The SeaSWIR dataset consists of 137 ASD (Analytical Spectral Devices, Inc.)
marine reflectances, 137 total suspended matter (TSM) measurements and 97
turbidity measurements gathered at three turbid estuarine sites (Gironde, La
Plata, Scheldt). The dataset is valuable because of the high-quality
measurements of the marine reflectance in the Short Wave InfraRed I region
(SWIR-I: 1000–1200
The SeaSWIR dataset has been made publicly available
(
The SeaSWIR dataset
(
Locations of the study sites: Scheldt (Belgium), Gironde (France), and Río de La Plata (Argentina).
The marine reflectance in the SWIR has not been reported before because
specialized optical instrumentation is required to measure in the SWIR. More
information on the SWIR marine reflectance is essential because several
satellites have sensors with spectral bands in this range. Examples include
Sentinel-3 OLCI with a spectral band at 1020
To measure marine reflectance in the SWIR-I and SWIR-II, a typical “land”
ASD (Analytical Spectral Devices, Inc.) spectroradiometer (FieldSpec) was
used, recording data from 350 to 2500
The measurement campaigns resulted in a large dataset of 137 ASD reflectance spectra in the visible/near-infrared (VNIR) and SWIR with corresponding TSM concentrations and turbidity. This paper describes the methodology for data acquisition and processing and analyzes the final reflectance, TSM concentrations and turbidity values. Additional quality control of the ASD reflectance is performed with simultaneously acquired TRIOS reflectance data.
Field data have been collected in three different estuaries with high to extremely high TSM concentrations: the Scheldt in Belgium, the Gironde in France and the Río de la Plata in Argentina (Fig. 1). These three estuaries were chosen because of their high concentration range of suspended particles and diversity of particle composition/type. It was expected that the high concentration of TSM would result in a measurable water reflectance signal in the SWIR.
The Scheldt River has its origin in France, flows through Belgium and reaches
the North Sea in the Netherlands. It forms a macrotidal estuary leading to
TSM concentrations within surface waters varying from a few to several
hundreds of
The Gironde is a macrotidal estuary where the water optical properties are
dominated by non-algal particles, i.e., by suspended sediments (silts and clays) delivered by the Garonne and
Dordogne rivers and trapped within the maximum turbidity zone of the estuary.
TSM concentrations typically vary from 10 to
4000
The Río de la Plata is a shallow (
Date and location of in situ campaigns.
Most of the measurements from the SeaSWIR dataset were collected from fixed
pontoons along the sides of the rivers/estuaries. These pontoons allow
measurement of a large variation of TSM concentration and
Five sampling campaigns and in total 23 measurement days were organized between 2010 and 2013 on all test sites (Table 1).
In 2010, the first field campaigns were organized at the Sint Anna pontoon on the Scheldt River (Antwerp, Belgium). The campaign included 2 days of measurement, 1 day in July and 1 day in October. The campaign was repeated in the same pontoon in June 2012 with a 2-day campaign.
Later in the same month, an extensive field campaign was organized in the
Gironde estuary (France) with 3 days of measurements from a vessel and 3 days
of measurements from fixed pontoons.
1st day (11 June 2012) onboard the research vessel: fixed location close
to the river mouth, 85 2nd day (12 June 2012) onboard the research vessel: fixed location
67 3rd day (13 June 2012) onboard the research vessel: fixed location
30 4th day (14 June 2012) from a pontoon located in Blaye, right shore of
the estuary 5th day (15 June 2012) from the harbor wall located in Pauillac, left
shore of the estuary
During 2 weeks (12–23 November 2012) a 9-day campaign was organized at the
Club de Pescadores Pier at the Río de la Plata in Buenos Aires
(Argentina).
Finally a last campaign was organized in the Gironde in August 2013. A 5-day sampling campaign was organized at the pontoons of Pauillac and Blaye.
All measurements are listed in the tables below.
At all test sites concomitant water reflectance, turbidity and TSM
measurements were made (
Data collection. Campaign site location and year, type of platform used, number of radiometry (using ASD and Trios radiometers), total suspended matter, and turbidity measurements collected.
Water reflectance (
As the ASD has only one radiometer, three consecutive measurements have to be
made to obtain the downwelling irradiance, the total upwelling radiance and
the downwelling sky radiance. To do so, the instrument was mounted on a steel
frame which can be rotated easily keeping a 40
Measurements were performed under consistently cloudy and sunny skies. Patchy clouds and highly variable light conditions were avoided. The three individual measurements are performed as follows.
The downwelling irradiance above the surface (
Instrument setup for
The total upwelling radiance from the water (
Instrument setup for
The downwelling sky radiance (
Instrument setup for
A measurement procedure was defined of 21 consecutive measurements which are
preceded by the optimization of the spectrometer integration time.
Optimization is performed by pointing the optic head of the spectrometer
towards the center of the Spectralon panel. Optimization results in automatic
setting of gains and offsets for the SWIR detector, automatic setting of the
integration time value for the VNIR detector and automatic dark current
collection. The instrument was re-optimized after a change in lighting
conditions. Then the measurement sequence is as follows. sequence 1: sequence 2 sequence 3:
The water reflectance (
The average and standard deviations of
Finally, an extra white reflectance correction was performed for residual
glint by subtracting the water reflectance at 1305 Evaluating the three The standard deviation of After subtracting 1350
The SeaSWIR dataset contains all water reflectance measurements retained
after quality control, as well as the downwelling irradiance (average of
ASD water reflectance for all data campaigns.
