Sei sulla pagina 1di 9

J Indian Soc Remote Sens (March 2013) 41(1):117125 DOI 10.

1007/s12524-012-0203-x

RESEARCH ARTICLE

Validation of Chlorophyll-a Algorithms in the Coastal Waters of Bay of Bengal Initial Validation Results from OCM-2
P. V. Nagamani & M. I. Hussain & S. B. Choudhury & C. R. Panda & P. Sanghamitra & R. N. Kar & A. Das & I. V. Ramana & K. H. Rao
Received: 9 September 2011 / Accepted: 6 January 2012 / Published online: 3 April 2012 # Indian Society of Remote Sensing 2012

Abstract In this paper we report chlorophyll measurements made during an ocean colour validation cruise in April 2011 of the research vessel, Sagar Paschimi in the coastal waters of Northern Bay of Bengal. The chlorophyll-a concentration in these waters range from 0.2 to 4.0 mg/m3. Chlorophyll-a concentration from OCM-2 was estimated using the global ocean colour algorithms namely, OC2, OC3, OC4 and Chl-a algorithms respectively. OCM data was processed using the global SeaWiFS Data Analysis System (SeaDAS) in which all the above mentioned algorithms are embedded for estimating the chlorophyll-a concentration. A comparative study was made between and in-situ and satellite derived chlorophyll-a concentration. Although the matchups between in-situ and satellite data from OCM-2 were sparse, it indicates that direct application of the standard SeaWiFS algorithm-the OC4-V4 algorithmin the coastal waters of the Bay of Bengal will underestimate chlorophyll-a by up to 30%. The results show a good
P. V. Nagamani (*) : S. B. Choudhury : I. V. Ramana : K. H. Rao National Remote Sensing Centre, ISRO, Hyderabad 500 625, India e-mail: pvnagamani@gmail.com M. I. Hussain S V P College of Engineering, Visakhapatnam, India C. R. Panda : P. Sanghamitra : R. N. Kar : A. Das Institute of Metals and Minerals Technology, CSIR, Bhubaneswar 751 013, USA

correlation with an R value of 0.61 using OC2 algorithm. However, all the other global algorithms over estimate the chlorophyll-a concentration even in low chlorophyll concentration range. The comparison between in-situ and all the existing chlorophyll algorithms shows the efficiency of these algorithms for quantification of chlorophyll in coastal waters and hence the need to develop regional algorithms and fluorescence based algorithms for better quantification. Keywords OCM-2 . Validation . Chlorophyll . Coastal waters

Introduction Monitoring and managing small coastal ecosystems require a considerable understanding of temporal dynamics of biophysical factors describing coastal water systems. For this reason, daily observation from space could be a very efficient tool and high resolution, multispectral satellite data have been successfully applied in the detection of chemical, biological and physical processes happening in coastal waters (Chauhan et al. 2002). However, monitoring coastal waters using the existing satellite instruments is a challenging task which requires very sophisticate procedures. We know that in open oceans, case I waters, phytoplankton is the principal agent responsible for the optical properties of waters. Instead, the optical properties of shallow and coastal waters, case II waters, are influenced not only by phytoplankton, but also by other substances, generally

118

J Indian Soc Remote Sens (March 2013) 41(1):117125

classified as suspended inorganic particles and yellow substance (IOCCG reports 2006; 2010). Under Oceansat-2 OCM validation program an important set of in situ bio-optical measurements has been carried out. Beside lab-measurements of water quality parameters, phytoplankton pigment concentrations like chlorophyll-a, b, c and carotinoids, turbidity and total suspended matter concentration. These measurements are performed using the TURNER made fluorometer. The areas of interest are located in the south-west of Bay of Bengal, Head of the Bay along the east coast of India. In this region, three sites have been selected: (1) Paradip a perineal port in Orissa shows interesting hydrographic condition. This site is characterised by (2) Mahandai wherein river Mahanadi meets the sea forming an estuarine environment and (3) Haldia again a estuarine port with hoggly and Haldia river meets in the sea. In Fig. 1 it has been shown the geographical distribution of the sampling location during the study period. The marine ecosystem of Bay of Bengal is characterized by its tropical climate. It is situated in the monsoon belt and receives high rainfall. Intensive fishing is the primary force driving the marine ecosystem, with climate as the secondary driving force. The marine ecosystem is affected by monsoons, storm surges, and cyclones but has no seasonal upwelling. However, in near shore areas, the mixing of nutrient rich bottom waters and warm surface waters creates conditions similar to upwelling (Dwivedi and Choubey 1998). The number and intensity of cyclones in the
Fig. 1 Geographical distribution of the sampling locations during the cruise period on 10th, 12th, 20th and 22nd of April 2011

