ISSN 1024-8560, Atmospheric and Oceanic Optics, 2018, Vol. 31, No. 3, pp. 225–231. © Pleiades Publishing, Ltd., 2018. Original Russian Text © V.V. Rostovtseva, I.V. Goncharenko, B.V. Konovalov, A.F. Alukaeva, 2017, published in Optika Atmosfery i Okeana.
OPTICAL WAVES PROPAGATION
Rapid Estimation of the Ecological State of Coastal Water Areas Based on Shipboard Passive Remote Optical Sensing of the Water Surface V. V. Rostovtsevaa, *, I. V. Goncharenkoa, **, B. V. Konovalova, and A. F. Alukaevaa a
Shirshov Institute of Oceanology, Russian Academy of Sciences, Moscow, 117997 Russia *e-mail:
[email protected] **e-mail:
[email protected] Received July 10, 2017
Abstract—The study of the ecological state of coastal areas often requires rapid acquisition of detailed data. In this work, sea radiance coefficient spectra are analyzed, which were measured with a new three-channel passive optical complex for ecological monitoring of marine aquatoria (EMMA), developed by us, which operates semiautomatically on board a moving ship. The measurements were carried out near the Brazilian coast at the Rio Grande river mouth. The spectra are processed using an original technique which is based on intrinsic properties of the pure sea water absorption spectrum (WASM – water absorption step method) and has been modified for eutrophic Case II waters. This enables us to estimate the absorption indices of the suspended matter and colored organic matter. The comparison of these remote estimates with the estimates made from water samples taken at special stations shows their high correlation. The distributions of the suspended matter and colored organic matter are retrieved from remote measurement data and compared to the satellite imagery for the water area under study. Keywords: ecological state of marine areas, spectra of the sea radiance coefficient, absorption and scattering of light by sea water, coastal water area at the river mouth, concentration of suspended matter and colored organic matter DOI: 10.1134/S1024856018030132
INTRODUCTION Acquisition of data on the state of coastal water areas is an urgent practical task, because coastal regions play an important role in human life and activity. Properties of waters in these regions are very different and variable, since, in addition to interactions with open waters, they are affected by water inflow from the land and mixing with bottom deposits. The rapid estimation of the water state is the most effective when using methods based on the correlation between the water composition and its optical properties [1–5]. Recently, a large bulk of data has been provided by satellite-based optical instruments [6, 7]. However, their spatial and temporal resolutions are insufficient sometimes. Clouds usually impede the measurements. To interpret the satellite data, many sea-truth measurements are required. All this urgently requires the development of noncontact techniques for rapid estimation of the surface water composition onboard a moving ship. Passive optical techniques are among the most promising. They do not require high energy, have a good spectral resolution, and measure the same parameters as satellite instruments [8–10]. To receive
the sea radiance coefficient (SRC) onboard a ship, it is necessary to measure the brightness of upward radiation and the brightness of the sky area contributing the most to the water-surface-reflected signal and to estimate the total irradiation of the sea surface. The measurement results of a single-channel spectrometer, which sequentially measured all the three signals, were described in [11]. In this work, we present data which have been received with a new semiautomatic threechannel device designed by us. Three-channel spectrometers are produced by leading world companies for estimation of the state of coastal waters [12]. For example, a handheld Water Insight WISP-3 spectrometer is intended for single measurements of the SRC by an operator; its weight is >2 kg, the spectral resolution is ~5 nm in the 350– 800 nm range. A TriOS Ramses radiometer is mounted onboard a ship; it includes three TriOS Optical sensors each of ∼0.9 kg in weight, 0.5 m long, with 256 spectral channels; the data measured are transmitted into an onboard computer via a RS-232 serial interface. Another example of a stationary device is a Sea-Bird Scientific (Satlantic LP) Surface Acquisition System; it includes three spectrometers ~40 cm long
225
226
ROSTOVTSEVA et al.
