophic and mesotrophic lakes consistently were estimated more accurately than oligotrophic lakes. Our results show that TM bands 1 and 3 can be used to estimate regional lake water clarity outside the Great Lakes Region and that the accuracy of estimates is improved with additional model variables that reflect physical lake characteristics and watershed conditions.