Regional measurements and spatial/temporal analysis of CDOM in 10,000+ optically variable Minnesota lakes using Landsat 8 imagery
Graphical abstract
Introduction
Research in recent decades has revealed a central role for colored (or chromophoric) dissolved organic matter (CDOM) in regulating major physical, chemical and biological processes in lakes and rivers (e.g., reviewed in Solomon et al., 2015, Williamson et al., 1999, Creed et al., 2018, and elsewhere). We now know that CDOM functions as one of a small number of “master variables, “ similar to phosphorus, pH and redox potential, that control important aspects of the composition and functioning of aquatic ecosystems and regulate their responses to environmental change (Williamson et al., 1999; Creed et al., 2018). Recent studies show that CDOM levels strongly influence: (a) light and thermal regimes in lakes (e.g., Houser, 2006; Ask et al., 2009; Thrane et al., 2014; Pilla et al., 2018; Snucins and Gunn, 2000), (b) biogeochemical cycles (e.g., Knoll et al., 2018; Corman et al., 2018), (c) food web processes and interactions (e.g., Karlsson et al., 2009; Solomon et al., 2015), (d) contaminant bioavailability (e.g., Tsui and Finlay, 2011), and (e) water clarity (e.g., Brezonik et al., 2019a). Knowledge of the sources, levels, and cycling of CDOM in freshwaters thus is important for aquatic resource management and for predicting the outcomes of environmental change.
Moderate to high levels of CDOM in freshwaters are determined largely by rates of transport from soils and wetlands in surrounding watersheds and thus are affected by a combination of factors related to vegetation and hydrology. The dependency of aquatic CDOM on dynamic external sources, combined with internal production and loss processes in aquatic systems, can lead to high variability of CDOM levels across landscapes and within lakes at time scales of seasons to years (Brezonik et al., 2015, Williamson et al., 1999). Human-driven changes in temperature, atmospheric chemistry, land use and watershed hydrology also can have strong effects on CDOM (Creed et al., 2018, Finstad et al., 2016, Kritzberg, 2017, Stanley et al., 2012, de Wit et al., 2016).
Although CDOM is easily measured in the laboratory, the availability of in situ CDOM data is surprisingly limited relative to its importance, even in states like Minnesota, where monitoring of its >10,000 surface waters is a major focus of many state, tribal and local agencies. Several recent, large-scale assessments of regional U.S. lake monitoring efforts (Stanley et al., 2012; Ross et al., 2019) showed that far fewer data were available for CDOM and related variables such as DOC compared to nutrients, chlorophyll, and water clarity, despite the strong effects of CDOM on those and other physicochemical variables. The spatial and temporal variation in CDOM in surface waters suggests the need for more CDOM data to improve understanding of drivers and better predict lake responses to stresses ranging from local land cover changes to global climate change. Some countries with large numbers of CDOM-rich lakes have incorporated routine monitoring of CDOM or a related parameter such as DOC (e.g., Sobek et al., 2007). The relative lack of CDOM data for U.S. lakes (Stanley et al., 2019) may stem from the fact that many monitoring programs initially started in relatively low-CDOM regions but also from the fact that the importance of CDOM as a driver of ecological conditions has been appreciated only recently.
Whatever the cause, the availability of CDOM data remains deficient compared to its importance. Remote sensing using satellite-based sensors could play an important role in providing CDOM data at high temporal and spatial resolution. Recent studies show that the Landsat sensors (Kutser et al., 2005; Brezonik et al., 2005; Kutser et al., 2009; Olmanson et al., 2016a), and Sentinel-2/MSI sensors (Toming et al., 2016; Chen et al., 2017) can provide such data at scales relevant for inland lakes as small as 4 hectares (ha).
Recent improvements in Earth-observing satellite sensors have expanded the capabilities to measure optically-related water quality characteristics, including CDOM, in lakes (Olmanson et al., 2016a; Tyler et al., 2016; Pahlevan et al., 2019; Page et al., 2019). Specifically, the Landsat 8 Operational Land Imager (L8/OLI) and the European Space Agency (ESA) Sentinel-2 MultiSpectral Imager (S2/MSI) have improved spatial, spectral, radiometric and temporal resolution compared with earlier sensors. With the L8/OLI and S2/MSI constellation collecting imagery every 3 to 5 days, frequent satellite-based measurements of a variety of key water quality variables on lakes are now possible.
