- Email: [email protected]
Analysis of estuarine colour components during non-monsoon period through Ocean Colour Monitor
Estuarine, Coastal and Shelf Science 66 (2006) 523e531 www.elsevier.com/locate/ecss
Analysis of estuarine colour components during non-monsoon period through Ocean Colour Monitor H.B. Menon a,*, A.A. Lotliker a, S.R. Nayak b b
a Department of Marine Science, Goa University, University P.O., 403 206, Goa State, India Marine Water and Resource Division, Space Application Center, Indian Space Research Organization, Ahmedabad 380 015, India
Received 6 July 2005; accepted 21 October 2005 Available online 13 December 2005
Abstract Simultaneous acquisition of water samples, radiance and irradiance measurements were carried out from 40 stations in the MandovieZuari estuaries during February to May 2002. From the samples collected, inherent and apparent optical properties (IOP and AOP) such as absorption coefficient (a), upwelling diffuse attenuation coefficient (ku) and subsurface reflectance (R) were derived. Using these optical properties, radiative transfer at each water column is examined. On the basis of the radiative transfer outcome, band-ratio algorithms are derived for three optically active substances (OAS), viz, chlorophyll-a, suspended sediment and coloured dissolved organic matter (CDOM). The respective algorithms are 670/555, 490/670 and 412/670 nm for chlorophyll-a, suspended sediment and CDOM. These algorithms are applied to Ocean Colour Monitor (OCM), onboard Indian Remote Sensing Satellite (IRS)-Polar Satellite Launch Vehicle (P4), scenes (digital data), to synoptically analyze these OAS. The synoptic analysis of OAS revealed different hydrodynamic characteristics of the estuaries during non-monsoon seasons. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: algorithm; IRS-P4-OCM; band ratio; CDOM; chlorophyll-a; suspended sediment; hydrodynamics; non-monsoon
1. Introduction Estuarine environment is characterized by dynamic water bodies of different temporal changes that occur due to hourly (tide) to seasonal variations (river discharge). This spatial heterogeneity of the water column is further augmented by change in the depth and width of the estuary. To monitor such changes of varying time scales due to tidal advection, freshwater influx, dispersion and resuspension, synoptic analysis is imperative. This could be accomplished through optical remote sensing. The ocean colour monitoring for a long period is an important tool to decipher the biophysical processes, ecology and coastal environmental changes (Tang et al., 2002; Babin et al., 2003; Darecki et al., 2003; Menon, 2004; Menon et al., 2005). On the basis of ocean colour constituents, waters are divided into case I and case II (Morel and Prieur, 1977). Case I refers to
* Corresponding author. E-mail address: [email protected] (H.B. Menon). 0272-7714/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ecss.2005.10.009
waters for which phytoplankton species and their accessory pigments play a dominant role in determining the colour while case II waters are those for which the colour is determined by suspended inorganic and dissolved organic matters along with phytoplankton and their accessory pigments. Adopting this criterion the study area falls under case II category. The general estuarine dynamics is exemplified by the need for better understanding of mechanism that control the water quality of the area of study, MandovieZuari estuarine system. The system is under strong anthropogenic influence due to transportation of iron ore, as a part of mining industry, through these estuaries. Hence the seepage of iron ore from the barges, carrying ore, completely alters the water quality and the photic zone. This is being augmented by the sediment flux from mining dumpsites through seasonal rain. On the basis of foraminiferal assemblages, Rao and Rao (1974) made an attempt to analyze the ecology of Mandovi and Zuari estuaries. Efforts had been made to study the variation of chlorophyll-a distribution in Zuari estuary (Bhargava and Dwivedi, 1974) and Mandovi estuary (Dalal and Goswami, 2001), circulation and mixing in the
524
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
estuaries (Murthy and Das, 1972; Verma et al., 1975; Qasim and Sen Gupta, 1981; Unnikrishnan et al., 1999) and sediment distribution in the Zuari estuary (Nayak and Bukhari, 1992). The results revealed that sediment distribution in Zuari estuary during non-monsoon period is between 15 and 22 mg l1 while chlorophyll-a is in the range 1.0e5.0 mg l1. No study is reported about the distribution of coloured dissolved organic matter (CDOM) in these estuaries. Hence it is clearly evident from the above references that little attention is paid in analyzing the water quality of estuaries either through in situ observation or using high-resolution synoptic data during the recent past. Therefore, an attempt is made to retrieve ocean colour components from the estuarine system through optical remote sensing, as this may be a valuable tool for monitoring the long-term ecological consequences due to anthropogenic activities in the estuarine system. This has led the authors to carry out the present study with the following objectives: 1. Examine the feasibility of Ocean Colour Monitor (OCM) data to map the spatial distribution of three optically active substances (OAS) viz, chlorophyll-a, CDOM and suspended sediments through suitable algorithms to understand the water quality. 2. Understand the hydrodynamics of the estuarine system through ocean colour component analysis.
2. Study area The area for the present study is Mandovi and Zuari, two estuaries, situated in Goa, along the west coast of India. These two estuaries are interconnected by a narrow canal, the Cumbarjua (Fig. 1). At the upstream ends, they receive large freshwater influx during southwest monsoon (JuneeSeptember) and little during the rest of the year. The average annual rainfall of Goa is about 3000 mm of which nearly 80% occurs during the southwest monsoon period (Qasim and Sen Gupta, 1981). Hence the estuaries are freshwater dominated during JuneeSeptember and seawater dominated during the rest of the year. Therefore, tropical southwest monsoon, riverine and seawater flows make the estuaries an ecologically complex ecosystem. 3. Method The waters in the estuarine system being highly optically complex, retrieval of chlorophyll-a, suspended sediment and CDOM through OCM is possible only by applying site specific algorithms developed through the analysis of hyperspectral water leaving radiance in the optical range (400e700 nm) of Electromagnetic Radiation (EMR). The hyperspectral water leaving radiance can be generated either through an in-water radiometer having hyperspectral bands or through radiative
Fig. 1. Study area showing the different station positions (stations 1e10). Dotted lines indicate depth contours.