The water reflectance is measured with three TriOS-RAMSES hyperspectral
spectroradiometers. Two spectroradiometers measure radiance and one measures
the downwelling irradiance. The instruments are mounted on a frame, which is
projected over the side of the platform using a 2
TSM was analyzed from water samples collected and filtered in the field. The
filters for the TSM analysis (Whatman GF/F filters with a nominal pore size
of 0.7
At the test sites, water samples were collected in brown bottles just below
the water surface. The water (volumes of 3 to 100
Systematically the water samples were filtered in triplicates in order to determine the precision of the TSM concentration measurements. After removal of outliers the average of the triplicates was used for further analysis. When TSM was measured by both the VITO and LOV laboratories, the average was also used here.
Turbidity was measured using portable HACH 2100P and 2100QIS turbidimeters as
in Nechad et al. (2009). The instrument records turbidity between 0 and 1000 FNU, with a
resolution of three significant figures. A 10
The SeaSWIR reflectance dataset is generally characterized by low reflectance
values at short wavelengths (
The spectra for the Scheldt (data campaigns Scheldt2010 and Scheldt2012,
Fig. 5a and b) have overall high reflectance values. In region A the spectra
have a similar slope related to light absorption of CDOM and non-algal
particles. In region B, maximum reflectance values are found of up to 0.1 and
the spectra show clear changes in shape and magnitude. The influence of light
absorption by chlorophyll
The spectra for La Plata are shown in Fig. 5e. All the spectra have a very
similar shape and have only small differences in magnitude. Maximum
reflectance values are found around 700
The Gironde spectra (data campaigns Gironde2012 and Gironde2013, Fig. 5c and
d) have extremely high values, up to 0.2 around 700
To compare the reflectance spectra from the three estuarine sites, all
reflectance spectra with a TSM concentration around 100
Water reflectance spectra with a TSM concentration of
Statistics for the ASD reflectance for the reflectance
spectra shown in Fig. 6 with a TSM concentration of
For all stations where both TRIOS and ASD reflectance is available, the
correlation is shown in Fig. 7 for six spectral bands. The best correlation
is obtained for the shortest wavelengths (412, 490 and 555
ASD water reflectance versus TRIOS water reflectance
The mean, minimum, maximum and standard deviation of the TSM concentrations
and turbidity measurements are shown in Table 4 and Fig. 8. Based on the
triplicates, the accuracy was excellent (
TSM concentrations measured over the three sites range from
48
TSM and turbidity.
Boxplot for TSM for all sites.
A few water samples from the Gironde and Scheldt campaigns were analyzed for
TSM at both the VITO and LOV laboratories. Figure 9 shows the relationship
between both. The correlation is high (
Intercomparison of TSM measured by the different laboratories (VITO and LOV).
The TSM–turbidity relationship for all sites is linear, with a correlation
coefficient of 0.96 (Fig. 10a). The outlier observed at 456 FNU and
176
Intercomparison TSM and turbidity
The SeaSWIR dataset has been made publicly available (
The SeaSWIR dataset presents the first hyperspectral measurements of the water reflectance in the SWIR. Measuring water reflectance in the SWIR was made possible by using an ASD spectrometer with one radiometer and a measurement protocol where the instrument is turned facing the sky, the water and the Spectralon panel. Using this measurement protocol and consecutive post-processing it was shown, by intercomparison with TRIOS measurements, that high-quality water reflectance measurements can be made with the ASD. It is however suggested to strictly follow the same procedure as outlined in this paper and apply a strict quality control. Measurements which passed quality control were all performed under relatively constant atmospheric conditions, being blue sky or completely overcast. Variable light conditions prohibit a correct retrieval of the water reflectance as lighting conditions change while turning the instrument and making the three individual measurements of downwelling irradiance, the total upwelling radiance and downwelling sky radiance. It should also be noted that most measurements were made from fixed pontoons, which simplifies the measurement significantly and improves the quality of the data. Still, even when measuring on a pontoon, the location on the pontoon should be chosen wisely and should not be close to any large obstacles which could alter the measurements of the downwelling irradiance and downwelling sky radiance.
The quality of this dataset opens up the possibility of including the ASD in future measurement campaigns on water, for the development of algorithms as well as for calibration and validation of satellite and airborne observations. This is important for current (e.g., Sentinel-3 OLCI) and future (e.g., Environmental Mapping and Analysis Program – EnMAP, Hyperspectral Precursor and Application Mission – PRISMA and Sentinel-10) satellite missions with more spectral bands in the SWIR-I and SWIR-II.
Next to the water reflectance measurements, the SeaSWIR dataset also contains
simultaneous TSM and turbidity measurements (using a HACH portable turbidity
meter). The high quality of these measurements was confirmed by the standard
deviation of the triplicates and the duplicate measurements performed in the
LOV and VITO laboratories. As all measurements were made in estuaries during
the tidal cycle, the TSM range of the SeaSWIR dataset is very high, ranging
from 48
EK, KR, AD and DD designed all the experiments. EK and SS defined the ASD measurement procedure. All the authors contributed to the field campaigns.
The authors declare that they have no conflict of interest.
The research leading to these results has received funding from the Belgian Science Policy Office through the STEREO program (SeaSWIR project). Edited by: David Carlson Reviewed by: Ken Voss and one anonymous referee