Freshwater ecosystem monitoring using satellite remote sensing and field surveys

Abstract

Freshwater ecosystems transfer and transform energy, nutrients and carbon. The color of lakes and rivers, as observed from space, can provide clues to their ecological function and response to anthropogenic activities. Despite this, remote sensing science has been slower to mature for freshwaters relative to terrestrial and marine ecosystems. This research pairs physical and biogeochemical measurements collected during field campaigns with airborne and satellite remote sensing to improve our understanding of the links between color and chemistry in lakes and rivers. Using remote sensing, we mapped over 3,000 river miles in order to better understand uncertainties introduced to remote sensing retrievals during atmospheric correction of satellite imagery. Building from this work, we then conducted remote sensing analysis of ~500,000 lakes combined with a subset of intensive field surveys to (1) establish the relationship between color and gross primary productivity in shallow arctic-boreal lakes, (2) quantify multi-decadal trends in lake color across arctic-boreal North America, and (3) evaluate the effect of climate variability on arctic-boreal lake color trends. We found that atmospheric correction can bias model results by 3 – 59% for estimates of chlorophyll-a and turbidity derived from Landsat-8 and Sentinel-2 for three large river systems: the Columbia, Amazon and Mississippi. We also discovered the green band (reflectance ~ 560 nm) was the least impacted by uncertainties from processor or sensor choice, implying that the green band is most suitable for historical or cross-sensor analysis. This chapter has been published in Remote Sensing of the Environment. Applying this framework in the high northern latitudes, moderate resolution (30m) satellite and high resolution (5 m) hyperspectral airborne remote sensing imagery were used to infer gross primary productivity rates from arctic-boreal lakes within the NASA Arctic-Boreal Experiment (ABoVE) domain. This study was published in Environmental Research Letters. We then use the Landsat archive to map annual growing season greenness from 1985 – 2019 for lakes throughout the entire ABoVE domain. Over a quarter of lakes showed significant color change. Declining greenness was the dominant trend and was most common in areas also undergoing warming and precipitation increases. This finding, which has been submitted to Proceedings of the National Academy of the Sciences, provides evidence in support of the hypothesis that warming is restructuring arctic-boreal lake ecological dynamics because of changes, in part due to changing hydrologic connectivity. Collectively, these studies have advanced our understanding of the ecological significance of lake and river color, including opportunities to uncover new relationships between lake color and chemistry by combining novel geochemistry data with remote sensing.