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
transfer analysis. In the present case the second approach is adopted. To carry out the radiative transfer analysis, the inherent optical properties (IOP) and apparent optical properties (AOP) of each water column, represented by different stations, are generated through water sample analysis. 3.1. In situ observations The in situ observations from the area of study involve water sampling, in-water radiometer measurements (using Satlantic radiometer), Eko Sunphotometer measurement for aerosol optical depth (AOD) observations and Secchi disk depth measurement for transparency of light. From each station, water sample was collected from a depth between sea surface and 1% of photosynthetically active radiations (PAR). Water sampling and radiometer observations were always from the area where the shadow of the boat did not fall. The observations were carried out during pre-monsoon period (FebruaryeMay) 2002. In each month, field survey was carried out from 10 stations (Fig. 1) between 1100 hours and 1400 hours (the satellite passes over the area around 1200 hours). Hence all the observations were within 3 h of satellite pass. As the estuarine width decreases drastically between mouth and head, positions of different stations were selected in such a way that the width of each station is about 1 km or more (three times the pixel size of OCM). This precaution was taken to prevent any overlapping of land radiance with water leaving radiance. Estuaries being a shallow water body, it is necessary to arrest any radiance from bottom of each station. This was done by selecting station where the depth of the water column is three times more than the Secchi disc depth of the corresponding station (Muller and Austin, 1995). 3.2. In-water radiometer measurement A Satlantic radiometer consisting of Profiling Multichannel Radiometer (SPMR) and Satlantic Multichannel Surface Reference (SMSR) was used to measure the upwelling radiance (Lu(l, 0)), downwelling irradiance (Ed(l, 0)) and downwelling irradiance reaching the sea surface (Es(l, 0þ)). The instrument operates in seven bands with wavelengths centered at 412, 443, 490, 510, 555, 670 and 780 nm. Due to ambiguity in the readings, the data obtained at 780 nm were not considered. The radiometer measurements were always taken in association with water sample collection. This precaution was taken to prevent any change in the computed water leaving radiance values from measured radiance values due to variation in solar zenith angle. While operating radiometer, utmost care was taken not to exceed the tilt of SPMR beyond 5 and that of SMSR 10 , a standard suggested by the manufacturer. 3.3. Inherent optical properties (IOP) and apparent optical properties (AOP) generated through water sample analysis Two water samples, 5 L, each were collected from each station using Niskin sampler. One sample for suspended particles
525
(chlorophyll-a and sediment) and the other for CDOM. In order to avoid the degradation of the samples, 1 ml of saturated MgCO3 was added to the sample for chlorophyll-a analysis (Strickland and Parson, 1972), whereas 0.4 ml of 1 M HgCl2 was added to the sample for analysis of CDOM (Kowalczuk and Kaczmarek, 1996). The samples were then preserved in dark under low temperature prior to laboratory analysis. As the details of the water sample analysis were given elsewhere (Menon et al., 2005), the same is not repeated here. As per Gordon (1978) subsurface reflectance R(l) is Rðl; 0 Þ ¼ Eu ðl; 0 Þ=Ed ðl; 0 Þ ¼ QLw ðl; 0 Þ=Ed ðl; 0 Þ ¼ QRrs ðl; 0 Þ
ð1Þ
where Eu(l) and Ed(l) are the upwelling and downwelling irradiances just below the sea surface, respectively, Lw(l) is the water leaving radiance and Q is the bi-directional reflectance (irradianceeradiance ratio) factor. Since Q is spectrally variant, radiometer observed values are used to calculate the Q factor. In order to derive hyperspectral water leaving radiance (Lw(l)), R(l) in Eq. (1) was derived as per the details given in Menon et al. (2005). 3.4. Satellite data processing The optical sensor, OCM onboard IRS-P4 has eight spectral bands. Out of which, six are in the visible spectrum of EMR and two are in the near infrared (NIR) region. Together they are in the range 402e885 nm with spectral resolution of 20 nm in the visible bands and 40 nm in NIR bands. The spatial resolution of the sensor is 360 m 250 m and the revisit cycle is 2 days (Kundu et al., 2001). The unsigned 16-bit data of OCM for the eight bands were imported using ERDAS IMAGINE 8.4. The images were then geo-referenced using ground control points (GCP) after which the area of study covering MandovieZuari estuarine system, Goa, India, bounded by 73 55#e74 5# E and 15 15#e15 25# N was extracted from the full scene. 3.5. Atmospheric correction Optical sensors, in general, have two NIR bands for applying atmospheric correction to visible bands to retrieve OAS. The concept is that radiance received by NIR bands are only from aerosols irrespective of the water column being highly turbid or not. But studies carried out by Yan et al. (2002) revealed that NIR bands of Sea WiFS (765 and 865 nm) had water leaving radiance component when the water column was turbid. Hence, in the present study, aerosol radiance is computed by deriving AOD through sunphotometer (Chylek et al., 2003). An Eko Sunphotometer having filters 368, 500, 675 and 862 nm was used to derive AOD. The transmission loss due to atmospheric gases is incorporated and Rayleigh radiance is computed (Doerffer, 1992). Thus the water leaving radiances from the selected bands of OCM are generated by subtracting Rayleigh and aerosol radiances.
526
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
3.6. OCM derived chlorophyll-a, sediment and CDOM There are many algorithms to retrieve case II water constituents. These are inverse modeling approach (Doerffer and Fisher, 1994), band-ratio method (O’Reilly et al., 1998) and Principal Component Analysis (PCA). In the present case the second approach is adopted. The basis in choosing wavelengths for band-ratio algorithm is differential absorption and scattering of solar flux associated with each wavelength in the visible spectrum of EMR, due to inherent characteristics of each OAS. For example, in the open ocean water optically active substance is mainly chlorophyll-a. Hence the interaction of visible spectra of EMR with this water is such that there is always a linear relation between ratio of water leaving radiance received by blue and green bands of OCM and chlorophyll-a concentration. But from the analysis of IOP due to chlorophyll-a, sediment and CDOM, it is realized that interaction of optical spectra in estuarine waters is non-linear and all the OAS utilize energy associated between 400 and 500 nm ranges. Hence to distinguish these OAS through OCM, hyperspectral water leaving radiance (for every 1 nm from 400 to 700 nm) is computed through a radiative transfer model and analyzed. The details about the calibrated radiative transfer modeling and some of the results including hyperspectral water leaving radiance are given in Menon (2004) and Menon et al. (2005). The procedure in identifying the wavelengths to develop band-ratio algorithm to retrieve OAS from estuarine waters through OCM is given below. The first step is to select a station with minimum concentration of OAS so that spectral radiance leaving this station can be the baseline spectrum. Hence water leaving radiance corresponding to station 7 sampled on 13 May 2002 is taken as baseline spectrum. Subsequently, water leaving radiances are computed for different concentrations of chlorophyll-a within the acceptable range prevailing in the estuarine waters during pre-monsoon season. While doing this, it is assumed that concentration of sediment and CDOM (ay (440)) are not varying. Then each hyperspectral radiance is divided by the baseline spectrum. This has resulted in the identification of wavelength having the least and maximum absorption by chlorophyll-a in the optical spectrum. The same procedure is repeated in the case of sediment and ay (440). Thus the analysis revealed that wavelengths suitable to retrieve chlorophyll-a are 555 and 670 nm, for sediment 490 and 670 nm while for CDOM 412 and 670 nm, respectively. The band-ratio algorithms are shown in Fig. 2a. From the graph it appears that as chlorophyll-a and sediment increase, the power correlation between the respective band ratios and concentrations approaches some constant value. Similarly in the case of CDOM, there is an exponential fit between ay (440) and ratio of water leaving radiance at 412 and 670 nm and the ratio approaches a constant. Therefore, significant inference drawn, from the analysis, is to retrieve OAS with concentrations more than the present case, different algorithms have to be developed. That means the algorithms developed are site and season specific. In all the cases R2 is more than 0.95. Therefore, by using these chosen wavelengths, bandratio algorithm is applied to OCM data and different optically
Fig. 2. (a) Correlation between (1) chlorophyll-a concentration and the ratio of water leaving radiance at 670 (Lw 670) and 555 nm (Lw 555) bands; (2) sediment concentration and the ratio of water leaving radiance at 490 (Lw 490) and 670 nm (Lw 670) bands and (3) absorption coefficient due to CDOM at 440 nm (ay (440)) and the ratio of water leaving radiance at 412 (Lw 412) and 555 nm (Lw 670) bands. Synoptic distribution of (1) chlorophyll-a, (2) suspended sediment and (3) ay (440) on (b) 8 January 2003, (c) 14 January 2003, (d) 14 February 2002 and (e) 24 May 2003.
active constituents are retrieved. The results are shown in Fig. 2bee. The respective algorithms are the following: , where R1 ¼ Lw 670/Lw 555; 1. Chlorophyll-a ¼ 5.5931R0.615 1 2. ay (440) ¼ 2.9393R2.2486 , where R2 ¼ Lw 412/Lw 670; 2 3. Suspended sediment ¼ 17.11R0.3596 , where R3 ¼ 3 Lw 490/Lw 670.
