Recovering low quality MODIS-Terra data over highly turbid waters through noise reduction and regional vicarious calibration adjustment: A case study in Taihu Lake

• Published in: Remote sensing of environment Vol. 197; pp. 72 - 84
• Main Authors: Li, Junsheng, Hu, Chuanmin, Shen, Qian, Barnes, Brian B., Murch, Brock, Feng, Lian, Zhang, Minwei, Zhang, Bing
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.08.2017
• Subjects: Atmospheric correction; Chaohu Lake; MODIS; Noise reduction; Ocean color; Remote sensing; Remote sensing reflectance; Shortwave infrared; Taihu Lake; Atmospheric correction; Noise reduction; Chaohu Lake; Ocean color; Shortwave infrared; Remote sensing reflectance; Remote sensing; MODIS; Taihu Lake
• ISSN: 0034-4257; 1879-0704
• DOI: 10.1016/j.rse.2017.05.027

Remote sensing of water quality in turbid coastal and inland waters requires accurate atmospheric correction, which is technically challenging. While previous efforts have shown the advantage of using the short-wave infrared (SWIR) bands instead of near-infrared (NIR) bands for atmospheric correction, such an approach could only be applied to the Moderate Resolution Imaging Spectroradiometer (MODIS) on the Aqua satellite (MODISA). This is because MODIS data from the Terra satellite (MODIST) contain more noise and other sensor artifacts, thus this sensor has been generally regarded by the ocean color research community as not being able to provide science quality data. Here, we address this technical challenge through noise reduction and regional vicarious calibration adjustment, and demonstrate preliminary success using turbid Taihu Lake as an example. The noise in the three SWIR bands was evaluated first, and then reduced through a noise reduction method. The SWIR bands were adjusted over open-ocean waters using the well-calibrated NIR ocean bands (1-km resolution) and radiative transfer, which were then used to adjust the land bands (250-m and 500-m resolution) in the visible and NIR over turbid waters where concurrent field-measured reflectance spectra are available. Of all three combinations of SWIR bands, the combination of 1240 and 1640-nm bands was found to perform the best, showing significantly improved retrieval accuracy for Taihu Lake, leading to recovery of low-quality MODIST data to higher-quality data comparable to MODISA, and thus doubling valid data coverage. Testing of this approach on another highly turbid lake (Chaohu Lake, China) showed similar results. While the general application of this approach to turbid lakes still needs to be tested as local tuning of the calibration coefficients may be required, these results suggest that MODIST may be used as effectively as MODISA for monitoring Taihu Lake water quality. •MODIS/Terra noise in short-wave infrared bands reduced to increase SNR•Calibration adjustment performed over both MODIS/Terra SWIR and VIS-NIR bands•Significantly improved atmospheric correction accuracy over turbid Taihu Lake•May be applicable to other waters with similar in-water and atmospheric conditions