Bio-optical retrieval algorithms based on blue-green reflectance ratios suffer severely in the coastal waters due to inappropriate atmospheric corrections and spectral interferences from organic and inorganic components. Furthermore, in coastal waters, standard near-infrared (NIR) atmospheric correction algorithms also often fail because of higher turbidities which result in increased elastic reflectance and significant radiance contributions in NIR bands. Recently, an atmospheric correction algorithm using the short-wave infrared (SWIR) bands has also been applied to turbid coastal waters. In this study we examine the performance of our recently proposed bloom detection technique, based on differences in selected red band spectra (RBD) and our toxic dinoflagellate Karenia brevis (K. brevis) bloom classification technique, with a designated K. brevis bloom index (KBBI) for use with Moderate Resolution Imaging Spectroradiometer (MODIS) data, corrected for the atmosphere, using NIR and SWIR atmospheric correction algorithms. Our analysis shows that both atmospheric correction algorithms are unsatisfactory, giving negative normalized water-leaving signals at the 412nm band for the bloomed area, which is an indication of failure of atmospheric correction, meaning that the signal is over corrected for the atmosphere. Although SWIR seems to give smaller negative values than the NIR in the bloomed region, both of the atmospheric correction failed in K. brevis bloomed areas. These negative values are primarily due to the fact that waters containing large accumulations of K. brevis species have a relatively strong water-leaving radiance in the near-infrared bands, while at the same time are absorbing strongly in the blue-green region, which leads to possible errors in the atmospheric correction and underestimation of in the blue and green bands of MODIS. Reflectance band ratio algorithms applied to this inappropriately atmospherically corrected signal gives different and inaccurate results with either atmospheric correction algorithm. Results for the RBD bloom detection technique, which we have previously defined as are found to be similar with either atmospheric correction algorithm. This is due to the fact that for this technique; we are using the difference in magnitude of the reflectance signal at two adjacent red bands, and since these two bands are relatively close spectrally 667nm and 678nm, the magnitude of the optical impact of the atmosphere will be very nearly the same on either band, and thus will have no impact on the difference of the two magnitudes. This is in marked contrast to the impact of the atmosphere on the ratio of signal magnitudes at each band (which would be significantly altered and dependent on the absolute magnitude of the atmospheric correction being correct, which would be near impossible to attain, for a ratio retrieval technique to retain its validity in retrievals). Our simulations for K. brevis, which is known to be characterized by weak backscatter, both because its own backscatter is low because of its low index of refraction, and because of the typically low cohort submicron particulate concentrations typically associated with it, show that the correlation of K. brevis chlorophyll concentration with RBD values departs from linearity as Chl concentrations increase. It may be possible to evolve empirically based relationships between RBD and K. brevis Chl concentrations. The near insensitivity of the RBD technique to atmospheric correction and less sensitivity to CDOM may enable us to retrieve chlorophyll more accurately than the blue-green reflectance ratio algorithms for low backscattering blooms such as K. brevis that blooms regularly in Gulf of Mexico. When we apply atmospheric correction algorithms to the KBBI technique, which uses the normalized difference of the two bands, and for which we have previously defined KBBI by and used it for distinguishing K. brevis bloomed waters from non-K. brevis bloomed waters, both atmospheric correction algorithms give slightly different KBBI values. However, it was still possible using either atmospheric correction to identify K. brevis bloomed areas. While in general, the KBBI technique could identify potential K. bervis bloomed area using data corrected with either atmospheric correction, the SWIR algorithm does more poorly in the offshore pixels and often gives noise values of KBBI (positive and negative false bloom alarm in nearby pixels). This is because the MODIS SWIR bands are designed for the land and have substantially lower signal-to-noise ratio (SNR) values. So the signals received in the SWIR bands in offshore pixels are low, and often within the noise level. Thus the overall performance of NIR atmospheric correction algorithm is found to be better for use with KBBI than the SWIR atmospheric correction algorithm, with the exception that NIR gave more false positive bloom alarms at the cloud edge pixels and spurious results for the stripe regions at the ends scan lines.