700 g and size 30 × 20 × 20 cm (Fig. 1). The specifications of the spectrometers are the following: ⎯operation range ⎯spectral resolution ⎯signal-to-noise ratio ⎯array resolution ⎯integration time
Fig. 1. Small-size semiautomatic EMMA complex for passive remote optical measurements onboard a moving ship.
and 1 kg in weight, which measure an optical signal in the 350–800 nm range. The monitoring complex we designed is lighter and allows semiautomatic measurements (dark current, integration time, and device orientation relative to the sun direction are manually specified before a measurement cycle). The spectra of the water reflectance coefficient Rrs remotely measured both with our three-channel SRC spectrometer and these foreign devices strongly depend on the observation conditions. (The coefficient Rrs differs from the SRC by the dimension (sr–1) and a constant factor.) To overcome this disadvantage, it is suggested either to measure in near-ideal conditions (large solar elevation angle, quiet weather, and cloudless sly), which is often impossible, especially in polar latitudes, or to exclude the measurements that have been received at nonhorizontal orientation of the device platform or other specific conditions (up to 60% of data are sometimes lost in this case). We, instead, use a new algorithm, which we have developed for adjusting the spectra by data on light absorption and scattering by pure sea water (WASM – water absorption step method) [13]. In this work, we modify the algorithm for processing waters rich in colored organic matter. IN SITU MEASUREMENT INSTRUMENTATION For the rapid estimation of the state of coastal sea waters onboard a moving ship, we created a complex for ecological monitoring of marine aquatoria (EMMA), where passive optical measurements are carried out by three small portable STS spectrometers (Ocean Optics Company) with a total weight of about
350–800 nm 3 nm 1500 : 1 1024 points 10–6–10 s.
The sea and sky brightnesses are measured in two channels, where the radiation comes via an optical fiber with a collimating lens (field-of-view is about 5°). The viewing angle in the sea channel is selected so as to avoid observation of foam formations of the moving ship (40°–45° with the nadir direction); the viewing angle in the sky channel is equal to that in the sea channel and is measured with the zenith direction. An optical fiber with a cosine corrector is used in the third channel, which allows estimation of the total irradiation of a horizontal surface. To calculate the SRC, the irradiation is converted into the brightness of a white horizontal reflector using a factor found during the device calibration. The measurements in each EMMA channel are carried out semiautomatically with a frequency of 1 Hz. The results are stored in the database of a controlling computer, each second in the form of spectra received from each of the three spectrometers plus GPS/GLONASS coordinates. EMMA operates continuously full time. The shipboard measurements make it possible to receive geoattached arrays of measured parameters with a good resolution (~3 m). The complex is computer-controllable; this allows realtime visualization of the measurements performed and decision-making about a change in the ship trajectory, e.g., for plume profiling. Weight−size parameters and portability of the complex allow its use at all ship types, including small ships, which is useful in the solution of problems of ecological monitoring of coastal water areas. SRC is assessed by three measured spectra:
R(λ) = ( Bsea (λ) – rBsky (λ)) Bws (λ),
(1)
where Bsea is the upward radiation brightness; Bsky is the brightness of adjacent sky area; r is the Fresnel reflection coefficient equal to 0.02 during the quiet weather at sounding angles close to vertical; Bws is the brightness of the white reflector calculated by the total irradiation of the sea surface. Since we deal with shipboard sounding of the sea surface, the light scattering and absorption in air can be neglected. Then, the spectra found during SRC measurements are processed using special WASM-based software with the aim of assessing the suspended matter and colored organic matter concentrations in a water area under study.
ATMOSPHERIC AND OCEANIC OPTICS
Vol. 31
No. 3
2018
RAPID ESTIMATION OF THE ECOLOGICAL STATE OF COASTAL WATER AREAS
To verify the assessments, water samples were taken along the ship route and then processed at a 2100Q turbidimeter (Hach Company, Germany). The measurements are performed according to the EPA Method 180.1 protocol in NTU (nephelometric turbidity units), which are nephelometric kaolin turbidity units, i.e., conventional units based on the effect of light scattering in a light beam by small suspended particles. Following the above protocol, the water samples were kept in a thermostat and their turbidity was calculated as the content of standard suspended particles, in mg per 1 L of water [14]. In addition, the samples were filtered, and the concentration of water admixtures was assessed by light absorption spectra [15].
