Effects of atmospheric correction on extracting cyanobacteria bloom information based on remote sensing indices

Abstract

Accurate maps of the spatial and temporal dynamics of cyanobacteria blooms are urgently needed in the Taihu Lake,which is a drinking water resource for cities around the lake. Satellite imagery can be used as a cost-effective method for remotely monitor trends in cyanobacteria blooms. However,atmospheric effects and sun-target-satellite geometry can make multi-temporal images of blooms inconsistent with each other and cause uncertainties in bloom data extraction. In this paper,four remote sensing approaches were applied to retrieve cyanobacteria bloom information in the Taihu Lake during the whole year of 2006. These approaches included the near infrared( NIR) single band data,the ratio vegetation index( RVI),the normalized difference vegetation index( NDVI),and the normalized difference water index( NDWI). Two kinds of MODIS( moderate-resolution imaging spectroradiometer) products,i. e.,the top-of-atmosphere( TOA) radiance images without atmospheric correction( MOD02) and the surface reflectance images with atmospheric correction( MOD09),were selected as the data source. Furthermore,three factors comprising the aerosol optical thickness( AOT),the solar zenith angle,and the sensor zenith angle were chosen as indicators of radiation transfer processes to evaluate their influence on the remote sensing indices during the extraction of cyanobacteria bloom information. Specifically,the relationships between retrieval threshold values and the three indicators were analyzed to evaluate the temporal influences quantitatively. The results showed that: 1 these three factors had more impact on NIR single band data and the NDWI,and less impact on the RVI and NDVI( RVI was less sensitive than NDVI in regard to the atmospheric factors); 2 both AOT and the solar zenith angle were positively correlated with the threshold values. Whether or not these relationships hold water for other cases needs to be further examined. It is thus held that these four remote sensing approaches should be used carefully for monitoring cyanobacteria blooms when atmospheric correction is not applied.