Remote-sensing reflectance and true colour produced by a coupled hydrodynamic, optical, sediment, biogeochemical model of the Great Barrier Reef, Australia: Comparison with satellite data

https://doi.org/10.1016/j.envsoft.2015.11.025Get rights and content

Highlights

  • We calculate remote-sensing reflectance from a biogeochemical model state.

  • Remote-sensing reflectance from the model is compared with observed values.

  • Combined simulated red, green and blue reflectances make simulated true colour.

  • Simulated reflectances and true colour are compared to observations.

  • Simulated true colour of the Great Barrier Reef powerfully displays model output.

Abstract

Aquatic biogeochemical models are vital tools in understanding and predicting human impacts on water clarity. In this paper, we develop a spectrally-resolved optical model that produces remote-sensing reflectance as a function of depth-resolved biogeochemical model properties such as phytoplankton biomass, suspended sediment concentrations and benthic reflectance. We compare simulated remote-sensing reflectance from a 4 km resolution coupled hydrodynamic, optical, sediment and biogeochemical model configured for the Great Barrier Reef with observed remote-sensing reflectance from the MODIS sensor at the 8 ocean colour bands. The optical model is sufficiently accurate to capture the remote-sensing reflectance that would arise from a specific biogeochemical state. Thus the mismatch between simulated and observed remote-sensing reflectance provides an excellent metric for model assessment of the coupled biogeochemical model. Finally, we combine simulated remote-sensing reflectance in a red/green/blue colour model to produce simulated true colour images during the passage of Tropical Cyclone Yasi in February 2011.

Introduction

Coastal environments are often impacted by multiple natural and anthropogenic forcings, including nutrient and sediment runoff, ocean acidification and ocean warming (Parry et al., 2007). Despite these stressors being known for decades, even well-resourced coastal environmental agencies have in the past struggled to optimise resource management (Brodie and Waterhouse, 2012). The state of the art in the management of coastal environments is now looking to combined monitoring and prediction systems to provide near-real-time information (Schiller et al., 2014) and to develop optimal management strategies. This paper will focus on optical model developments to better integrate biogeochemical modelling outputs and satellite ocean colour observations.

The assessment of marine biogeochemical models presently relies primarily on either in-situ observations, that are generally sparse in space and often in time, or remotely-sensed ocean colour (Kidston et al., 2013, Jones et al., 2012, Jones et al., 2015). The observed in-situ biogeochemical quantities are often not direct analogues of model quantities. For example, phytoplankton biomass is typically represented in models using mass concentration of elements such as nitrogen contained within the cells, while the most cost-effective observations of phytoplankton are based on active fluorescence of cells (Earp et al., 2011).

The frequency of satellite observations depends on latitude, cloud cover, orbit paths, glint etc., but, in general, they provide a vastly superior coverage to in situ observations of biogeochemical properties. Remotely-sensed ocean colour observations are more regular, but still do not correlate directly with model quantities except through imperfect remote-sensing reflectance to chlorophyll algorithms (Mobley et al., 2015). Thus what is lacking is a single quantity that can be calculated without unreliable assumptions from biogeochemical models, and that is routinely produced by remote-sensing teams with small errors. In this paper, the common quantity is the remote-sensing reflectance, Rrs, evaluated at multiple wavelengths.

Remote-sensing reflectance is a measure of the water-leaving radiance normalized by the at-surface downwelling solar irradiance and has units of sr−1. Thus both model outputs and observations can be quantified as a remote-sensing reflectance. To avoid confusion, we will use the term observed remote-sensing reflectance to refer to satellite-derived Rrs, and simulated remote-sensing reflectance to refer to Rrs calculated from the outputs of the biogeochemical model.

There is a long history of including optical calculations within an aquatic biogeochemical model (Jassby and Platt, 1976). Initially these models were based on simple exponential decay of scalar, spectrally-averaged irradiance with depth (Fasham, 1993, Taylor et al., 1997), and were developed primarily as a means of forcing light-limited phytoplankton growth functions. Later spectrally-resolved models were also considered (Gregg and Carder, 1990, Baird et al., 2007, Nerger and Gregg, 2007) as a means of providing better-constrained optical parameters values to improve predictive capabilities. It has also become apparent that with improved optical observations, such as in-situ radiometers, more sophisticated optical models facilitated a more complete comparison between model and observations (Fujii et al., 2007, Mobley et al., 2015).

Recently a global biogeochemical model has been coupled to a spectrally-resolved optical model to produce remote-sensing reflectance values that can be compared with remotely-sensed observations (Dutkiewicz et al., 2015). The Dutkiewicz et al. (2015) model produced annually averaged patterns of remote-sensing reflectance in 3 bands. Their simulated remote-sensing reflectances were similar to the Moderate-resolution Imaging Spectroradiometer (MODIS) composites, with a small positive bias. Being a global scale, blue water study, Dutkiewicz et al. (2015) were able to ignore the influence of the benthic reflectance and the scattering of suspended sediment, but these processes must be included for coastal waters such as considered in this paper. Nonetheless Dutkiewicz et al. (2015) is an important study that shows the utility of assessing biogeochemical models using remote-sensing reflectance calculated by an optical model from biogeochemical model output. Similarly, Mobley et al. (2015) presents simulated remote-sensing reflectance, although the study was idealised so no comparison with observations was possible.

