RSPSoc2008 NERC Field Spectroscopy Facility Aquatic
Instrument Demonstration
28 November 2008
John Hedley1, Alasdair MacArthur2, Tim Malthus2
1. School of Biosciences, University of Exeter. E-mail: j.d.hedley@exeter.ac.uk
2. NERC Field Spectroscopy Facility, University of Edinburgh. E-mail: fsf@nerc.ac.uk
Introduction
This report is a summary of the results obtained during the field demonstration of the NERC Field Spectroscopy
Facilities aquatic optical sampling equipment, at Stithians Lake in September 2008, as part of the Remote
Sensing and Photogrammetric Society conference. A combination of equipment was demonstrated, a WETLabs
AC-S and ECO-BB3 for measuring optical properties of the water itself, known as “Inherent Optical Properties“
(IOPs), in conjunction with Satlantic HyperOCR sensors for measuring the actual light field in the water – from
which “Apparent Optical Properties” (AOPs) can be derived. AOPs include parameters such as the spectral
diffuse attenuation of downwelling irradiance, kd(λ), and are derived from an actual measured light field so their
extrapolation to other situations is only approximate, since for example the calculated values may depend on the
position of the sun or the directional nature of the radiance field. IOPs on the other hand represent the physical
optical characteristics of the water itself, measured independently from any actual light field. To estimate actual
light fields or AOPs from IOPs numerical modelling is required to propagate the incident above-water radiance
distribution from the sky through the water column. This report also introduces PlanarRad - a free software tool
written by Dr. John Hedley that will be available from the FSF in early 2009 (Fig. 1). PlanarRad is an opensource C++ implementation of the Invariant Imbedded method for radiative transfer in plane-parallel waters as
detailed in Curt Mobley’s book Light and Water (Mobley, 1994). In this report we demonstrate a simple model
closure experiment – where we compare of actual light fields and measured AOPs at Stithians Lake with those
predicted from modelling with the concurrently collected IOP data.
Methods
IOPs - Beam attenuation and
absorption were measured using a
WetLABS AC-S in 84 bands from
400 nm to 760 nm. Temperature,
depth and salinity were measured
with a SBE19 CTD and backscatter
coefficients at 117º at 470 nm, 532
nm and 660 nm were measured
using
WetLABS
ECO-BB3
backscatterometer. All IOP data was
post-processed
and
corrected
according to the manufacturers
protocols, and the backscatter data
was used to model Fournier-Forand
phase functions describing the
angular distribution of scattering, as
required for input to the PlanarRad
model (Mobley et al. 2002).
Fig. 1. PlanarRad open-source C++ software for modelling plane-parallel
radiative transfer in natural waters. The software was specifically designed to
be integrated with the FSF instrumentation suite.
AOPs – downwelling planar
irradiance and upwelling radiance
were measured using Satlantic
HyperOCR sensors with 182 bands
(a)
(b)
(c)
Fig. 2. Inherent Optical Properties (IOPs) at Stithians Lake
from 350 nm to 950 nm. The above-water incident radiance directional distribution was estimated by using an
additional air-calibrated Satlantic HyperOCR sensor to take a total downwelling irradiance reading and a shaded
reading (for separating the direct and diffuse sky irradiance components). From this data the PlanarRad software
builds as input a directional sky radiance model based on the empirical sky parameterisation of Grant et al.
(1996). PlanarRad was then used to model the in-water light field based on a total depth of 3 m and an assumed
Lambertian substrate reflectance of 0.03. The visualisation tools of PlanarRad allowed for immediate
investigation of model closure, since the measured and modelled data can be plotted overlayed, and updated
interactively as the depth of interest is adjusted (Fig. 1).
Results and Discussion
IOPs - Fig. 2a shows the depth averaged spectral beam attenuation and absorption measured by the WetLABS
AC-S, note this data includes absorption by pure water itself. Features to note are that CDOM (Coloured
Dissolved Organic Matter) absorption is significant – this is the steep exponential shape to the curve from 400
nm to 570 nm and gave the water its brown colour. Absorption in the Near-Infra Red (NIR) is also significant
and is due to the NIR absorption by pure water itself. A detailed plot (Fig. 2b) shows three spectral absorption
features at approximately 583 nm, 640 nm, and 690 nm, the latter two are features consistent with the pattern of
specific spectral absorptions published for some species of phytoplankton (690 nm feature is due to Chlorophyll,
Mobley, 1994). Fig. 2c shows the depth-averaged Fournier-Forand phase functions estimated from the
backscatter coefficients, note there is very little spectral variability in angular nature of scattering although
backscatter in the green wavelength (532 nm) is slight lower than red and blue.
AOPs - Fig. 3 (left panel) shows downwelling spectral planar irradiance, Ed, as measured at 0.5 m and 1.0 m
compared to the PlanarRad model output based on the IOPs and modelled sky radiance input. The spectral fit
between modelled and measured irradiance is quite good, but the model outputs result in an underestimation of
the attenuation of irradiance. Fig, 3 (right panel) shows modelled and measured upwelling radiance, Lu, at 1.0
and 1.5 m. Model closure is good from 400 nm to 600 nm, but in this case the model underestimates upwelling
radiance at greater depths and longer wavelengths, and spectral features are more prominent in the measured
data. Fig. 3 also shows upward and downward looking hemispherical projections of the directional radiance
from the PlanarRad model at the same depths as the spectral plots.
Model Closure - There are several possible reasons for the observed deviations between the measured light
levels and the model predictions:
1) The available depth at the site was inadequate for this kind of experiment – since the instrumentation frame
itself is about 1 m high only a very small sampling depth range was possible. Relating data at depths
differing by only 0.5 m will be subject to large potential errors due to fluctuations in the light field
Upward-looking hemispherical radiance projections
0.5 m
1.0 m
Down-looking hemispherical radiance projections
1.0 m
1.5 m
Fig 3. Measured and modelled downwelling irradiance and upwelling radiance. Hemispherical radiance
projections are a visualisation output from PlanarRad to indicate the directional distribution of radiance at
different depths in the water column (the final version of PlanarRad will provide a scale for these plots).
2) The reflectance of the substrate was not measured, but was assumed to be 0.03 and spectrally flat.
3) Recent work has shown that the standard data correction protocol for the AC-S can lead to underestimations
of absorption (McKee et al. 2008). This may be an area of future research interest for the Field Spectroscopy
Facility.
In conclusion, given the limited time available and restrictions of collecting only a single data set, the results
from the AOP and IOP data are quite consistent. Model closure experiments and issues surrounding uncertainty
in IOP measurements and the propagation of uncertainty through models are currently hot topics in hydrological
optics. The combination of AOP and IOP profiling sensors available from the FSF, together with the soon to be
available open-source PlanarRad software make an excellent test bed for investigating these issues, and for
building a reliable understanding the nature of underwater light field.
Acknowledgments
The authors extend thanks to Ian Walsh of WETLabs, the RSPSoc2008 organising committee and the attendees
at the field demonstration.
References
Grant, R. H., Hiesler, G. M., and Gao, W. (1996). Photosynthetically-active radiation: sky radiance distributions under clear
and overcast conditions. Agricultural and Forest Meteorology, 82, 267-292.
McKee, D., Piskozub, J., and Brown, I. (2008). Scattering error corrections for in situ absorption and attenuation
measurements. Optics Express, 16, 19480-19492.
Mobley, C. D. (1994). Light and Water. Academic Press, San Diego.
Mobley, C. D., Sundman, L. K., and Boss, E. (2002). Phase function effects on oceanic light fields. Applied Optics, 41,10351050.