4. Results and discussion Using the algorithms developed, four satellite scenes (satellite data) are processed. These are 8 January 2003 (Fig. 2b), 14 January 2003 (Fig. 2c), 14 February 2002 (Fig. 2d), and 24 May 2003 (Fig. 2e). These scenes are chosen to show the ability of the algorithms in retrieving OAS during different hydrodynamics of the estuary prevailing in different non-monsoon
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
527
Fig. 2 (continued).
periods. The studies on the hydrography and circulation made by Murthy and Das (1972), Verma et al. (1975) and Qasim and Sen Gupta (1981) revealed that estuaries behaved like a mixed/homogeneous one during pre-monsoon, partially mixed during post-monsoon and saltwedge type during monsoon. Hence they are tidally dominated during pre-monsoon, tide and freshwater influx control them during monsoon and post-monsoon seasons. Therefore, January data are chosen as a representative of the estuarine water quality during the final stages of post-monsoon, February to represent the initial stages of pre-monsoon and May the final stages of pre-monsoon season. Another reason for choosing these scenes is to show the ability of the algorithms in retrieving OAS during non-monsoon seasons of any year in addition to the period of in situ observations used for algorithm development. In general, the OCM derived products and in situ observations are in good agreement and are in the same range. A graphical representation of validation is given in Fig. 3. Though, good correlation is observed for all the constituents, an offset is seen in the case of chlorophyll-a and sediment. The offset is on the
axis where satellite derived values are taken. For validation, all the available data (chlorophyll-a and sediment) published so far, from the area, are considered. As some of the data are from the midstream of the estuary, where the width of the estuary is less than 1 km (the resolution of the sensor is 360 m 250 m), selecting corresponding pixel of the in situ observation from OCM imagery will result in an erroneous value as the water leaving radiance can be entangled by the radiance from land. Therefore, to correlate those in situ data, pixels are not considered corresponding to the station positions rather from the adjacent area where the width of the estuary is around 1 km (three times the resolution of the sensor) or more. This resulted in the offset. As there are no published data on CDOM from the area, these values are derived during separate field surveys. Hence no offset is seen in this case. The spatial distribution of chlorophyll-a, sediment and CDOM in Mandovi and Zuari estuaries derived through IRSP4-OCM during the above days are given in Fig. 2bee. In the estuarine waters chlorophyll-a is in the range 3.3e5.0 mg l1, sediment 14e22 mg l1 and ay (440) 1.4e2.2 m1 during
528
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
Fig. 2 (continued).
January. The distribution of these OAS are in the range 3.8e 5.0 mg l1, 18e22 mg l1 and 0.4e2.2 m1, respectively, during February whereas they are in the range 4.5e5.5 mg l1, 15 mg l1 and 0.8 m1, respectively, during May. The concentration of chlorophyll-a in the estuarine waters is always more than the coastal waters. Chlorophyll maxima are always in the upper reaches of estuarine waters irrespective of month. The maximum concentration, 5.0 mg l1, is during May. Further, an examination of distribution of chlorophyll-a revealed an anomalous pattern in Mandovi estuary whereas a gradual increase from downstream to upstream (towards the head) in Zuari estuary during January (Fig. 2b(1), c(1)) and February (Fig. 2d(1)). Barring this anomalous low, distribution of chlorophyll-a in both the estuaries during these months is in the range mentioned above. Similarly at the mouth and upstream of Mandovi, a plume of sediment is depicted during February while the distribution in January is near uniform with even distribution in May. Zuari estuary shows a different picture. Here, in the midstream, a well defined sediment plume is seen during January (Fig. 2b(2), c(2)) and February (Fig. 2d(2)) while it is distributed uniformly during May (Fig. 2e(2)). A noticeable feature seen
in both the estuaries is the high incidence of CDOM associated with the regions of sediment plumes. The respective concentrations of sediment and absorption due to CDOM at 440 nm during January and February are 19 mg l1 and 2.2 m1. The spatial distribution of CDOM shows least concentration in the coastal region with a gradual increase to midstream of Zuari estuary during January (Fig. 2b(3), c(3)), February (Fig. 2d(3)) and May (Fig. 2e(3)). In their studies Nayak and Bukhari (1992) observed a region of high sediment concentration (22 mg l1) in the midstream of Zuari estuary for which the observations were carried out in February. The reason for such a high concentration is ascribed to the drastic decrease in estuarine depth from mouth to upstream and the effect of tide. Zuari estuary is funnel shaped with a wide and deep mouth, which narrows and shallows down to upstream. The width at the mouth of the Zuari estuary is 7.0 km while at the midstream it is around 3.0 km. The shallowing is abrupt at the place where suspended sediment and CDOM concentrations are high. At this point the depth shallows down to 3 m from 10 m at the mouth. Also the time of in situ observation at this point and the satellite
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
Fig. 3. Correlation between satellite derived and in situ values of (a) chlorophyll-a, (b) suspended sediment, (c) absorption due to CDOM at 440 nm (ay (440)).
pass (1200 hours) coincided with high tide (1157 hours) during February. As a result the tidal energy associated with seawater inflow generated turbulence at the bottom and resuspension of sediment and CDOM. This had led to the formation of a zone of high concentration of sediment and CDOM in the midstream. This process is enhanced by softer nature of the sediments (Rao and Rao, 1974) and by convergence of currents at the constriction and increasing currents towards the narrow part of the estuary (Pundare, 1989). Similar to Zuari, Mandovi estuary is also funnel shaped with a broad mouth of 4.5 km and a very narrow upstream. In the estuary a sediment plume is observed at the mouth, upstream and further north along the coast during February (Fig. 2d(2)). The occurrence of sand bars, popularly called as Aguada and Reis Magos bar, near the entrance of the Mandovi is known for centuries. The important feature is the presence of the ramp-like structure in the navigation channel of the Mandovi with a depth gradient from 10 m to 4 m (Murthy et al., 1976; Qasim and Sen Gupta, 1981). Also the northwestern part of the mouth of Mandovi is exposed to swell from south,
529
southwest, northwest and north. Therefore, the shoal on both sides of the entry channel became the main sight of wave breaking. During the initial phase of pre-monsoon season (February) the waves at the entrance of the Mandovi estuary (northern portion) are small with a significant period of 68 s. Heavy swells and intensification of wave activity by several folds during the monsoon season led to highly turbulent conditions over shoals resulting in the removal of large quantities of material from the beaches and their transfer close to the navigational channel. During the post-monsoon period (January), these materials are returned and deposited on either side of the entrance channel and resulting in a uniform sediment pattern. During the pre-monsoon season (February) substantial amount of sediment are added from the coastal region to estuary in association with swell of west and northwest (Murthy et al., 1976). Therefore, high concentrations of sediment seen at the head of the estuary and along the coast north of the mouth of Mandovi during February and a uniform sediment distribution during January are the remnants of the sediment transported in the estuary during monsoon season. However, the sediment distribution in Zuari during January (Fig. 2b(2), c(2)) is due to different factors. Though the in situ observations were in conjunction with low tide, a clear plume of sediment and CDOM are seen at the midstream of Zuari and a near uniform distribution of sediment and a patchy distribution of CDOM in Mandovi (Fig. 2b(3), c(3)). It is an established fact that in shallow estuaries the momentum balance is primarily pressure gradient and friction. Friedrichs and Aubrey (1994) had shown that in such systems if the estuarine channel converges (if it shallows and narrows in the upstream direction) then amplification due to convergence of the channel cancels decay due to friction, leaving the amplitude unchanged over long distances along the estuarine channels. This explains the reason for the distribution in both the estuaries. In May (Fig. 2e) the scenario is different. Though the observation coincided with low tide, rather than a plume, the sediment and CDOM are evenly distributed throughout the estuaries. This could be attributed to the hydrodynamic characteristics of the estuaries. Both the estuaries are well mixed/homogenous during pre-monsoon (Unnikrishnan et al., 1999). The OCM data/scene selected in the study are of May, they represent the complete evolution of estuary into a homogenous one from partially mixed. The explanation is given below. Let R be the volume of freshwater being brought to the estuary by river and V the volume of seawater due to tide. During monsoon, the ratio of R to V is always greater than one and hence leads to a salt-wedge estuary. But during post-monsoon period though R decreases, V is the same. Hence the ratio decreases. This leads to the formation of a partially mixed estuary. During pre-monsoon R is negligible so that ratio may be very less. Hence turbulent mixing becomes so efficient that locally all salinity differences are eliminated and the estuary becomes vertically mixed. This is the reason for the typical distribution of sediment and CDOM during the final phase of pre-monsoon (Fig. 2e(2), e(3)). A similarity in the distribution pattern of CDOM with