The GBR ecosystem, described as one of the seven natural wonders of the world, is under increasing pressure from local and global anthropogenic stressors (De'ath et al., 2009). Decreasing water clarity due to nutrient and sediment pollution is one of the most serious threats to the GBR ecosystem (Thompson et al., 2014), with the primary concern being the impact of lower benthic light levels on coral and seagrass communities (Collier et al., 2012, Petrou et al., 2013).

In addition to water clarity being critical for the functioning of shallow-water systems, remote-sensing of water clarity is increasingly being used to manage shallow-water ecosystems. For example, Devlin et al. (2013) has categorised plumes into primary, secondary and tertiary extents using remotely-sensed ocean colour, and then used these estimates of extent as a means of determining the frequency of impact of river plumes on coral reefs on the GBR in the vicinity (∼50 km) of large tropical rivers. The accurate simulation of remote-sensing reflectance by a coupled circulation, optical and biogeochemical model will allow it to be analysed using the same technique to that developed for remotely-sensed ocean colour.

Thus, a modelling system that produces remote-sensing reflectance as a model output will be better assessed on its skill at determining water clarity and will have improve utility for management planning than the biogeochemical models presently used in coastal applications.

This paper describes a spectrally-resolved optical model for optically-complex coastal and open-ocean waters, detailing the calculation of inherent optical properties (IOPs) and apparent optical properties (AOPs), including remote-sensing reflectance. The model is sophisticated enough to provide realistic remote-sensing reflectance at any chosen wavelength, but computationally simple enough to be used within a ∼4 km resolution, 47 layer, multi-year simulation of the 2000 km long Great Barrier Reef, with a 60 state variable biogeochemical model and 19 optically-significant components. The simulated remote-sensing reflectance fields are calculated from the biogeochemical model state and then compared to remotely-sensed observations. Finally, the simulated remote-sensing reflectances are used to produce simulated true colour images of the Great Barrier Reef at the time of Tropical Cyclone (TC) Yasi when the ocean surface was obscured by clouds, providing a new appreciation of the magnitude and spatial extent of changes in water clarity during an extreme event.

Section snippets

Methods

A brief description of the modelling system is given in Appendix A. Here we concentrate on the bio-optical model.

Simulated remote-sensing reflectance

A hindcast of the coupled model was run from 1 Sep 2010 to near present, producing 3 dimensional distributions of optically-significant water column constituents and spatially- and temporally-varying benthic substrates. We use the above-described optical model to produce simulated remote-sensing reflectance from the hindcast biogeochemical output at midday each day. We focus on 2011 as it contained an extremely wet summer (Oubelkheir et al., 2014) and a severe tropical cyclone (Great Barrier

Discussion

A spectrally-resolved optical model has been developed to calculate remote-sensing reflectance from biogeochemical state in order to assess the performance of the biogeochemical model. Thus, the success of this study relies on the optical model being more accurate than the biogeochemical model's ability to predict the distribution and concentration of optically-significant constituents, such that when a mis-match occurs between observed remote-sensing reflectance and simulated remote-sensing

Acknowledgements

The model simulations were developed as part of the eReefs project, a public-private collaboration between Australia's leading operational and scientific research agencies, government, and corporate Australia. Atmospherically-corrected MODIS products were sourced from the Integrated Marine Observing System (IMOS) – IMOS is supported by the Australian Government through the National Collaborative Research Infrastructure Strategy and the Super Science Initiative.We thank the many colleagues

References (79)

  • E.A. King et al.

    A pre-operational system for satellite monitoring of the Great Barrier Reef Marine water quality

    Tech. Rep.

    (2014)
  • N. Margvelashvili

    Stretched Eulerian coordinate model of coastal sediment transport

    Comput. Geosci.

    (2009)
  • T.S. Moore et al.

    A class-based approach to characterizing and mapping the uncertainty of the MODIS ocean chlorophyll product

    Remote Sens. Env.

    (2009)
  • L. Nerger et al.

    Assimilation of SeaWiFS data into a global ocean-biogeochemical model using a local SEIK filter

    J. Mar. Sys

    (2007)
  • C. Petus et al.

    Monitoring spatio-temporal variability of the Adour River turbid plume (Bay of Biscay, France) with MODIS 250-m imagery

    Cont. Shelf Res.

    (2014)
  • T. Quaife et al.

    Assimilating canopy reflectance data into an ecosystem model with an Ensemble Kalman Filter

    Remote Sens. Env.

    (2008)
  • K.R. Ridgway et al.

    Mesoscale structure of the mean East Australian Current system and its relationship with topography

    Prog. Oceanogr.

    (2003)
  • A. Schiller et al.

