| Cyanobacteria is the main dominant algae specie of blooms in inland eutrophic water. The abundance of cyanobacteria is of great significance to the monitoring and early warning of cyanobacteria blooms. Monitoring cyanobacteria abundance by remote sensing technology can be macroscopic,fast and real-time,which can reflect whether the dominant algae is cyanobacteria and the temporal and spatial variability of cyanobacteria abundance. Therefore, it is of great significance to develop the remote sensing monitoring model of cyanobacteria abundance. In this paper, a typical inland eutrophic lake, Taihu Lake, was used to study the relationship between cyanobacteria abundance and different bio-optical quantities, including apparent optical quantity(remote sensing reflectance), intrinsic optical quantity (phytoplankton absorption coefficient) and pigment (PC: Chla),the difference of optical characteristics of different cyanobacteria abundance was analyzed, and the remote sensing estimation method of cyanobacteria abundance was studied. On the basis of these analysis, the partial least squares estimation model of cyanobacteria abundance based on MERIS data and the semi-analytical estimation model of cyanobacteria abundance were developed. The main research contents and conclusions include the following four aspects:(1) The relationship between the cyanobacteria abundance and the bio-optical quantities.Based on the analysis of the relationship between cyanobacteria abundance and remote sensing reflectance (Rrs), phytoplankton absorption coefficient (aph) and pigment concentration (PC: Chla), the differences in samples with different cyanobacteria abundance were found in remote sensing reflectance and phytoplankto n absorption coefficient. The difference of remote sensing reflectance of different cyanobacteria abundance was mainly in the range of 560nm ～700nm, and the depth of reflection valley at 624nm was increasing with the increase of cyanobacteria abundance.The absorption coefficient of phytoplankton had a more obvious indication of cyanobacteria abundance. The band ratio indexs of the absorption coefficient of phytoplankton [aph(624) - aph(550)]/[aph(624)+aph(550)] and [aph(665) -aph(624)]/[aph(665) + aph(624)] and the band ratio index of the absorption coefficient of unit phytoplankton [aph*(624) - aph*(550)]/[[aph*(624) +aph*(550)] and [aph*(665) - aph*(624)]/[[aph*(665) + aph*(624)] had a good correlation with cyanobacteria abundance, which can indicate cyanobacteria abundance preferably. The correlation between cyanobacteria abundance and [aph(624)-aph(550)]/[aph(624) + aph(550)] and [a*ph(624) - aph*(550)]/[[aph*(624) +aph*(550)] was more closely related. However, the study found that the PC: Chla had a poor correlation with cyanobacteria abundance, PC: Chla can’t be used to indicate the abundance of cyanobacteria.(2) The developing of empirical estimation model of cyanobacteria abundanceThe hyperspectral remote sensing reflectance data of the field were simulated to 15 bands of MERIS, based on the relationship between cyanobacteria abundance and remote sensing reflectance, the fifth bands (the center wavelength is 560nm) ,the sixth band (the center wavelength is 620nm) , the seventh band (the center wavelength is 665nm), the eighth band (the center wavelength is 681nm) and the ninth band (the center wavelength is 709nm) of simulated MERIS bands were selected to develop an empirical estimation model of cyanobacteria abundance, the root mean square error was 8.58, the average relative error was 14.09%, the prediction accuracy of the model was good,which can better estimate the cyanobacteria abundance of the water sample.(3) The developing of semi-analysis algorithm for cyanobacteria abundanceBecause the empirical model is relatively dependent on the experimental data set,so a semi - analytical estimation model of cyanobacteria abundance by remote sensing was proposed on the basis of analyzing the difference in optical characteristics of different cyanobacteria abundance samples. The absorption of phycocyanin at 620 nm apc (620) and the absorption of chlorophyll a at 665 nm achl(665) were obtained by isolating the absorption coefficient of the phytoplankton (aph). And based on the relationship between cyanobacteria abundance and apc (620) / achl (665), a semi -analytical estimation model of cyanobacteria abundance based on Rrs (620), Rrs (665),Rrs (709) and Rrs (778) was developed. The mean square error of the model was 19.57,and the averag relative error was 41.37%. Although the accuracy of the semi-analytical algorithm was slightly less than that of the partial least squares algorithm, the semi-analytical algorithm has more physical basis and significance, its generality in inland lakes has better potential.(4) The distribution and seasonal variation of cyanobacteria abundance in Taihu Lake regionThe two - step method can be used to estimate the abundance of cyanobacteria in Taihu Lake region. First, the spectral shape SS (665) was used to determine the existence of cyanobacteria. Then, the inversion of cyanobacteria abundance in TaihuLake region was realized based on MERIS data and the partial least squares model. In time dimension, the abundance of cyanobacteria in different seasons was different, and the abundance of cyanobacteria was higher in summer and autumn. The abundance of cyanobacteria in spring and winter was at a low level. In spatial dimension, the cyanobacteria abundance in different lakes district was different, and the cyanobacteria abundance in Xukou Bay and East Taihu Lake is relatively stable, which was at a low level in four seasons. The cyanobacteria abundance from Zhushan Bay, Meilang Bay and the Northwest Lake Showed a tendency to increase - decrease - decrease from spring to winter. The abundance of cyanobacteria from Gonghu Bay, the Central Lake and the Southwest Lake Showed a tendency to flat - increase - decrease from spring to winter. |
