Calibration and Validation of Sentinel3 OLCI Products in the Highly Dynamic Area of the Eems Dollard Estuary
| Main Authors: | Peters, Steef, Spaias, Lazaros, Poser, Kathrin, Hommersom, Annelies, Laanen, Marnix, Salama, Suhyb |
|---|---|
| Format: | Proceeding poster eJournal |
| Bahasa: | eng |
| Terbitan: |
, 2020
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/3604018 |
Daftar Isi:
- Tidal and estuarine areas pose a big challenge for validation of satellite observations of water quality parameters and reflectance. General criteria for spatial and temporal homogeneity do not apply. Still, Sentinel 3 OLCI can provide a wealth of information on the dynamical coastal processes in those areas, if the products are well calibrated and validated. The Dutch/German Eems Dollard area is such a region where many questions around climate change effects, primary production and human induced turbidity changes need to answered. Due to the high spatial expression of the tidal dynamics, remote sensing is the only viable information source to understand the processes in space and time over long time periods. Based on a comparison with in-situ reflectances, the augmented C2RCC atmospheric correction processor was preferred over the regular C2RCC processor. The difference between the 2 versions was found to be significant at 511 nm and lower wavelengths, above 511 nm both schemes produce similar results. Based on literature, the Nechad-705 algorithm (Nechad, 2010) was selected for SPM and the Gons algorithm (Gons, 2005) for Chl-a. Both algorithms use spectral information above 600 nm, so the differences in atmospheric correction schemes might not affect the retrieval results. For validation of SPM and Chl-a we used two different datasets, regular monitoring data from the Dutch ship based sampling program (MWTL) and high frequency data from a fixed position measurement station at Eemshaven. The ship based measurements were performed at optimal tide situations, and sometimes quite close to the shoreline, but not synchronous to satellite overpass times. At the fixed monitoring stations a time interval of 1 minute could be realised. From the shipborne measurements no significant correlation was found with the Sentinel 3A OLCI results. At the fixed monitoring station we found a good relationships for SPM (R2=0.72) (although with some explainable outliers) and Chl-a (R2=0.9 before outlier removal and 0.96 after outlier removal). We conclude from the validation analysis that satellite products based on Sentinel 3A observations form a reliable and time-consistent information source for estuarine situations, but validation data at a high frequency is required, both for reflectance and for water quality parameters. The new Water Insight semi-continuously registering hyperspectral instrument the WISPstation is installed on the Eemshaven platform since November 2018 and will contribute to future calibration and validation.