0.05 0.04 0.03 0.02 0.01 0 400
450
500
–1
SRC, sr
550 (b)
600
0.09
650
700
1
0.07
In this work, we use the original technique for processing SRC spectra measured on board a ship with the three-channel EMMA complex, which has been modified for eutrophic waters. The technique takes into account peculiarities of the spectra of light absorption by pure water and makes it possible to retrieve sea water absorption spectra and concentrations of its main components from the SRC spectra measured. The need for modification of the technique is dictated by the following considerations. During the development of the technique for mesotrophic waters [13], it was assumed that the main absorption of light in the wavelength range >580 nm is caused by the absorption by pure, i.e., free of admixtures, sea water and nonselective absorption by suspension. Figure 2a shows SRC spectra measured in the Black Sea water area with oligotrophic and mesotrophic waters. All these spectra have a common feature in the form of a step in the 550–650 nm range. During the calculations at the first stage, this step was considered caused exclusively by pure water. During the next stage of WASM algorithm, it was suggested to assess the total absorption by colored organic matter iteratively from SRC values in the short wavelength region. However, the absorption by this water component was as weak as it did not affect the general behavior of the spectra. The situation changes in eutrophic waters. Let us consider SRC spectra at the Rio Grande river mouth near the Brazilian coast (Fig. 2b). It is seen that the spectra in eutrophic waters differ in shape. The step becomes flatter and then completely disappears in more turbid waters (spectra 1 and 3–5). This means that the absorption by colored organic matter should be considered during the first stage of the algorithm. This parameter can be estimated in eutrophic waters from the following assumptions. In contrast to transparent waters, radiation scattered by the water column mainly contributes to the Vol. 31
(a)
SRC, sr–1 0.06
MODIFICATION OF WASM ALGORITHM FOR EUTROPHIC WATERS
ATMOSPHERIC AND OCEANIC OPTICS
227
3 2
0.05
4
5 6
0.03
7
0.01 400
450
550 600 500 Wavelength, nm
650
700
Fig. 2. SRC spectra measured with EMMA complex in (a) oligotrophic and mesotrophic waters in the Black Sea (Gulf of Feodosia) and (b) in eutrophic waters near the Brazilian coast (Rio Grande river mouth).
upward radiation intensity in eutrophic waters, and the part of the surface-reflected radiation is small (no more than 10–15%). Therefore, the SRC error, caused by the selection of the sky area which mainly contributes to the reflected signal, is small. In this case the true SRC value different from the measured value can be calculated by the equation (2) ρ(λ) = kR(λ), where k is the unknown factor, which differs from unity during an asynchronous jump of the total irradiation or of a signal from the surface due to sea waves. We take into account that
k0bb (3) a(λ) + bb in the two-stream approximation (bb is the backscattering coefficient of the water column, nonselective since the molecular scattering in eutrophic waters is small as compared to the scattering by suspension; k0 is the constant determined by the light transmission through the water–air interface; a(λ) is the water absorption coefficient). We assume that the absorption coefficient in eutrophic waters at λ > 500 nm is determined by the absorption by pure water aw(λ), nonselective absorption by suspension asm, and
No. 3
ρ(λ) =
2018
ROSTOVTSEVA et al.