530
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
that of sediment can be a pointer to the fact that the reasons attributed for the high concentration of sediment are responsible for high CDOM concentration too. In contrast to the distribution of sediment and CDOM, the distribution of chlorophyll-a in the Mandovi estuary shows a gradual increase towards the head during May (Fig. 2e(1)) while a scattered pattern with a patch of low concentration in the midstream and upstream during January (Fig. 2b(1), c(1)) and February (Fig. 2d(1)). Similarly a gradual increase in concentration of chlorophyll-a is seen from mouth towards the upstream of Zuari estuary during all the months. Though there is a variation in the chlorophyll-a distribution between mouth and head among the two estuaries, the chlorophyll-a concentration is maximum at the head of the estuaries during all months. Heavy precipitation of the order of 250e 300 cm yr1 and land runoff (Qasim and Sen Gupta, 1981) during southwest monsoon period (JuneeSeptember) bring huge quantities of nutrients making the estuaries highly productive (Devassy and Goes, 1989). Therefore, remnants of the nutrient rich water of the previous monsoon are responsible for the high chlorophyll-a and hence primary production at the head of the estuaries. Though chlorophyll-a concentration is maximum and almost of the same order of magnitude at the head of the estuaries during all the months, it is observed that the concentration at the lower stream of the estuaries increases from January to May. In their studies on the distribution of phytoplankton in Zuari estuary, Bhargava and Dwivedi (1974) observed a similar pattern. According to them the maxima in chlorophyll-a distribution generally coincided with high salinity. This could be attributed to the influence of neritic waters at the lower stream of the estuary. The neritic waters being chlorophyll rich phytoplankton, during flood (high tide), they dominate over the ebb tide crop of phytoplankton. Therefore, a tide controlled estuarine circulation led to the increase in the distribution of chlorophyll-a at the lower stream from January to May. 5. Conclusion The paper presents the procedure and development of local band-ratio algorithms for three major OAS (chlorophyll-a, suspended sediment and CDOM) in terms of IOP, AOP, in-water constituent measurements and retrieval of OAS through OCM. This is the first time that a highly turbid estuarine water column has been successfully analyzed optically through OCM data. The hydrodynamics prevailing in the estuarine system during different non-monsoon seasons could be well explained through mapping the colour components. Knowledge of distribution of water quality parameters and the location of zone of high biomass and plumes is valuable in establishing management policies for these ecologically important waters of Goa. Acknowledgement The first author HBM is thankful to Indian Space Research Organization (ISRO), Bangalore for the project sanctioned
‘‘algorithm development’’ under the National programme IRS-P4-OCM. The critical comments made by the unknown referees are highly acknowledged. References Babin, M., Stramski, D., Ferrari, G.M., Claustre, H., Bricaud, A., Obolensky, G., Hoepffner, N., 2003. Variations in the light absorption coefficients of phytoplankton, nonalgal particles and dissolved organic matter in coastal waters around Europe. Journal of Geophysical Research 108 (C7), 3211. Bhargava, R.M.S., Dwivedi, S.N., 1974. Diurnal variation in phytoplankton pigment in Zuari estuary, Goa. Indian Journal of Marine Science 3, 142e145. Chylek, P., Henderson, B., Mishchenko, M., 2003. Aerosol radiative forcing and the accuracy of satellite aerosol optical depth retrieval. Journal of Geophysical Research 108, 4764. Dalal, S.G., Goswami, S.C., 2001. Temporal and ephemeral variations in copepod community in the estuaries of Mandovi and Zuari e west coast of India. Journal of Plankton Research 23, 19e26. Darecki, M., Weeks, A., Sagan, S., Kowalczuk, P., Kaczmarek, S., 2003. Optical characteristics of two contrasting Case 2 waters and their influence on remote sensing algorithms. Continental Shelf Research 23, 237e250. Devassy, V.P., Goes, J.I., 1989. Seasonal patterns of phytoplankton biomass and productivity in tropical estuarine complex (west coast of India). Proceedings of the Indian Academy of Sciences: Plant Sciences 99, 485e501. Doerffer, R., Fisher, J., 1994. Concentration of chlorophyll, suspended matter and gelbstoff in case II waters derived from satellite coastal zone colour scanner data with inverse modeling method. Journal of Geophysical Research 99, 7457e7466. Doerffer, R., 1992. Imaging Spectroscopy for Detection of Chlorophyll and Suspended Matter. GKSS 92/E/54, pp. 215e257. Friedrichs, C.T., Aubrey, D.G., 1994. Tidal propagation in strongly convergent channels. Journal of Geophysical Research 99, 3321e3336. Gordon, H.R., 1978. Remote sensing of optical properties in continuously stratified waters. Applied Optics 17, 1631e1636. Kowalczuk, P., Kaczmarek, S., 1996. Analysis of temporal and spatial variability of yellow substance absorption in the southern Baltic. Oceanologia 38, 3e32. Kundu, S.N., Sahoo, A.K., Mohapatra, S., Sing, R.P., 2001. Change analysis using IRS-P4 OCM data after the Orissa cyclone. International Journal of Remote Sensing 22, 1383e1389. Menon, H.B., 2004. Calibration of an optical equation to analyze the atmospheric turbidity and water quality of an estuarine environment. Photonirvachak, Journal of Indian Society of Remote Sensing 32, 287e300. Menon, H.B., Lotliker, A., Nayak, S.R., 2005. Pre-monsoon bio-optical properties in estuarine, coastal and Lakshadweep waters. Estuarine, Coastal and Shelf Science 63, 211e223. Morel, A., Prieur, L., 1977. Analysis of variation in ocean colour. Limnology and Oceanography 22, 709e722. Muller, J.L., Austin, R.W., 1995. Ocean Optics Protocol for Sea WiFs Validation, Revision. Goddard Space Flight Center, Maryland. Murthy, C.S., Das, P.K., 1972. Pre-monsoon tidal flow characteristics of Mandovi estuary. Indian Journal of Marine Science 1, 220e225. Murthy, C.S., Das, P.K., Nair, R.R., Veerayya, M., Varadachari, V.V.R., 1976. Circulation and sedimentation process in and around the Aguada bar. Indian Journal of Marine Science 5, 9e17. Nayak, G.N., Bukhari, S.S., 1992. Spatial and temporal distribution of total suspended matter and other associated parameters in the Zuari estuary, Goa. Journal of Indian Association of Sedimentologists 11, 55e69. O’Reilly, J.E., Maritorena, S., Mitchell, G., Siegel, D.A., Carder, K.L., Garver, S.A., Kahru, M., McClain, C., 1998. Ocean colour chlorophyll-a algorithms for Sea WiFS. Journal of Geophysical Research 103 (C11), 24937e24953. Pundare, U.V., 1989. Studies for improvement of navigation in Aguada bay at Goa. Third National Conference on Dock and Harbour Engineering, Surathkal, pp. 589e595.
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531 Qasim, S.Z., Sen Gupta, R., 1981. Environmental characteristics of the MandovieZuari estuarine system of Goa. Estuarine, Coastal and Shelf Science 13, 557e578. Rao, G.D., Rao, T.C.S., 1974. Textural characteristics of sediments of Marmagao bay, central west coast of India. Indian Journal of Marine Science 3, 93e98. Strickland, J.D., Parson, T.R., 1972. A practical handbook of seawater analysis. Bulletin of the Fisheries Research Board of Canada 167e310. Tang, D.L., Kawamura, H., Luis, A.J., 2002. Short-term variability of phytoplankton blooms associated with a cold eddy in the northeastern Arabian Sea. Remote Sensing of Environment 81, 82e89.
531
Unnikrishnan, A.S., Shetye, S.R., Guveia, A.D., 1999. Tidal propagation in MandovieZuari estuarine network, west coast of India: impact of freshwater influx. Estuarine, Coastal and Shelf Science 45, 737e744. Verma, K.K., Rao, L.V.G., Cherian, T., 1975. Temporal and spatial variations in hydrographic conditions of Mandovi estuary. Indian Journal of Marine Science 4, 11e17. Yan, B., Stamnes, K., Toratani, M., Li, W., Stamnes, J.J., 2002. Evaluation of a reflectance model used in the Sea WiFS ocean colour algorithm: implications for chlorophyll concentration retrievals. Applied Optics 41 (30), 6243e6259.
