In NOAA STAR’s land product team, we have operational land surface reflectance product from both GOES-16 and GOES-18 Advanced Baseline Imager (ABI) sensors. The consistency between GOES-16 and GOES-18 surface reflectance is vital for maintaining the integrity of a seamless and reliable data record to support long-term climate analyses. Surface Reflectance can be described by the Bidirectional Reflectance Distribution Function (BRDF), which is a function of the Solar Zenith Angle (qS), the View Zenith Angles (qV), and the Relative Azimuth Angle (f). In this way, for a specific surface pixel in the overlapping region, two Surface Reflectances from different qV can be obtained at the same time under a given qS. In the past 30 years, many studies have proved that Surface Reflectance anisotropy may be caused by the different qV, which is a major source of discrepancy that cannot be neglected in surface reflectance inter-comparison. It is called the directional effects on the Surface Reflectance, or the so-called BRDF effects. Although the method of BRDF effects correction has been proposed, previous studies are mainly based on the Moderate Resolution Imaging Spectroradiometer (MODIS) data or airborne platform observations. Thus, understanding the directional effects is vital for inter-comparison or data fusion from GOES-16 and GOES-18 Bidirectional Reflectance Factors (BRF).
To this end, an experiment is designed over pixels with valid BRDF retrievals and covering different surface types. All five shortwave channels (1,2,3,5,6) are compared over different seasons. The operational GOES-16 BRDF parameter datasets are used to estimate BRFs, which are applied to assess the BRDF effects between GOES-16 and GOES-18 observations. It would provide essential reference information for sensor data comparison, calibration, and joint application. Three major steps are described in brief.
First, the coincident pixel pairs between GOES-16 and GOES-18 are selected based on their Full Disk navigation datasets. Second, BRFs at 1900 UTC are estimated, considering the lowest qS at that time in the overlapping region. In this step, qS and f at 1900 UTC are calculated by a well-designed theoretical model and the Ross-Thick-Li-Sparse kernel is applied to calculate the corresponding kernel matrix. Finally, BRFs of five shortwave channels for both satellites are compared in spatiotemporal scale. To eliminate the impact of terrain altitude, we focus on the comparison in the plain within the ranges of 30ºN~40ºN and 90ºW~100ºW. The metrics of Pearson correlation coefficient (PCC), bias and precision are calculated to quantify the consistency of BRFs from the two satellites.
Preliminary results show that major BRF sample pairs concentrate along the diagonal line closely for each band in the scatter plot. To the time series of metrics, the overall BRDF effects show a significant decrease from January to March, and then their trends keep stable. Take the precision as an example, its value drops down sharply from 0.02 to 0.008 from January to early March, and then oscillates between 0.008 and 0.006. Three typical days in winter, spring, and summer are selected to analyze its influence on BRF comparison in different seasons. Generally, BRFs from GOES-18 are higher than those from GOES-16 in winter while BRFs are much equivalent between the two satellites in the other two seasons. In conclusion, BRFs from GOES-16 and GOES-18 are consistent to some extent and the BRDF effects has smaller impact in Surface Reflectance under low terrain and snow-free conditions. Meanwhile, a longer time series of BRDF parameter datasets is necessary for better understanding the seasonal variation of BRDF effects.

