ADVANCES IN OPTICAL POLARIZATION REMOTE SENSING FOR MARINE OBSERVATION: A CASE STUDY IN NANCHANG RIVER PARK

Earlier databases ignore the effects of atmospheric particles on polarized land signals. Using ground data and modelling, we processed two missions and built two databases. Ignoring particles changed the surface signal by over 10 % in 60–75 % of observations. Records corresponded under similar conditions nearly a decade apart. Coarser observations showed stronger land-dependent links between surface mixing and the signal. The method supports long-term records and atmospheric particle estimates.

A global AOD product from Particulate Observing Scanning Polarimeter (POSP) has been proposed and validated using Aerosol Robotic Network (AERONET). Results show a high accuracy, with correlation coefficients (R) of 0.914, a root mean square error (RMSE) of 0.085, outperforming Moderate Resolution Imaging Spectroradiometer (MODIS). Error analysis reveals seasonal variation with lower accuracy in autumn/winter, and increased uncertainty with lower Normalized Difference Vegetation Index NDVI.

In order to conduct accurate aerosol retrieval over snow, the Remote Sensing of Trace Gases and Aerosol Products (RemoTAP) algorithm is extended with a bi-directional reflection distribution function for snow surfaces. The experiments with both synthetic and real data show that the extended RemoTAP maintains capability for snow-free pixels and has obvious advantages in accuracy and the fraction of successful retrievals for retrieval over snow, especially over surfaces with snow cover > 75 %.