Analysis of estuarine colour components during non-monsoon period through Ocean Colour Monitor H.B. Menon a,*, A.A. Lotliker a, S.R. Nayak b b
a Department of Marine Science, Goa University, University P.O., 403 206, Goa State, India Marine Water and Resource Division, Space Application Center, Indian Space Research Organization, Ahmedabad 380 015, India
Received 6 July 2005; accepted 21 October 2005 Available online 13 December 2005
Abstract Simultaneous acquisition of water samples, radiance and irradiance measurements were carried out from 40 stations in the MandovieZuari estuaries during February to May 2002. From the samples collected, inherent and apparent optical properties (IOP and AOP) such as absorption coefficient (a), upwelling diffuse attenuation coefficient (ku) and subsurface reflectance (R) were derived. Using these optical properties, radiative transfer at each water column is examined. On the basis of the radiative transfer outcome, band-ratio algorithms are derived for three optically active substances (OAS), viz, chlorophyll-a, suspended sediment and coloured dissolved organic matter (CDOM). The respective algorithms are 670/555, 490/670 and 412/670 nm for chlorophyll-a, suspended sediment and CDOM. These algorithms are applied to Ocean Colour Monitor (OCM), onboard Indian Remote Sensing Satellite (IRS)-Polar Satellite Launch Vehicle (P4), scenes (digital data), to synoptically analyze these OAS. The synoptic analysis of OAS revealed different hydrodynamic characteristics of the estuaries during non-monsoon seasons. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: algorithm; IRS-P4-OCM; band ratio; CDOM; chlorophyll-a; suspended sediment; hydrodynamics; non-monsoon
1. Introduction Estuarine environment is characterized by dynamic water bodies of different temporal changes that occur due to hourly (tide) to seasonal variations (river discharge). This spatial heterogeneity of the water column is further augmented by change in the depth and width of the estuary. To monitor such changes of varying time scales due to tidal advection, freshwater influx, dispersion and resuspension, synoptic analysis is imperative. This could be accomplished through optical remote sensing. The ocean colour monitoring for a long period is an important tool to decipher the biophysical processes, ecology and coastal environmental changes (Tang et al., 2002; Babin et al., 2003; Darecki et al., 2003; Menon, 2004; Menon et al., 2005). On the basis of ocean colour constituents, waters are divided into case I and case II (Morel and Prieur, 1977). Case I refers to
* Corresponding author. E-mail address: [email protected] (H.B. Menon). 0272-7714/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ecss.2005.10.009
waters for which phytoplankton species and their accessory pigments play a dominant role in determining the colour while case II waters are those for which the colour is determined by suspended inorganic and dissolved organic matters along with phytoplankton and their accessory pigments. Adopting this criterion the study area falls under case II category. The general estuarine dynamics is exemplified by the need for better understanding of mechanism that control the water quality of the area of study, MandovieZuari estuarine system. The system is under strong anthropogenic influence due to transportation of iron ore, as a part of mining industry, through these estuaries. Hence the seepage of iron ore from the barges, carrying ore, completely alters the water quality and the photic zone. This is being augmented by the sediment flux from mining dumpsites through seasonal rain. On the basis of foraminiferal assemblages, Rao and Rao (1974) made an attempt to analyze the ecology of Mandovi and Zuari estuaries. Efforts had been made to study the variation of chlorophyll-a distribution in Zuari estuary (Bhargava and Dwivedi, 1974) and Mandovi estuary (Dalal and Goswami, 2001), circulation and mixing in the
524
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
estuaries (Murthy and Das, 1972; Verma et al., 1975; Qasim and Sen Gupta, 1981; Unnikrishnan et al., 1999) and sediment distribution in the Zuari estuary (Nayak and Bukhari, 1992). The results revealed that sediment distribution in Zuari estuary during non-monsoon period is between 15 and 22 mg l1 while chlorophyll-a is in the range 1.0e5.0 mg l1. No study is reported about the distribution of coloured dissolved organic matter (CDOM) in these estuaries. Hence it is clearly evident from the above references that little attention is paid in analyzing the water quality of estuaries either through in situ observation or using high-resolution synoptic data during the recent past. Therefore, an attempt is made to retrieve ocean colour components from the estuarine system through optical remote sensing, as this may be a valuable tool for monitoring the long-term ecological consequences due to anthropogenic activities in the estuarine system. This has led the authors to carry out the present study with the following objectives: 1. Examine the feasibility of Ocean Colour Monitor (OCM) data to map the spatial distribution of three optically active substances (OAS) viz, chlorophyll-a, CDOM and suspended sediments through suitable algorithms to understand the water quality. 2. Understand the hydrodynamics of the estuarine system through ocean colour component analysis.
2. Study area The area for the present study is Mandovi and Zuari, two estuaries, situated in Goa, along the west coast of India. These two estuaries are interconnected by a narrow canal, the Cumbarjua (Fig. 1). At the upstream ends, they receive large freshwater influx during southwest monsoon (JuneeSeptember) and little during the rest of the year. The average annual rainfall of Goa is about 3000 mm of which nearly 80% occurs during the southwest monsoon period (Qasim and Sen Gupta, 1981). Hence the estuaries are freshwater dominated during JuneeSeptember and seawater dominated during the rest of the year. Therefore, tropical southwest monsoon, riverine and seawater flows make the estuaries an ecologically complex ecosystem. 3. Method The waters in the estuarine system being highly optically complex, retrieval of chlorophyll-a, suspended sediment and CDOM through OCM is possible only by applying site specific algorithms developed through the analysis of hyperspectral water leaving radiance in the optical range (400e700 nm) of Electromagnetic Radiation (EMR). The hyperspectral water leaving radiance can be generated either through an in-water radiometer having hyperspectral bands or through radiative
Fig. 1. Study area showing the different station positions (stations 1e10). Dotted lines indicate depth contours.
