Generating Diurnal to Seasonal Bidirectional Reflectance Factors from GK-2A Geostationary Satellite

Abstract:

Satellite derived surface reflectance products are the key inputs to downstream products such as vegetation indices, leaf area index and canopy photosynthesis. Geostationary satellite offer new opportunities to track surface reflectance at very fine temporal scales across seasons, which have been not explored fully. Here, we use a recently launched GEO-KOMPSAT-2A (GK-2A) satellite data to retrieve diurnal to seasonal GK-2A/AMI bidirectional reflectance factor (BRF) which is evaluated against multi-angle field measured surface reflectance data from hyperspectral imager attached to an unmanned aerial vehicle (UAV). We use RossThick-LiSparse Reciprocal BRDF model which is used for Moderate Resolution Imaging Spectroradiometer (MODIS) to retrieve BRDF parameters after applying an atmospheric correction with 6S model. Diurnal GK-2A/AMI surface BRF is compared with in-situ and UAV BRF data which were measured at the same sun-sensor geometry of GK-2A/AMI at specific cloud-free dates in each season over a heterogeneous rice paddy landscape. We also develop two GK-2A/AMI surface reflectance products for seasonal comparison with multi-satellite data. First, GK-2A/AMI daily BRF by averaging observations in each day from10 am to 2 pm local solar time with GK-2A’s own view zenith angle. Second, GK-2A/AMI daily Nadir BRDF Adjusted Reflectance (NBAR) with nadir view zenith angle and mean solar zenith angle during composite period (10 am to 2 pm local solar time). Seasonal Impact of changing GK-2A/AMI sun-sensor geometry on seasonal profile of each VIS/NIR band reflectance and vegetation indices are compared with MODIS and ground observations. We found good agreements of NBAR between MODIS and GK-2A/AMI across seasons (R²=0.68, RMSE = 2.4%; red band, R² = 0.88, RMSE = 2.4%; NIR band). On daily average, GK-2A/AMI achieve 27% more data retrievals than MODIS (MCD43A4) during the cloudy summer season. Furthermore, we present diurnal BRF variations could be greater than seasonal variations of NBAR, highlighting normalizing angular effects is an essential step to apply geostationary satellite for monitoring ecosystem dynamics.