Quantiative Analysis Of Suspended Sediment And Chlorophyll Concentration Using MERIS Data In Taiwan Strait

MEdium Resolution Imaging Spectrometer (MERIS) on board the ENVISAT-1 is one of the most advanced ocean color sensors. It has the virtues of high temporal resolution, high spatial resolution, high spectral resolution and high sensitivity, and is widely applied. Based on the introduction of ocean color mechanism, a processing flow involved MERIS data pre-processing and estimation of suspended sediment (SSC) and chlorophyll (Ch l) concentration in Taiwan Strait was proposed. The main contents are as follows:(1) MERIS data pre-processing. After analyzing the basic component of MERIS L1B data sets, MERIS data pre-processing method was introduced. This included interpolation of tie point data sets, extracting area of the ocean and cloud removal.(2) Models for calculating SSC concentration.Historical data whose particle size distribution is similar to that of Taiwan Strait was selected as in situ data. Bands most sensitive to SSC concentration variation were chosed to establish the quantitative remote model in Taiwan Strait, which included the logarithmic expressions of single band(510nm or 620nm) reflectance , logarithmic expressions of multi-band ratio R₅₁₀/(R₄₉₀/R₅₆₀) or R₆₂₀/(R₄₉₀/R₅₆₀), and spectral mixing analysis models using two bands(510nm and 620nm) or three bands (510nm,560nm and 620nm) and semi empirical model. Results shown that multi-band ratio reflected the SSC distribution in Taiwan Strait well. The reasons may be that this multi-band ratio would reduce the impact of atmospheric correcion and the ratio of (R₄₉₀/R₅₆₀) partly eliminates the impact of chlorophyll.(3) Models for calculating Chl concentration. Six mostly offen used empirical models, like OC4V4, OC2,OC2V2, OC2V4, CaCOFIcubic and CaCOFI_linear were used to calculate the distribution of Chl concentration in Taiwan Strait. The neural network model provided by ESA (Europe Space Agence) was also adopted, and its results were compared to those of the six empirical models. Results shown that in coastal waters, Chl concentration acquired by the empirical model based on the blue/green band ratio was a little higher than historical in situ data, but in off-shore area, the result matched better. The Chl concentration calcutated from neural network model reflect chlorophyll distribution well and the results are more reasonable than those of empirical models. For Chl estimation, the...