1.5 Estimates on filters, m–1
Absorption index, m–1
7 6 5 1
4 3
3
4
2
5
1
2
6 W
18 2
R = 0.88 2
R = 0.96 0.5
7
450
500 550 600 Wavelength, nm
650
absorption by colored organic matter acom. The last one consists of the absorption by phytoplankton pigments, detritus, and “yellow substance”; its spectral dependence is similar to the yellow substance in this range (g = 0.015 nm–1):
a(λ) = aw (λ) + acom500 exp {– g(λ – 500)} + asm . (4) Then, choosing three wavelengths that characterize the step (e.g., 540, 580, and 602 nm) and enumerating them from short to long waves, the set of three equations with three unknowns is derived from Eqs. (1)–(4):
⎧ 1 = k R1 ⎪a – Δ + (a + b ) + a k0bb 21 sm b com500e1 ⎪ w2 ⎪ 1 k (5) = R2 ⎨ ⎪aw2 + (asm + bb ) + acom500e2 k0bb 1 ⎪ = k R3, ⎪⎩a + Δ + (a + b ) + a k0bb w2 32 sm b com500e3 where Ri is SRC measured at the wavelength λi; Δ21 = aw2 – aw1, Δ32 = aw3 – aw2 is the magnitude of Table 1. Calculation results of the sum of absorption and scattering indices of suspension (asm + bb) and the absorption index of colored organic matter at a wavelength of 500 nm (acom500) Absorption index Station no. asm + bb, m–1 acom500, m–1 2.40 0.275 1.04 0.89 0.63 0.23 0.14
2.66 0.42 1.42 1.19 0.82 0.39 0.26
6
R2 = 0.80
0.5 1.0 1.5 2.0 2.5 3.0 Estimates from remote measurements, m–1
700
Fig. 3. Spectra of absorption indices of admixtures calculated with the use of the WASM algorithm from SRCs measured, as well as the spectrum of light absorption index of pure water (curve W).
1 2 3 4 5 6 7
10
2 0
0 400
14
1.0
Ksm(estimates in samples), mg/L
228
Fig. 4. Comparison between estimates of the suspension concentration (asm + bb) from EMMA measurements, estimates of absorption by suspension on filters at λ = 750 nm (Δ), and data of tirbidimeter in water samples (r), as well as between estimates of the absorption index of colored organic matter (acom500) calculated from remote measurements and estimates of absorption by phytoplankton found from the difference in absorption on filters at λ = 680 and 750 nm (h).
the step stipulated by pure water absorption; ei = exp {– g(λi – 500)} . Solving the set of equations (5), we estimate the coefficient of absorption by colored organic matter at a wavelength of 500 nm (acom500) and the sum of indices of absorption and scattering by the suspension (asm + bb), as well as the coefficient k/k0bb, which can be used for calculation of the spectra of absorption indices of admixtures and the water (Fig. 3). It is seen that the absorption by admixtures is several times stronger that the light absorption by pure water at the Rio Grande river mouth, and the spectrum exponentially decreases in the longwave region as expected. In addition, a characteristic step in SRC spectra appears only in waters where the increase in the water absorption spectrum exceeds the decrease in the absorption in the spectrum of admixtures in the 580–600 nm range. Table 1 includes the calculated absorption index of colored organic matter and the sum of absorption and scattering indices by suspension for SRC spectra shown in Fig. 2b. It should be noted that for Figs. 2 and 3 we chose the spectra calculated from measurements made at the time of water sampling at the stations (the number of stations coincide with those in Fig. 5) were selected for Figs. 2 and 3. To verify the estimates made from shipboard remote measurements, the surface water samples were taken at the same time. The samples were analyzed with a turbidimeter adjusted to the suspension concentration Ksm (mg/L). Then, the samples were filtered and the light absorption spectra on the filters were measured. The concentration of mineral suspension was estimated as the absorption at 750 nm in absorption units (m–1); the
ATMOSPHERIC AND OCEANIC OPTICS
Vol. 31
No. 3
2018
RAPID ESTIMATION OF THE ECOLOGICAL STATE OF COASTAL WATER AREAS
229
(а) N W
7
4 5
11
5
13 5 11. .5 10
S
3.5 5.5
10
1 16.5
E
5
6
7.5
8
9.5 9
7.5
7
3
6.
7
77
2
6
5 5.