    Cross-shelf exchanges between the Coral Sea and the Great Barrier Reef lagoon determined from a regional-scale numerical model

    Cont. Shelf Res.

    (2015)
  • T. Schroeder et al.

    Inter-annual variability of wet season freshwater plume extent into the Great Barrier Reef lagoon based on satellite coastal ocean colour observations

    Mar. Poll. Bull.

    (2012)
  • K. Wild-Allen et al.

    Applied coastal biogeochemical modelling to quantify the environmental impact of fish farm nutrients and inform managers

    J. Mar. Sys

    (2010)
  • D. Antoine et al.

    Variability in optical particle backscattering in contrasting bio-optical oceanic regimes

    Limnol. Oceanogr.

    (2011)
  • M.J. Atkinson et al.

    C:N:P ratios of benthic marine plants

    Limnol. Oceanogr.

    (1983)
  • R.C. Babcock et al.

    Towards an Integrated Study of the Gladstone Marine System

    (2015)
  • M.E. Baird et al.

    CSIRO Environmental Modelling Suite: Scientific Description of the Optical, Carbon Chemistry and Biogeochemical Models Parameterised for the Great Barrier Reef

    (2014)
  • M.E. Baird et al.

    A dynamic model of the cellular carbon to chlorophyll ratio applied to a batch culture and a continental shelf ecosystem

    Limnol. Oceanogr.

    (2013)
  • M.E. Baird et al.

    The effect of packaging of chlorophyll within phytoplankton and light scattering in a coupled physical-biological ocean model

    Mar. Fresh. Res.

    (2007)
  • D. Blondeau-Patissier et al.

    Bio-optical variability of the absorption and scattering properties of the Queensland inshore and reef waters, Australia

    J. Geophys. Res. (Oceans)

    (2009)
  • V.E. Brando et al.

    Adaptive semianalytical inversion of ocean color radiometry in optically complex waters

    Appl. Opt.

    (2012)
  • D.G. De’ath et al.

    Declining coral calcification on the Great Barrier Reef

    Science

    (2009)
  • A.G. Dekker et al.

    Intercomparison of shallow water bathymetry, hydro-optics, and benthos mapping techniques in Australian and Caribbean coastal environments

    Limnol. Oceanogr. Methods

    (2011)
  • M. Devlin et al.

    Combining in-situ water quality and remotely sensed data across spatial and temporal scales to measure variability in wet season chlorophyll-a: Great Barrier Reef lagoon (Queensland, Australia)

    Ecol. Process.

    (2013)
  • H.M. Dierssen et al.

    Red and black tides: quantitative analysis of water-leaving radiance and perceived color for phytoplankton, colored dissolved organic matter, and suspended sediments

    Limnol. Oceanogr.

    (2006)
  • S. Dutkiewicz et al.

    Capturing optically important constituents and properties in a marine biogeochemical and ecosystem model

    Biogeosci. Discuss.

    (2015)
  • L.N.M. Duysens

    The flattening of the absorption spectra of suspensions as compared to that of solutions

    Biochim. Biophys. Acta

    (1956)
  • A. Earp et al.

    Review of fluorescent standards for calibration of in situ fluorometers: recommendations applied in coastal and ocean observing programs

    Opt. Express

    (2011)
  • M.J.R. Fasham

    Modelling the marine biota

  • D. Ficek et al.

    Spectra of light absorption by phytoplankton pigments in the Baltic; conclusions to be drawn from a Gaussian analysis of empirical data

    Oceanologia

    (2004)
  • M. Fujii et al.

    The value of adding optics to ecosystem models: a case study

    Biogeosci. Discuss.

    (2007)
  • M. Furnas

    Catchments and Corals: Terrestrial Runoff to the Great Barrier Reefs

    (2003)
  • Cited by (64)

    • An ensemble neural network atmospheric correction for Sentinel-3 OLCI over coastal waters providing inherent model uncertainty estimation and sensor noise propagation

      2022, Remote Sensing of Environment
      Citation Excerpt :

      Recent validation studies comparing various AC algorithms applied to Sentinel-3 OLCI over coastal and inland waters show mean absolute percentage retrievals errors in the range of 30–440% in the blue spectral region at 412.5 nm, 7–52% in the green at 560 nm and 11–78% in the red at 708 nm (Bi et al., 2018; Li et al., 2019; Alikas et al., 2020; Shen et al., 2020; Giannini et al., 2021; Vanhellemont and Ruddick, 2021). In addition to a good overall retrieval accuracy certain applications, such as the assimilation of marine reflectance into biogeochemical models (BGCM) (Baird et al., 2016; Jones et al., 2016; Steven et al., 2019)⁠⁠, also require information about the pixel-by-pixel retrieval uncertainty. A new study of the impact of measurement uncertainties has been presented recently by McKinna et al. (2021) who demonstrated how standard uncertainties might be included into ocean colour validation by correcting their validation results of an empirical ocean colour algorithm for model and observation uncertainties.

    View all citing articles on Scopus
    View full text