|
|
|
March 2003 |
Estimating Chlorophyll-a Concentration
Image satellite data
For SeaWiFS data, are uses in this study are Level 1A (LAC) data dated on 24th August 2000 and 29th August 2000 with FTP format can be downloaded from the Internet. Both images are cloudy and only apart of sampling point can be further analysed.
Methology
This study can segregated into two parts, firstly, chlorophyll-a sampling and analysis and secondly to map the chlorophyll-a concentration from SeaWiFS data.
Laboratory analysis
Seawater samples were taken using Van Dorn Sampler for maximum chlorophyll-a layer according the information from acoustic equipment. Chlorophyll-a concentrations were measured in the laboratory.
The chlorophyll-a concentrations were estimated by using the spectrophotometer 6300 Jensey at the Aquatic laboratory. The chlorophyll-a concentration was measured by using the technique and calculation described by Parsons et.al (1984). Five litres of seawater were filtered through a filter paper (Millipore, size 0.5 mm). As the seawater is being filtered, a few drops of suspension of magnesium carbonate were added to prevent acidity on the filter. Pigments were extracted from the filters in 90% acetone. The wavelengths involved are 750nm, 664nm, 647nm, 630nm.
All the extinction was corrected for a small turbidity blank by subtracting the 750nm from the 664nm, 647nm and 630nm absorptions. Then, the amounts of pigments in the original seawater sample were calculated using the equation given below.
For Chlorophyll-a(Ca)
Ca = 11.85E664 - 1.54 E647 - 0.08E630 (3.1)
Where:
E stands for the absorbance at different wavelength obtained above (Corrected by the 750 nm reading). | Ca is the amount of chlorophyll in mg/ml
Chlorophyll - a is Chlorophyll-a concentration, and its obtained from the following equation:
Chlorophyll - a(mg / m3) = C x v/Vx10 (3.2)
Where:
C are substituted for Chlorophyll-a from equation (3.1) | v is the volume of
acetone in ml (10ml) | V is the volume of seawater in l (5l)
Image Processing
Image processing for SeaWiFS data in this study, was processed using an image processing software PCI Easi/Pace Version 8.0 and ENVI Version 3.6. The main operation in this study can be categorized onto four types of processing. There are geometric correction, radiometric correction, chlorophyll-a extraction and finally mapping chlorophyll-a concentration.
Chlorophyll-a Extraction
Based from curve fitting, a linear regression analysis was carried out between water parameters and spectral radiances were performed and their significant level examined. In this study, spectral attention was given to channel 2 (445nm), channel 3 (490nm) and channel 5 (555nm). Linear regression technique was applied between ratio of reflectance values and chlorophyll-a concentration samples from in-situ sampling. The best
correlation coefficient (r2), would be used to extract the distribution of chlorophyll-a concentration.
Results
Different algorithms were applied to the images. The in-situ data and the measured chlorophyll-a was highly correlated using empirical algorithm using level 1 data with r2 = 0.9472. This algorithm could be applied to that particular site. Table 1 shows the correlation between In-situ chlorophyll-a and measured chlorophyll-a extracted from models. SeaBAM algorithm (NASA, 1997) was modified to suit with the local condition and produced good correlation with in-situ data r2 =0.924. Morel model (Morel, 1996) produced r2 =0.8589 between In-situ chlorophyll-a and measured chlorophyll-a extracted from models.
Discussion
The results show that the empirical model has significantly highest correlation to the in-situ data. SeaWiFS level 1 data gives correlation of r2 = 0.6882 and level 2 data gives correlation of r2 = 0.6677. The ratio between channel 2, channel 3 and channel 5 is a good combination to extract chlorophyll-a from SeaWiFS data. For SeaWiFS data, ratio derived using blue channel (443nm), blue-green channel (490nm) and green channel (555nm) was used to extract the chlorophyll-a concentration from SeaWiFS data. The ratio of was used for implementing the empirical algorithm (linear regression) and Morel algorithm. For SeaBAM algorithm, the ratio of was applied. The Morel and SeaBAM algorithms were modified to suit with the Tropical area. Details discussion can be found in Arnis (2001). Fig. 1(a)(b) and Fig. 2(c)(d) show the chlorophyll-a concentration map. From the map, the highest chlorophyll-a concentrations are found in the coastal waters of Terengganu and decreased to offshore. In short, remote sensing technique with suitable extracting chlorophyll-a algorithm offers a useful technique for estimating of chlorophyll-a concentration.
Conclusion
Remote sensing technique is a very useful tool for studying the distribution of chlorophyll-a concentration in a large water body area such as the Exclusive Economic Zone. In this work, channel 2, channel 3 and channel 5 of SeaWiFS data have been found to be the most suitable channel to extract the chlorophyll-a concentration from SeaWiFS data. Correlation analysis between remotely sensed data and chlorophyll-a in-situ data has indicated the possibility of mapping chlorophyll-a concentration with some degrees of success. The strong correlation of radiance ratio corresponding to above channel with in-situ data provided the basis for the development of equation and constant for the estimated chlorophyll-a concentration in South China Sea. However, the use of satellite remote sensing for mapping chlorophyll-a concentration in South China Sea is limited by the presence of cloud cover. Still, satellite data is preferable to filed measurements if one aim is to follow the temporal of phytoplankton over large area.
|
|
|
|
|
|
|