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
transfer analysis. In the present case the second approach is adopted. To carry out the radiative transfer analysis, the inherent optical properties (IOP) and apparent optical properties (AOP) of each water column, represented by different stations, are generated through water sample analysis. 3.1. In situ observations The in situ observations from the area of study involve water sampling, in-water radiometer measurements (using Satlantic radiometer), Eko Sunphotometer measurement for aerosol optical depth (AOD) observations and Secchi disk depth measurement for transparency of light. From each station, water sample was collected from a depth between sea surface and 1% of photosynthetically active radiations (PAR). Water sampling and radiometer observations were always from the area where the shadow of the boat did not fall. The observations were carried out during pre-monsoon period (FebruaryeMay) 2002. In each month, field survey was carried out from 10 stations (Fig. 1) between 1100 hours and 1400 hours (the satellite passes over the area around 1200 hours). Hence all the observations were within 3 h of satellite pass. As the estuarine width decreases drastically between mouth and head, positions of different stations were selected in such a way that the width of each station is about 1 km or more (three times the pixel size of OCM). This precaution was taken to prevent any overlapping of land radiance with water leaving radiance. Estuaries being a shallow water body, it is necessary to arrest any radiance from bottom of each station. This was done by selecting station where the depth of the water column is three times more than the Secchi disc depth of the corresponding station (Muller and Austin, 1995). 3.2. In-water radiometer measurement A Satlantic radiometer consisting of Profiling Multichannel Radiometer (SPMR) and Satlantic Multichannel Surface Reference (SMSR) was used to measure the upwelling radiance (Lu(l, 0)), downwelling irradiance (Ed(l, 0)) and downwelling irradiance reaching the sea surface (Es(l, 0þ)). The instrument operates in seven bands with wavelengths centered at 412, 443, 490, 510, 555, 670 and 780 nm. Due to ambiguity in the readings, the data obtained at 780 nm were not considered. The radiometer measurements were always taken in association with water sample collection. This precaution was taken to prevent any change in the computed water leaving radiance values from measured radiance values due to variation in solar zenith angle. While operating radiometer, utmost care was taken not to exceed the tilt of SPMR beyond 5 and that of SMSR 10 , a standard suggested by the manufacturer. 3.3. Inherent optical properties (IOP) and apparent optical properties (AOP) generated through water sample analysis Two water samples, 5 L, each were collected from each station using Niskin sampler. One sample for suspended particles
525
(chlorophyll-a and sediment) and the other for CDOM. In order to avoid the degradation of the samples, 1 ml of saturated MgCO3 was added to the sample for chlorophyll-a analysis (Strickland and Parson, 1972), whereas 0.4 ml of 1 M HgCl2 was added to the sample for analysis of CDOM (Kowalczuk and Kaczmarek, 1996). The samples were then preserved in dark under low temperature prior to laboratory analysis. As the details of the water sample analysis were given elsewhere (Menon et al., 2005), the same is not repeated here. As per Gordon (1978) subsurface reflectance R(l) is Rðl; 0 Þ ¼ Eu ðl; 0 Þ=Ed ðl; 0 Þ ¼ QLw ðl; 0 Þ=Ed ðl; 0 Þ ¼ QRrs ðl; 0 Þ
ð1Þ
where Eu(l) and Ed(l) are the upwelling and downwelling irradiances just below the sea surface, respectively, Lw(l) is the water leaving radiance and Q is the bi-directional reflectance (irradianceeradiance ratio) factor. Since Q is spectrally variant, radiometer observed values are used to calculate the Q factor. In order to derive hyperspectral water leaving radiance (Lw(l)), R(l) in Eq. (1) was derived as per the details given in Menon et al. (2005). 3.4. Satellite data processing The optical sensor, OCM onboard IRS-P4 has eight spectral bands. Out of which, six are in the visible spectrum of EMR and two are in the near infrared (NIR) region. Together they are in the range 402e885 nm with spectral resolution of 20 nm in the visible bands and 40 nm in NIR bands. The spatial resolution of the sensor is 360 m 250 m and the revisit cycle is 2 days (Kundu et al., 2001). The unsigned 16-bit data of OCM for the eight bands were imported using ERDAS IMAGINE 8.4. The images were then geo-referenced using ground control points (GCP) after which the area of study covering MandovieZuari estuarine system, Goa, India, bounded by 73 55#e74 5# E and 15 15#e15 25# N was extracted from the full scene. 3.5. Atmospheric correction Optical sensors, in general, have two NIR bands for applying atmospheric correction to visible bands to retrieve OAS. The concept is that radiance received by NIR bands are only from aerosols irrespective of the water column being highly turbid or not. But studies carried out by Yan et al. (2002) revealed that NIR bands of Sea WiFS (765 and 865 nm) had water leaving radiance component when the water column was turbid. Hence, in the present study, aerosol radiance is computed by deriving AOD through sunphotometer (Chylek et al., 2003). An Eko Sunphotometer having filters 368, 500, 675 and 862 nm was used to derive AOD. The transmission loss due to atmospheric gases is incorporated and Rayleigh radiance is computed (Doerffer, 1992). Thus the water leaving radiances from the selected bands of OCM are generated by subtracting Rayleigh and aerosol radiances.
526
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
3.6. OCM derived chlorophyll-a, sediment and CDOM There are many algorithms to retrieve case II water constituents. These are inverse modeling approach (Doerffer and Fisher, 1994), band-ratio method (O’Reilly et al., 1998) and Principal Component Analysis (PCA). In the present case the second approach is adopted. The basis in choosing wavelengths for band-ratio algorithm is differential absorption and scattering of solar flux associated with each wavelength in the visible spectrum of EMR, due to inherent characteristics of each OAS. For example, in the open ocean water optically active substance is mainly chlorophyll-a. Hence the interaction of visible spectra of EMR with this water is such that there is always a linear relation between ratio of water leaving radiance received by blue and green bands of OCM and chlorophyll-a concentration. But from the analysis of IOP due to chlorophyll-a, sediment and CDOM, it is realized that interaction of optical spectra in estuarine waters is non-linear and all the OAS utilize energy associated between 400 and 500 nm ranges. Hence to distinguish these OAS through OCM, hyperspectral water leaving radiance (for every 1 nm from 400 to 700 nm) is computed through a radiative transfer model and analyzed. The details about the calibrated radiative transfer modeling and some of the results including hyperspectral water leaving radiance are given in Menon (2004) and Menon et al. (2005). The procedure in identifying the wavelengths to develop band-ratio algorithm to retrieve OAS from estuarine waters through OCM is given below. The first step is to select a station with minimum concentration of OAS so that spectral radiance leaving this station can be the baseline spectrum. Hence water leaving radiance corresponding to station 7 sampled on 13 May 2002 is taken as baseline spectrum. Subsequently, water leaving radiances are computed for different concentrations of chlorophyll-a within the acceptable range prevailing in the estuarine waters during pre-monsoon season. While doing this, it is assumed that concentration of sediment and CDOM (ay (440)) are not varying. Then each hyperspectral radiance is divided by the baseline spectrum. This has resulted in the identification of wavelength having the least and maximum absorption by chlorophyll-a in the optical spectrum. The same procedure is repeated in the case of sediment and ay (440). Thus the analysis revealed that wavelengths suitable to retrieve chlorophyll-a are 555 and 670 nm, for sediment 490 and 670 nm while for CDOM 412 and 670 nm, respectively. The band-ratio algorithms are shown in Fig. 2a. From the graph it appears that as chlorophyll-a and sediment increase, the power correlation between the respective band ratios and concentrations approaches some constant value. Similarly in the case of CDOM, there is an exponential fit between ay (440) and ratio of water leaving radiance at 412 and 670 nm and the ratio approaches a constant. Therefore, significant inference drawn, from the analysis, is to retrieve OAS with concentrations more than the present case, different algorithms have to be developed. That means the algorithms developed are site and season specific. In all the cases R2 is more than 0.95. Therefore, by using these chosen wavelengths, bandratio algorithm is applied to OCM data and different optically
Fig. 2. (a) Correlation between (1) chlorophyll-a concentration and the ratio of water leaving radiance at 670 (Lw 670) and 555 nm (Lw 555) bands; (2) sediment concentration and the ratio of water leaving radiance at 490 (Lw 490) and 670 nm (Lw 670) bands and (3) absorption coefficient due to CDOM at 440 nm (ay (440)) and the ratio of water leaving radiance at 412 (Lw 412) and 555 nm (Lw 670) bands. Synoptic distribution of (1) chlorophyll-a, (2) suspended sediment and (3) ay (440) on (b) 8 January 2003, (c) 14 January 2003, (d) 14 February 2002 and (e) 24 May 2003.
active constituents are retrieved. The results are shown in Fig. 2bee. The respective algorithms are the following: , where R1 ¼ Lw 670/Lw 555; 1. Chlorophyll-a ¼ 5.5931R0.615 1 2. ay (440) ¼ 2.9393R2.2486 , where R2 ¼ Lw 412/Lw 670; 2 3. Suspended sediment ¼ 17.11R0.3596 , where R3 ¼ 3 Lw 490/Lw 670.