−32.20
8.5
8 8.5 1
0
1
2
3
4
4 km
5
−52.10
−52.00 (b) N W
E S
7
0.8
0.6
1
0.8
1.6
0.6
1.8
1
2
0.4
2.2
−32.20
2
1 2.6
6
1.4
3
1.2
4 1
0
1
2
3
5
4 km
−52.10
−52.00
Fig. 5. Distributions of (a) suspension concentration, mg/L, and (b) colored organic matter, m–1, retrieved from remote EMMA measurements on board a ship moving along route 1–7. Dots show the water sampling stations. The distributions are superimposed on a hyperspectral image taken on the same day from the Russian Resurs-P satellite near the Brazilian coast in the region of Rio Grande river mouth.
concentrations of phytoplankton pigments and detritus were estimated as the difference between the total absorption at 680 nm (red maximum in the chlorophyll a absorption) and suspension absorption. The results of comparison between these estimates and the remote measurements are shown in Fig. 4. The estimates of suspension concentration from remote measurements correlate well with estimates of absorption on the filters, being higher, since not only the absorption on the water surface, but also the layer ATMOSPHERIC AND OCEANIC OPTICS
Vol. 31
average absorption are estimated remotely. The turbidimeter measurements evidently also depend on both the light absorption and scattering; therefore, the correlation with the estimates from the remote measurements is even stronger. In addition, the absorption coefficients of colored organic matter derived from EMMA measurements have been compared with the absorption on the filters at λ = 680 nm, which corresponds to the red peak in the chlorophyll a absorption. The correlation is also strong in this case.
No. 3
2018
230
ROSTOVTSEVA et al.
DISCUSSION OF RESULTS Strong correlation between the estimates of suspension concentration derived from EMMA remote measurements and from processing of water samples by two common techniques allows a conclusion about the high efficiency of the new complex and processing techniques applied to the measurement results. To date, we have suggested and tested two versions of the technique based on the peculiarities of light absorption by pure sea water. When processing the SRC measurement results for oligotropic and mesotrophic waters in [13], we considered the absorption in the red spectral region to be mainly caused by pure water and gray suspension during the first stage, and only then estimated the absorption index of colored organic matter using one or several iterations. This approach turns out to be inapplicable to eutrophic waters, since the absorption by colored organic matter strongly affects the shape of SRC spectra in this water type. Therefore, in this work we have modified WASM: first, the absorption index of colored organic matter is estimated from SRC values at three wavelengths in the red spectral region, and the index of absorption by suspension is derived from it. This is possible since the upward radiation from the sea, mainly caused by backscattering on suspension, is much brighter than the surface-reflected sky light in eutrophic waters. In these conditions, one may ignore the error which shifts the SRC spectrum caused by measurements of brightness of the sky area adjacent to the required one. The absorption indices of colored organic matter and suspension measured allow calculation of the absorption spectra of waters under study in the whole range of wavelengths used and estimation of the concentrations of phytoplankton pigments and yellow substance from them by the technique [11]. In this work, we do not present these calculations. However, we have found a quite strong correlation between the estimate of the colored organic matter concentration and the estimate of the phytoplankton pigment concentration by the red peak of the chlorophyll a absorption. This means that the concentration of dissolved yellow substance is proportional to the phytoplankton concentration in the waters near the Brazilian coast, like in open ocean waters. Based on data received with the EMMA complex we have plotted the concentrations of suspension and colored organic matter in Fig. 5. The ship route and water sampling stations are also shown. The suspension concentrations (mg/L) have been calibrated by seven stations. The distributions have been constructed by a large number (∼20000) of values of the suspension concentration and absorption index of colored organic matter using the common method of the natural neighbor. They are superimposed on the hyperspectral image which was taken on the same day from the Russian Resurs-P satellite. It is seen that both