The use of satellite imagery to measure CDOM at large regional scales and over multiple time periods requires analysis of multiple images. Unless ground-based data are available to calibrate each image (a requirement difficult to achieve), accurate methods are needed for atmospheric correction of images to produce surface reflectance data directly representative of optical signals from waterbodies. Although various approaches have been reported to accomplish this (e.g., Pahlevan et al., 2017a, Pahlevan et al., 2017b; Vanhellemont and Ruddick, 2015, Vanhellemont and Ruddick, 2016), we have found that many of them yield unreliable results for inland lakes (Olmanson et al., 2011; Page et al., 2019). The recent availability of surface reflectance products from the EROS Center appears to have overcome this obstacle for Landsat 8 imagery (Kuhn et al., 2019), and Page et al. (2019) described a workflow process to atmospherically correct and harmonize S2/MSI and L8/OLI satellite imagery in Google Earth Engine (GEE) (Gorelick et al., 2017).
This paper describes application of these advances to measure CDOM on all waterbodies larger than 4 ha across a large geographic region (the state of Minnesota) that encompasses >226,000 km2 and contains officially 11,842 lakes 4 ha or larger in area (https://www.dnr.state.mn.us/faq/mnfacts/water.html). The paper describes a robust semi-empirical approach for routine monitoring of CDOM using L8/OLI imagery. We demonstrate the consistency and reliability of two atmospheric correction methods to generate remote sensing reflectance (Rrs) products and use these products to assemble a CDOM database on >10,500 lakes for both 2015 and 2016. We assess the accuracy of retrieved CDOM data for both low- and high-CDOM waters and summarize distributions of CDOM in Minnesota lakes at the ecoregion level.
Section snippets
Study area
Minnesota, a large, lake-rich state in the Upper Midwest of the U.S., comprises parts of seven ecoregions (Omernik and Griffith, 2014) that differ in land cover, geology, soils, vegetation and hydrologic conditions (Fig. 1). Known popularly as “the land of 10,000 lakes,” Minnesota actually has approximately 12,000 waterbodies with surface areas ≥4 ha (Olmanson et al., 2014) and many more that are smaller than that. The lakes are distributed broadly (but not uniformly) across the ecoregions. Two
CDOM model results
After exploration of various two-term regression models using L8/OLI data, we identified the best model as having the form:where coefficients, a, b, and c were fit to the calibration data by regression analysis, ln(a440) is the natural logarithm of the L8/OLI-derived a440 for a given sample location and B represents the corresponding L8/OLI spectral band. From the combined L8/OLI dataset, the ln(a440) prediction model generated a strong fit with R2 = 0.85 and
Conclusions
This paper demonstrates that remote sensing using satellite-based sensors can play an important role in providing census-level CDOM data over large areas at high temporal and spatial resolution. The constellations of L8/OLI, upcoming Landsat 9/OLI and Sentinel 2/MSI will greatly expand the capabilities to measure several optically-related water quality characteristics, including CDOM.
Strong relationships for CDOM (a440) were found using both MAIN and OLI-SR atmospheric correction methods.
Funding
This work was supported in part by National Science Foundation grant (CBET 1510332), Minnesota Environment and Natural Resources Trust Fund, Minnesota Agricultural Experiment Station, and University of Minnesota's U-Spatial Program, Sea Grant Program and Office of the VP for Research and Retirees Association.
Credit authorship contribution statement
Leif G. Olmanson: Conceptualization, Methodology, Formal analysis, Data curation, Writing - original draft, Writing - review & editing. Benjamin P. Page: Conceptualization, Methodology, Writing - original draft, Writing - review & editing. Jacques C. Finlay: Conceptualization, Data curation, Writing - original draft, Writing - review & editing. Patrick L. Brezonik: Conceptualization, Formal analysis, Data curation, Writing - original draft, Writing - review & editing. Marvin E. Bauer: Writing -
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
We thank the Minnesota Pollution Control Agency's lake water quality assessment program and numerous collaborators, research staff, and students for assistance in sample collection and analysis.
References (50)
- et al.
Factors affecting the measurement of CDOM by remote sensing of optically complex inland waters
Remote Sens. Environ.
(2015) - et al.
Assessment of the chlorine demand and disinfection byproduct formation potential of surface waters via satellite remote sensing
Water Res.
(2019) - et al.
Google earth engine: planetary-scale geospatial analysis for everyone
Remote Sens. Environ.
(2017) - et al.
Limitations on using CDOM as a proxy for DOC in temperate lakes
Water Res.
(2018) - et al.
Performance of Landsat-8 and Sentinel-2 surface reflectance products for river remote sensing retrievals of chlorophyll-a and turbidity
Remote Sens. of Environ.
(2019) - et al.
Mapping lake CDOM by satellite remote sensing
Remote Sens. Environ.
(2005) - et al.
Development and analysis of a 20-year Landsat water clarity census of Minnesota’s 10,000 lakes
Remote Sens. Environ.
(2008) - et al.