4. Results and discussion Using the algorithms developed, four satellite scenes (satellite data) are processed. These are 8 January 2003 (Fig. 2b), 14 January 2003 (Fig. 2c), 14 February 2002 (Fig. 2d), and 24 May 2003 (Fig. 2e). These scenes are chosen to show the ability of the algorithms in retrieving OAS during different hydrodynamics of the estuary prevailing in different non-monsoon
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
527
Fig. 2 (continued).
periods. The studies on the hydrography and circulation made by Murthy and Das (1972), Verma et al. (1975) and Qasim and Sen Gupta (1981) revealed that estuaries behaved like a mixed/homogeneous one during pre-monsoon, partially mixed during post-monsoon and saltwedge type during monsoon. Hence they are tidally dominated during pre-monsoon, tide and freshwater influx control them during monsoon and post-monsoon seasons. Therefore, January data are chosen as a representative of the estuarine water quality during the final stages of post-monsoon, February to represent the initial stages of pre-monsoon and May the final stages of pre-monsoon season. Another reason for choosing these scenes is to show the ability of the algorithms in retrieving OAS during non-monsoon seasons of any year in addition to the period of in situ observations used for algorithm development. In general, the OCM derived products and in situ observations are in good agreement and are in the same range. A graphical representation of validation is given in Fig. 3. Though, good correlation is observed for all the constituents, an offset is seen in the case of chlorophyll-a and sediment. The offset is on the
axis where satellite derived values are taken. For validation, all the available data (chlorophyll-a and sediment) published so far, from the area, are considered. As some of the data are from the midstream of the estuary, where the width of the estuary is less than 1 km (the resolution of the sensor is 360 m 250 m), selecting corresponding pixel of the in situ observation from OCM imagery will result in an erroneous value as the water leaving radiance can be entangled by the radiance from land. Therefore, to correlate those in situ data, pixels are not considered corresponding to the station positions rather from the adjacent area where the width of the estuary is around 1 km (three times the resolution of the sensor) or more. This resulted in the offset. As there are no published data on CDOM from the area, these values are derived during separate field surveys. Hence no offset is seen in this case. The spatial distribution of chlorophyll-a, sediment and CDOM in Mandovi and Zuari estuaries derived through IRSP4-OCM during the above days are given in Fig. 2bee. In the estuarine waters chlorophyll-a is in the range 3.3e5.0 mg l1, sediment 14e22 mg l1 and ay (440) 1.4e2.2 m1 during
528
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
Fig. 2 (continued).
January. The distribution of these OAS are in the range 3.8e 5.0 mg l1, 18e22 mg l1 and 0.4e2.2 m1, respectively, during February whereas they are in the range 4.5e5.5 mg l1, 15 mg l1 and 0.8 m1, respectively, during May. The concentration of chlorophyll-a in the estuarine waters is always more than the coastal waters. Chlorophyll maxima are always in the upper reaches of estuarine waters irrespective of month. The maximum concentration, 5.0 mg l1, is during May. Further, an examination of distribution of chlorophyll-a revealed an anomalous pattern in Mandovi estuary whereas a gradual increase from downstream to upstream (towards the head) in Zuari estuary during January (Fig. 2b(1), c(1)) and February (Fig. 2d(1)). Barring this anomalous low, distribution of chlorophyll-a in both the estuaries during these months is in the range mentioned above. Similarly at the mouth and upstream of Mandovi, a plume of sediment is depicted during February while the distribution in January is near uniform with even distribution in May. Zuari estuary shows a different picture. Here, in the midstream, a well defined sediment plume is seen during January (Fig. 2b(2), c(2)) and February (Fig. 2d(2)) while it is distributed uniformly during May (Fig. 2e(2)). A noticeable feature seen
in both the estuaries is the high incidence of CDOM associated with the regions of sediment plumes. The respective concentrations of sediment and absorption due to CDOM at 440 nm during January and February are 19 mg l1 and 2.2 m1. The spatial distribution of CDOM shows least concentration in the coastal region with a gradual increase to midstream of Zuari estuary during January (Fig. 2b(3), c(3)), February (Fig. 2d(3)) and May (Fig. 2e(3)). In their studies Nayak and Bukhari (1992) observed a region of high sediment concentration (22 mg l1) in the midstream of Zuari estuary for which the observations were carried out in February. The reason for such a high concentration is ascribed to the drastic decrease in estuarine depth from mouth to upstream and the effect of tide. Zuari estuary is funnel shaped with a wide and deep mouth, which narrows and shallows down to upstream. The width at the mouth of the Zuari estuary is 7.0 km while at the midstream it is around 3.0 km. The shallowing is abrupt at the place where suspended sediment and CDOM concentrations are high. At this point the depth shallows down to 3 m from 10 m at the mouth. Also the time of in situ observation at this point and the satellite
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
Fig. 3. Correlation between satellite derived and in situ values of (a) chlorophyll-a, (b) suspended sediment, (c) absorption due to CDOM at 440 nm (ay (440)).
pass (1200 hours) coincided with high tide (1157 hours) during February. As a result the tidal energy associated with seawater inflow generated turbulence at the bottom and resuspension of sediment and CDOM. This had led to the formation of a zone of high concentration of sediment and CDOM in the midstream. This process is enhanced by softer nature of the sediments (Rao and Rao, 1974) and by convergence of currents at the constriction and increasing currents towards the narrow part of the estuary (Pundare, 1989). Similar to Zuari, Mandovi estuary is also funnel shaped with a broad mouth of 4.5 km and a very narrow upstream. In the estuary a sediment plume is observed at the mouth, upstream and further north along the coast during February (Fig. 2d(2)). The occurrence of sand bars, popularly called as Aguada and Reis Magos bar, near the entrance of the Mandovi is known for centuries. The important feature is the presence of the ramp-like structure in the navigation channel of the Mandovi with a depth gradient from 10 m to 4 m (Murthy et al., 1976; Qasim and Sen Gupta, 1981). Also the northwestern part of the mouth of Mandovi is exposed to swell from south,
529
southwest, northwest and north. Therefore, the shoal on both sides of the entry channel became the main sight of wave breaking. During the initial phase of pre-monsoon season (February) the waves at the entrance of the Mandovi estuary (northern portion) are small with a significant period of 68 s. Heavy swells and intensification of wave activity by several folds during the monsoon season led to highly turbulent conditions over shoals resulting in the removal of large quantities of material from the beaches and their transfer close to the navigational channel. During the post-monsoon period (January), these materials are returned and deposited on either side of the entrance channel and resulting in a uniform sediment pattern. During the pre-monsoon season (February) substantial amount of sediment are added from the coastal region to estuary in association with swell of west and northwest (Murthy et al., 1976). Therefore, high concentrations of sediment seen at the head of the estuary and along the coast north of the mouth of Mandovi during February and a uniform sediment distribution during January are the remnants of the sediment transported in the estuary during monsoon season. However, the sediment distribution in Zuari during January (Fig. 2b(2), c(2)) is due to different factors. Though the in situ observations were in conjunction with low tide, a clear plume of sediment and CDOM are seen at the midstream of Zuari and a near uniform distribution of sediment and a patchy distribution of CDOM in Mandovi (Fig. 2b(3), c(3)). It is an established fact that in shallow estuaries the momentum balance is primarily pressure gradient and friction. Friedrichs and Aubrey (1994) had shown that in such systems if the estuarine channel converges (if it shallows and narrows in the upstream direction) then amplification due to convergence of the channel cancels decay due to friction, leaving the amplitude unchanged over long distances along the estuarine channels. This explains the reason for the distribution in both the estuaries. In May (Fig. 2e) the scenario is different. Though the observation coincided with low tide, rather than a plume, the sediment and CDOM are evenly distributed throughout the estuaries. This could be attributed to the hydrodynamic characteristics of the estuaries. Both the estuaries are well mixed/homogenous during pre-monsoon (Unnikrishnan et al., 1999). The OCM data/scene selected in the study are of May, they represent the complete evolution of estuary into a homogenous one from partially mixed. The explanation is given below. Let R be the volume of freshwater being brought to the estuary by river and V the volume of seawater due to tide. During monsoon, the ratio of R to V is always greater than one and hence leads to a salt-wedge estuary. But during post-monsoon period though R decreases, V is the same. Hence the ratio decreases. This leads to the formation of a partially mixed estuary. During pre-monsoon R is negligible so that ratio may be very less. Hence turbulent mixing becomes so efficient that locally all salinity differences are eliminated and the estuary becomes vertically mixed. This is the reason for the typical distribution of sediment and CDOM during the final phase of pre-monsoon (Fig. 2e(2), e(3)). A similarity in the distribution pattern of CDOM with