distributions agree well with the satellite image though they have their own characteristic features. CONCLUSIONS The study in eutrophic waters near the Brazilian coast has shown the high efficiency of the new EMMA optical complex. The complex allows monitoring of coastal areas, which are characterized by large gradients of concentrations of natural admixtures and strong temporal variability, for example, regions of influx of large rivers into the sea. The complex provides a large volume of data on the content and distribution of different natural admixtures over a water area. Due to this, it can also be effectively used during sea-truth measurements. ACKNOWLEDGMENTS The work was supported by the Ministry of Education and Science of the Russian Federation (agreement 14.613.21.0050, identifier RFMEFI61315X0050). REFERENCES 1. A. Morel and L. Prieur, “Analysis of variations in ocean color,” Limnol. Oceanogr. 22 (4), 709–722 (1977). 2. V. N. Pelevin, “Optical phenomena on the ocean surface,” in Phenomena on the Ocean Surface, Ed. by A.S. Monin and V.P. Krasitskii (Gidrometeoizdat, Leningrad, 1985), p. 318–329 [in Russian]. 3. D. G. Bowers and C. E. Binding, “The optical properties of marine suspended particles: A review and synthesis,” Estuarine, Coastal Shelf Sci. 67, 219–230 (2006). 4. V.S. Shamanaev, “Airborne lidars of the IAO SB RAS for sensing of optically dense media,” Atmos. Ocean. Opt. 28 (4), 359–365 (2015). 5. G. P. Kokhanenko, Yu. S. Balin, I. E. Penner, and V. S. Shamanaev, “Lidar and in situ measurements of the optical parameters of water surface layers in Lake Baikal,” Atmos. Ocean. Opt. 24 (5), 478–486 (2011). 6. C. B. Mouw, S. Greb, D. Aurin, P. M. DiGiacomo, Z.-P. Lee, M. Twardowski, C. Binding, C. Hu, R. Ma, T. Moore, W. Moses, and S. E. Craig, “Aquatic color radiometry remote sensing of coastal and inland waters: Challenges and recommendations for future satellite missions,” Remote Sens. Environ. 160, 15–30 (2015). 7. V. I. Burenkov, O. V Kopelevich, T. N. Rat’kova, and S. V. Sheberstov, “Satellite observations of the coccolithophorid bloom in the Barents Sea,” Oceanology 51 (5), 766–774 (2011). 8. V. A. Matyushenko, V. N. Pelevin, and V. V. Rostovtseva, “Measurement of the sea radiance coefficient with a three–channel spectrophotometer from aboard a research ship,” Atmos. Ocean. Opt. 9 (5), 421–424 (1996).
ATMOSPHERIC AND OCEANIC OPTICS
Vol. 31
No. 3
2018
RAPID ESTIMATION OF THE ECOLOGICAL STATE OF COASTAL WATER AREAS 9. C. D. Mobley, “Estimation of the remote-sensing reflectance from above-surface measurements,” Appl. Opt. 38 (36), 7442–7455 (1999). 10. S. C. J. Palmer, T. Kutser, and P. D. Hunter, “Remote sensing of inland waters: Challenges, progress and future directions,” Remote Sens. Environ. 157, 1–8 (2015). 11. V. V. Rostovtseva, B. V. Konovalov, I. V. Goncharenko, and D. V. Khlebnikov, “Method for estimating admixture content in seawater using operative spectrophotometry,” Oceanology 57 (4), 505–519 (2017). 12. A. Hommersom, S. Kratzer, M. Laanen, I. Ansko, M. Ligi, M. Bresciani, C. Giardino, J. M. BeltranAbaunza, G. Moore, M. Wernand, and S. Peterset, “Intercomparison in the field between the new WISP-3 and other radiometers (TriOS Ramses, ASD FieldSpec,
ATMOSPHERIC AND OCEANIC OPTICS
Vol. 31
231
and TACCS),” J. Appl. Remote Sens. 6 (1), 063615–1 (2012). 13. V. V. Rostovtseva, “Method for sea water absorption spectra estimation on the basis of shipboard passive remote sensing data and pure sea water properties,” Atmos. Ocean. Opt. 29 (2), 162–170 (2016). 14. www.epa.gov/sites/production/files/201508/documents/ method_180-1_1993.pdf. Cited May 17, 2017. 15. B. V. Konovalov, M. D. Kravchishina, N. A. Belyaev, and A. N. Novigatskii, “Determination of the concentration of mineral particles and suspended organic substance based on their spectral absorption,” Okeanologiya 54 (5), 660–667 (2014).
No. 3
Translated by O. Ponomareva
2018