northern part of the Bay of Bengal are likely to increase due to global warming. Details on the southwest monsoon has been studied by Desai and Bhargava (1998). Major rivers such as the Brahmaputra and Ganges discharge large quantities of fresh water into the Bay of Bengal (Dwivedi and Choubey 1998).This input of freshwater and silt impacts the salinity of the coastal and estuarine waters as well as coastal circulation patterns. It influences and governs marine ecosystem dynamics during the southwest monsoon. Hydrography and productivity has been studied extensively by Dwivedi (1993). Wetlands, marshes, and mangroves play an important role in the overall productivity. The Bay of Bengal marine ecosystem is considered as moderately productive (150300 gC/m2-yr) ecosystem based on SeaWiFS global primary productivity estimates. Changing environmental conditions are influencing currents, productivity and coastal pollution. Lakes connected to the Bay of Bengal marine ecosystem are changing. Some coastal areas serving as nursery grounds for commercially valuable species of prawns are polluted. Distribution of zooplankton biomass and biological production in the Bay of Bengal ecosystem is studied by Desai and Bhargava (1998). Benthic biomass production in the shelf region of the Bay of Bengal by Parulekar et al. (1982). In this present study we have used the available global bio-optical algorithms for estimating the chlorophyll-a concentration and validated with the in-situ data collected in the turbid coastal waters along the east coast of India. We tried to show the errors

J Indian Soc Remote Sens (March 2013) 41(1):117125

119

involved in estimating the chlorophyll-a concentration using band ratio based methods and need to develop regional, absorption based as well as fluorescence based techniques for estimating the chlorophyll-a concentration in the coastal waters. Efforts are on to understand the gap areas and accordingly address the issues to refine/modify the method with local coefficients.

Methodology In-Situ Data Collection The in-situ chlorophyll-a concentration was collected using TURNER fluorometer for estimating the chlorophyll-a concentration. The measurements were made using round-robin experiment for collecting the samples at all the stations. In these areas a total of 41 stations with 100 sampling points out of which 25 stations with surface observation located within 3 km from the shore have been selected, and the measurements have been performed about twice per month.

Study Area and Data Sets Used The study area covers Ocean Colour Monitor (OCM-2) data of path:13 and row:10 covering Paradip (20N, 86E) Mahanadi (20.15N, 86.12E) and Haladia (21.6N, 88.02E) regions with 360 m resolution has been utilised for this present study. The in-situ chlorophyll data collected was collected at surface (0 m), middle (5 m) and deep (1015 m) on 10th, 12th 20th and 22nd April 2011 are used in this study. The geographical distribution of the sampling locations was shown in Fig. 1.

Satellite Data Relatively cloud free data of Ocean Colour Monitor (OCM-2) data during the study period covering Paradip, Mahanadi and Haladia of path: 13 and row: 10 has been selected for processing the data. The global processing software namely SeaWiFS Data Analysis

Fig. 2 Comparison of chlorophyll-a concentration collected during the study period shows the vertical distribution for (a) Paradeep (b) Mahanadi and Haldia transects

120

J Indian Soc Remote Sens (March 2013) 41(1):117125

System (SeaDAS) was used for processing the OCM-2 data. The albedo value was set to 1.1 for land and cloud masking. Chlorophyll-a concentration was estimated using the available chlorophyll algorithms in SeaDAS software and compared with the in-situ measured chlorophyll-a concentration data for validation. The global ocean colour algorithms namely Ocean Colour 2 (OC2), Ocean Colour 4 (OC4) and Ocean colour 3 (OC3) and chlorophyll-a algorithms (Chl-a) (OReilly et al. 1998) are used for estimating the chlorophyll concentration in the coastal waters off
Fig. 3 Spatial distribution of OCM-2 derived Chlorophyll-a concentration along the east coast of India on 10th, 12th, 20th and 22nd of April 2011 using SeaDAS software (OC2 algorithm)

Mahanadi, Paradeep and Haldia regions as shown in the study area.

Results and Discussion As an initial analysis of the data distribution and chlorophyll-a variability the data collected for all the dates were plotted along the transect, as shown in Fig. 2. In this figure except the last date data i.e., the data collected on 22nd April 2011 was not plotted as