530
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531
that of sediment can be a pointer to the fact that the reasons attributed for the high concentration of sediment are responsible for high CDOM concentration too. In contrast to the distribution of sediment and CDOM, the distribution of chlorophyll-a in the Mandovi estuary shows a gradual increase towards the head during May (Fig. 2e(1)) while a scattered pattern with a patch of low concentration in the midstream and upstream during January (Fig. 2b(1), c(1)) and February (Fig. 2d(1)). Similarly a gradual increase in concentration of chlorophyll-a is seen from mouth towards the upstream of Zuari estuary during all the months. Though there is a variation in the chlorophyll-a distribution between mouth and head among the two estuaries, the chlorophyll-a concentration is maximum at the head of the estuaries during all months. Heavy precipitation of the order of 250e 300 cm yr1 and land runoff (Qasim and Sen Gupta, 1981) during southwest monsoon period (JuneeSeptember) bring huge quantities of nutrients making the estuaries highly productive (Devassy and Goes, 1989). Therefore, remnants of the nutrient rich water of the previous monsoon are responsible for the high chlorophyll-a and hence primary production at the head of the estuaries. Though chlorophyll-a concentration is maximum and almost of the same order of magnitude at the head of the estuaries during all the months, it is observed that the concentration at the lower stream of the estuaries increases from January to May. In their studies on the distribution of phytoplankton in Zuari estuary, Bhargava and Dwivedi (1974) observed a similar pattern. According to them the maxima in chlorophyll-a distribution generally coincided with high salinity. This could be attributed to the influence of neritic waters at the lower stream of the estuary. The neritic waters being chlorophyll rich phytoplankton, during flood (high tide), they dominate over the ebb tide crop of phytoplankton. Therefore, a tide controlled estuarine circulation led to the increase in the distribution of chlorophyll-a at the lower stream from January to May. 5. Conclusion The paper presents the procedure and development of local band-ratio algorithms for three major OAS (chlorophyll-a, suspended sediment and CDOM) in terms of IOP, AOP, in-water constituent measurements and retrieval of OAS through OCM. This is the first time that a highly turbid estuarine water column has been successfully analyzed optically through OCM data. The hydrodynamics prevailing in the estuarine system during different non-monsoon seasons could be well explained through mapping the colour components. Knowledge of distribution of water quality parameters and the location of zone of high biomass and plumes is valuable in establishing management policies for these ecologically important waters of Goa. Acknowledgement The first author HBM is thankful to Indian Space Research Organization (ISRO), Bangalore for the project sanctioned
‘‘algorithm development’’ under the National programme IRS-P4-OCM. The critical comments made by the unknown referees are highly acknowledged. References Babin, M., Stramski, D., Ferrari, G.M., Claustre, H., Bricaud, A., Obolensky, G., Hoepffner, N., 2003. Variations in the light absorption coefficients of phytoplankton, nonalgal particles and dissolved organic matter in coastal waters around Europe. Journal of Geophysical Research 108 (C7), 3211. Bhargava, R.M.S., Dwivedi, S.N., 1974. Diurnal variation in phytoplankton pigment in Zuari estuary, Goa. Indian Journal of Marine Science 3, 142e145. Chylek, P., Henderson, B., Mishchenko, M., 2003. Aerosol radiative forcing and the accuracy of satellite aerosol optical depth retrieval. Journal of Geophysical Research 108, 4764. Dalal, S.G., Goswami, S.C., 2001. Temporal and ephemeral variations in copepod community in the estuaries of Mandovi and Zuari e west coast of India. Journal of Plankton Research 23, 19e26. Darecki, M., Weeks, A., Sagan, S., Kowalczuk, P., Kaczmarek, S., 2003. Optical characteristics of two contrasting Case 2 waters and their influence on remote sensing algorithms. Continental Shelf Research 23, 237e250. Devassy, V.P., Goes, J.I., 1989. Seasonal patterns of phytoplankton biomass and productivity in tropical estuarine complex (west coast of India). Proceedings of the Indian Academy of Sciences: Plant Sciences 99, 485e501. Doerffer, R., Fisher, J., 1994. Concentration of chlorophyll, suspended matter and gelbstoff in case II waters derived from satellite coastal zone colour scanner data with inverse modeling method. Journal of Geophysical Research 99, 7457e7466. Doerffer, R., 1992. Imaging Spectroscopy for Detection of Chlorophyll and Suspended Matter. GKSS 92/E/54, pp. 215e257. Friedrichs, C.T., Aubrey, D.G., 1994. Tidal propagation in strongly convergent channels. Journal of Geophysical Research 99, 3321e3336. Gordon, H.R., 1978. Remote sensing of optical properties in continuously stratified waters. Applied Optics 17, 1631e1636. Kowalczuk, P., Kaczmarek, S., 1996. Analysis of temporal and spatial variability of yellow substance absorption in the southern Baltic. Oceanologia 38, 3e32. Kundu, S.N., Sahoo, A.K., Mohapatra, S., Sing, R.P., 2001. Change analysis using IRS-P4 OCM data after the Orissa cyclone. International Journal of Remote Sensing 22, 1383e1389. Menon, H.B., 2004. Calibration of an optical equation to analyze the atmospheric turbidity and water quality of an estuarine environment. Photonirvachak, Journal of Indian Society of Remote Sensing 32, 287e300. Menon, H.B., Lotliker, A., Nayak, S.R., 2005. Pre-monsoon bio-optical properties in estuarine, coastal and Lakshadweep waters. Estuarine, Coastal and Shelf Science 63, 211e223. Morel, A., Prieur, L., 1977. Analysis of variation in ocean colour. Limnology and Oceanography 22, 709e722. Muller, J.L., Austin, R.W., 1995. Ocean Optics Protocol for Sea WiFs Validation, Revision. Goddard Space Flight Center, Maryland. Murthy, C.S., Das, P.K., 1972. Pre-monsoon tidal flow characteristics of Mandovi estuary. Indian Journal of Marine Science 1, 220e225. Murthy, C.S., Das, P.K., Nair, R.R., Veerayya, M., Varadachari, V.V.R., 1976. Circulation and sedimentation process in and around the Aguada bar. Indian Journal of Marine Science 5, 9e17. Nayak, G.N., Bukhari, S.S., 1992. Spatial and temporal distribution of total suspended matter and other associated parameters in the Zuari estuary, Goa. Journal of Indian Association of Sedimentologists 11, 55e69. O’Reilly, J.E., Maritorena, S., Mitchell, G., Siegel, D.A., Carder, K.L., Garver, S.A., Kahru, M., McClain, C., 1998. Ocean colour chlorophyll-a algorithms for Sea WiFS. Journal of Geophysical Research 103 (C11), 24937e24953. Pundare, U.V., 1989. Studies for improvement of navigation in Aguada bay at Goa. Third National Conference on Dock and Harbour Engineering, Surathkal, pp. 589e595.
H.B. Menon et al. / Estuarine, Coastal and Shelf Science 66 (2006) 523e531 Qasim, S.Z., Sen Gupta, R., 1981. Environmental characteristics of the MandovieZuari estuarine system of Goa. Estuarine, Coastal and Shelf Science 13, 557e578. Rao, G.D., Rao, T.C.S., 1974. Textural characteristics of sediments of Marmagao bay, central west coast of India. Indian Journal of Marine Science 3, 93e98. Strickland, J.D., Parson, T.R., 1972. A practical handbook of seawater analysis. Bulletin of the Fisheries Research Board of Canada 167e310. Tang, D.L., Kawamura, H., Luis, A.J., 2002. Short-term variability of phytoplankton blooms associated with a cold eddy in the northeastern Arabian Sea. Remote Sensing of Environment 81, 82e89.
531
Unnikrishnan, A.S., Shetye, S.R., Guveia, A.D., 1999. Tidal propagation in MandovieZuari estuarine network, west coast of India: impact of freshwater influx. Estuarine, Coastal and Shelf Science 45, 737e744. Verma, K.K., Rao, L.V.G., Cherian, T., 1975. Temporal and spatial variations in hydrographic conditions of Mandovi estuary. Indian Journal of Marine Science 4, 11e17. Yan, B., Stamnes, K., Toratani, M., Li, W., Stamnes, J.J., 2002. Evaluation of a reflectance model used in the Sea WiFS ocean colour algorithm: implications for chlorophyll concentration retrievals. Applied Optics 41 (30), 6243e6259.