Researchers are evaluating the performance of MODIS land reflectance products in monitoring water

Satellite ocean color tools are used to characterize physical, chemical and biological variables in oceanic, coastal and inland waters. However, the huge loss and uncertainty in remote sensing reflectance data lead to difficulty in monitoring coastal and inland nearshore waters.

Most coastal and inland waters are generally turbid. In these regions, the contribution of water to the upper atmosphere signals is elevated, indicating that the Earth’s reflection products can be used to monitor water. The surface reflectance product of the Moderate Resolution Imaging Spectroradiometer (MODIS) (R_land) was used to monitor the clarity of the water and the suspended solids in the water.

Recently, a research group led by Prof. Duan Hongtao and Ma Ronghua of the Nanjing Institute of Geography and Lakes of the Chinese Academy of Sciences presented a comprehensive evaluation of the performance of MODIS R_land products against a field visual dataset. The dataset contains 4,143 reflectance spectra, 2,320 chlorophyll-a samples, and 1,467 particulate matter samples from the coast near the global shore and inland waters.

This work was published in Earth Science Reviews.

“Although this product is easier to use and has higher spatial resolution than MODIS ocean scales, according to our assessment, R_land may not be an ideal data source for monitoring inland and coastal waters,” said Professor Duan.

The results showed that R_land overestimated it Remote Sensing reflectance, particularly in the 469 nm and 859 nm bands. In addition, the land reflectivity showed a clear overestimation compared to ocean color products derived using the SeaDAS program in the East China Sea.

The researchers also reported significant negative values ​​and discontinuities in the R_land images. R_land was negative in pixels covered by cyanobacterial scum. The negatives and choppiness in R_land are likely to be caused by [an] An inappropriate mechanism for aerosol removal in R_land generation over water,” said Prof. MA.

Current algorithms have not estimated chlorophyll-a disease and suspended solids from R_land across global inland and coastal waters. Machine learning models have outperformed state-of-the-art algorithms to retrieve suspended particles in the global murky waters of R_land.

However, not all models, including machine learning models, can reliably retrieve chlorophyll from R_land, due to limited spectral information and uncertainty in R_land products. This means that R_land may be able to identify parameters closely related to suspended solids (eg, water purity and extinction coefficients) in most waters; However, pigments such as chlorophyll a are difficult to identify.

“MODIS R_land does not contain enough information to make these current algorithms usable. Hence, many water color parameters are difficult to retrieve from R_land except for several parameters, such as suspended particles, turbidity, and Clarity of waterProfessor Duan said.