J Indian Soc Remote Sens (March 2013) 41(1):117125

121

that data set was discarded due to the presence of thick clouds over that region. From Fig. 2 the spatial distribution of chlorophyll-a along that transect, shows decrease in the chlorophyll-a concentration while moving away from the coast towards offshore waters. Chlorophyll-a concentration vary from a minimum value of 0.96 to a maximum value of 2.81 mg m3 with a mean value of 1.69 mg m3 at the surface; 1.09, 3.54 and 2.132 mg m3 at the middle depth and 0.45, 2.54 and 1.67 mg m3 at bottom depth as minimum, maximum and mean values respectively in Paradeep
Fig. 4 Spatial distribution of OCM-2 derived Chlorophyll-a concentration along the east coast of India on 10th, 12th, 20th and 22nd of April 2011 using SeaDAS software (OC3 algorithm)

area. Chlorophyll-a concentration vary from a minimum value of 0.38 to a maximum value of 1.4 mg m3 with a mean value of 0.76 mg m3 at the surface; 0.41, 1.6 and 0.69 mg m3 at the middle depth and 0.71, 1.65 and 1.11 mg m3 at bottom depth as minimum, maximum and mean values respectively in Mahanadi region. Chlorophyll-a concentration vary from a minimum value of 0.28 to a maximum value of 3.39 mg m3 with a mean value of 1.72 mg m3 at the surface; 1.4, 1.8 and 1.62 mg m3 at the middle depth and 1.22, 2.85 and 1.71 mg m3 at bottom depth as minimum, maximum and mean values respectively in Haldia region.

122

J Indian Soc Remote Sens (March 2013) 41(1):117125

As mentioned in the methodology SeDAS processing software was used to process the chlorophyll-a concentration using OC2, OC3, OC4 and Chl-a algorithms from OCM-2 data. The new gain co-efficients modified on June 2010 provided by National Data Centre (NDC) along with the L1B data product (the gain coefficients were given in the header file of each scene of OCM-2) were used for converting the digital numbers into radiance values. The cloud masking value was set to 0.07 for OCM-2 based on the reflectance values of band 8 i.e., 865 nm (according to the existing and proposed cloud masking algorithms of

SeaWiFS and MODIS sensors). However, this threshold was set to 0.07 based on our internal studies for OCM-2 that was carried out to set the cloud masking threshold for processing OCM-2. The chlorophyll-a images generated using OC2, OC3, OC4 and Chl-a algorithms for the study period are shown in Figs. 3, 4, 5 and 6 respectively. In all these figures the chlorophyll-a image generated on 22nd April 2011 shows most of the region that was masked, represented with black colour. The land portion that was masked is shown in white colour for demarcating from cloud and land pixels. The valid

Fig. 5 Spatial distribution of OCM-2 derived Chlorophyll-a concentration along the east coast of India on 10th, 12th, 20th and 22nd of April 2011 using SeaDAS software (OC4 algorithm)

J Indian Soc Remote Sens (March 2013) 41(1):117125 Fig. 6 Spatial distribution of OCM-2 derived Chlorophyll-a concentration along the east coast of India on 10th, 12th, 20th and 22nd of April 2011 using SeaDAS software (Chl-a algorithm)

123

ocean pixels were only considered for interpreting the data during the study period and the chlorophyll-a variability of those pixels are represented in colour for better understanding of chlorophyll-a distribution in the open as well as coastal ocean regions. From these figures it is clearly understood that the chlorophyll-a variability in the open ocean region vary from 0.05 to 0.1 mg m3 only and the concentration increases towards the coast. As a part of validation the in-situ measured chlorophyll-a concentration has been compared with satellite derived chlorophyll-a during the study period processed using various chlorophyll-a algorithms. These results were plotted as a scatter plot

for better understanding the algorithms performance in the coastal waters using OCM-2 data, which was shown in Fig. 7. From this figure it has been shown that the existing global ocean colour algorithms over estimates in the near shore coastal waters as well as estuarine regionHaladia. Where not only the chlorophyll concentration plays a major role like the sediment re-suspension, pigment and the shell structure which is being formed by calcification. The costal system is more optically complex because of short-term coastal upwelling, enhancement of nutrients because of river run-off and anthropogenic activity and biological process like nitrification, de-nitrification, carbonate system etc.

124

J Indian Soc Remote Sens (March 2013) 41(1):117125

Fig. 7 Comparison of in-situ Vs satellite derived chlorophyll-a concentration using different bio-optical algorithms, for surface, middle and bottom depths of the sampling locations during the study period

which has direct bearing with phytoplankton growth and their pigment characterization. Besides the area is dominated by unicellular diatoms basically contains silica. Hence the costal process becomes more complex

in terms of bio-geo chemical behavior. Hence retrieval of chlorophyll becomes erroneous in its quantification. This requires specific, suitable regional bio-optical algorithms to overcome this problem.