Comparison of Landsat 8 and Landsat 7 for regional measurements of CDOM and water clarity in lakes
Remote Sens. Environ.
(2016) - et al.
A harmonized image processing workflow using Landsat-8 and Sentinel-2 for mapping water clarity in optically complex lake systems
Remote Sens. Environ.
(2019) - et al.
Landsat 8 remote sensing reflectance (Rrs) products: evaluations, intercomparisons and enhancements
Remote Sens. Environ.
(2017)
Sentinel-2 multi spectral instrument (MSI) data processing for aquatic science applications: demonstrations and validations
Remote Sens. Environ.
Sentinel-2/Landsat-8 product consistency and implications for monitoring aquatic systems
Remote Sens. Environ.
Developments in Earth observation for the assessment and monitoring of inland, transitional, coastal and shelf-sea waters
Sci. Total Environ.
Advantages of high quality SWIR bands for ocean colour processing: examples from Landsat-8
Remote Sens. Environ.
Terrestrial organic matter and light penetration: effects on bacterial and primary production in lakes
Limnol. Oceanogr.
Landsat-based remote sensing of lake water quality characteristics, including chlorophyll and colored dissolved organic matter (CDOM)
Lake Reserv. Manage.
Color, chlorophyll a, and suspended solids effects on Secchi depth in lakes: implications for trophic state assessment
Ecol. Appl.
Iron influence on dissolved color in lakes of the Upper Great Lakes States
PLoS One
Remote estimation of colored dissolved organic matter and chlorophyll-a in Lake Huron using Sentinel-2 measurements
J. Appl. Remote. Sens.
Nitrogen and phosphorus loads to temperate seepage lakes associated with allochthonous dissolved organic carbon loads
Geophys. Res. Lett.
Global change-driven effects on dissolved organic matter composition: implications for food webs of northern lakes
Glob. Chang. Biol.
From greening to browning: catchment vegetation development and reduced S-deposition promote organic carbon load on decadal time scales in Nordic lakes
Sci. Rep.
Retrieval of water-leaving radiance and aerosol optical thickness over the oceans with SeaWiFS: a preliminary algorithm
Appl. Opt.
Spatial and interannual variability of dissolved organic matter in the Kolyma River, East Siberia, observed using satellite imagery
Journal of Geophysical Research– Biogeosciences
Water color affects the stratification, surface temperature, heat content, and mean epilimnetic irradiance of small lakes
Can. J. Fish. Aquat. Sci.
Cited by (36)
Monitoring of reservoir CDOM degradation process by remote sensing of its surface waters signal
2024, Remote Sensing Applications: Society and EnvironmentApplication of airborne hyperspectral imagery to retrieve spatiotemporal CDOM distribution using machine learning in a reservoir
2022, International Journal of Applied Earth Observation and GeoinformationAssessing the current water clarity status of ~100,000 lakes across southern Canada: A remote sensing approach
2022, Science of the Total EnvironmentCitation Excerpt :Indeed, light absorption and scattering characteristics of substances in water bodies interact with incoming solar radiation to produce spectral signatures that can be measured through developing remote sensing algorithms (Matthews, 2011; Odermatt et al., 2012). To date, the publication record for remote sensing of inland water quality has focused mostly on optical water property retrieval algorithm development (Topp et al., 2020), with significantly fewer studies taking the step of applying algorithms to estimate broad patterns in these optical properties to unsampled lakes (Hicks et al., 2013; Ho et al., 2019; Hou et al., 2017; Lehmann et al., 2018; Lymburner et al., 2016; Olmanson et al., 2020, 2014, 2008; Song et al., 2020). Part of the problem is the complex optical nature of Case II waters, where most developed algorithms rely on locally or regionally calibrated empirical or semi-empirical relationships between optically active in situ water quality parameters and satellite reflectance information (Niroumand-Jadidi et al., 2020; Odermatt et al., 2012; Shi et al., 2013).
Using CDOM spectral shape information to improve the estimation of DOC concentration in inland waters: A case study of Andean Patagonian Lakes
2022, Science of the Total EnvironmentCitation Excerpt :Moreover, the inherent optical properties of CDOM provide a direct link to remote sensing retrievals. The remote estimation of CDOM absorption in the vast majority of lakes is now feasible due to the recent launches of satellites with sensors that provide the proper spectral, radiometric, and spatial resolutions (e.g., Brezonik et al., 2015; Kutser et al., 2016; Olmanson et al., 2016, 2020; Toming et al., 2016). Light absorption by CDOM can present a strong correlation with the bulk of DOC in different estuarine and inland waters (e.g., Brezonik et al., 2015; Osburn and Stedmon, 2011; Spencer et al., 2012).