Table 1 Statistics showing the correlation coefficient and the other parameters for chlorophyll-a concentration validated using various chlorophyll algorithms

Parameter Mean-data no.of.pnts SD RMSE %error Slope Intercept R bias

Insitu.Vs.OC2 0.362 25 0.258 0.445 34.287 0.701 0.055 0.828 0.362

Insitu.Vs.OC4 0.349 25 0.339 0.487 42.25 1.054 0.404 0.821 0.349

Insitu.Vs.OC3 0.355 25 0.254 0.436 35.937 0.809 0.159 0.842 0.355

Insitu.Vs.cHla 0.135 25 0.452 0.472 29.526 1.318 0.462 0.817 0.135

J Indian Soc Remote Sens (March 2013) 41(1):117125

125

A rigorous statistical analysis was carried out for the in-situ estimated and the chlorophyll-a concentration estimated using the global ocean colour algorithms. For this purpose the values like standard deviation (SD), root mean square error (RMSE), % error, slope intercept, the regression coefficient (R), bias and the scatter index (SI) were calculated for all the data points collected used for validation. For this purpose the data was segregated into three sets for surface, middle and bottom depths by combining all the stations data from Paradeep, Mahanadi and Haldia regions. The statistics provided are for the surface waters only as listed in Table 1. Although the OC4 and OC3 algorithms over estimate in the coastal waters, OC2 performs better in the lower concentration range near Paradip, Mahanadi and Hugli estuary near Haldia. The total distribution of chlorophyll-a concentration in the Hugli estuary some times over estimated due to the presence of overlap or dominance of chlorophyll-b pigments will lead to mislead the estimation of chlorophyll-a concentration in this region.

develop the chlorophyll algorithm which can perform better in the coastal and estuarine waters.
Acknowledgement Our sincere thanks are due to Dr. V.K.Dhadwal, director, NRSC, Dr. G. Behera, DD, RS&GISAA, Dr.CBS Dutt, GD, ESAG and Dr. M M Ali, Head Oceanography for their support and encouragement to carry out this work. The authors would like to express their sincere thanks to the respective institutions and funding agencies for providing the infrastructure and opportunity to participate in the cruises for collecting the bio-optical data.

References
Chauhan, P., Mohan, M., Sarangi, R. K., Kumari, B., Nayak, S., & Matondkar, S. G. P. (2002). Surface chlorophyll estimation in the Arabian Sea using IRS-P4 Ocean Colour Monitor (OCM) satellite data. International Journal of Remote Sensing, 23, 16631676. Desai, B. N., & Bhargava, R. M. S. (1998). Biologic production and fishery potential of the Exclusive Economic Zone of India. In: K. Sherman, E. Okemwa, & M. Ntiba (Eds.), Large marine ecosystems of the Indian Ocean: Assessment, sustainability, and management. Blackwell Science. Dwivedi, S. N. (1993). Long-term variability in the food chains, biomass yield, and ocenaography of the Bay of Bengal ecosystem. In K. Sherman et al. (Eds.), Large marine ecosystems: stress, mitigation, and sustainability (pp. 4352). Washington: American Association for the Advancement of Science. Dwivedi, S. N., & Choubey, A. K. (1998). Indian Ocean large marine ecosystems: Need for national and regional framework for conservation and sustainable development. In K. Sherman, E. Okemwa & M. Ntiba (Eds.), Large marine ecosystems of the Indian Ocean: Assessment, sustainability, and management. Blackwell Science IOCCG Report, Remote sensing of inherent optical properties: fundamentals, tests of algorithms, and applications. Edited by ZhongPing Lee, 2006. IOCCG Report, remote sensing of ocean colour in coastal, and other optically-complex, waters. Edited by Shubha Sathyendranath, 2010. OReilly, J. E., Maritorena, Mitchel, B. G., Seigal, D. A., Carder, K. L., Garver, S. A., Kahru, M., & McClain, C. R. (1998). Ocean color chlorophyll algorithms for SeaWiFS. Journal of Geophysical Research, 103, 2493724953. Parulekar, A. H., Harkantra, S. N., & Ansari, Z. A. (1982). Benthic production and assessment of demersal fishery resources of the Indian Seas. Indian Journal of Marine Sciences, 11, 107114.

Conclusions The global ocean colour algorithms over estimate the chlorophyll-a concentration in the turbid coastal and estuarine waters. In these areas due to the influence of the riverine influxes, include the high concentrations of suspended sediments, fine suspended particles with dissolved organic and inorganic constituents makes the waters more complex for estimating or quantifying the chlorophyll-a concentration. Ratio based algorithms fail in the coastal waters for estimating the chlorophyll-a concentration and there is need to develop regional chlorophyll algorithms for better quantification. This study gives a very clear picture about the problems involved in estimating the chlorophyll-a concentration in the coastal waters and the need to develop fluorescence based techniques or inherent optical properties based algorithms for estimating the chlorophyll-a concentration. However, more of in-situ observations in the turbid coastal waters are needed to

Potrebbero piacerti anche