Performance Assessment of the SNPP VIIRS Cloud Optical Thickness, Effective Particle Size, Cloud Top Height and Cloud Top Temperature Products

In this paper we will present the performance estimates for the VIIRS Cloud Optical Thickness (COT), Effective Particle Size (EPS), Cloud Top Height (CTH) and Cloud Top Temperature (CTT) obtained from comparison with Calipso CALIOP, NOAA and MODIS cloud products. In general we made comparison of the NPP products with the Calipso and Aqua MODIS cloud products at the same geolocations where they could be viewed by VIIRS and the other sensor within 10 minutes. However, comparisons of COT and EPS for day conditions were performed with cloud products generated by the NOAA AWG algorithms using direct input from the NPP SDR radiances. In this way, comparisons were made without errors introduced by the cloud mask and the cloud reflectance look-up-tables.

In the comparison with Calipso 5-km COT products multiple NPP pixels were matched up by geolocations to one single Calipso 5 km-cell. The NPP cloud properties of these pixels were then aggregated to yield an equivalent cell-level product. This was done to ensure consistency with the Calipso 5-km products which were generated by aggregating data collected from multiple laser pulses along the track.

SNPP VIIRS night time COT and EPS products represent the first cloud property products ever produced operationally for night conditions. However, the performance expectation for these night products is not high for the lack of algorithm maturity and furthermore for the limitation of the IR bands that have little sensitivity in COT and EPS. Indeed, the IR method for COT retrievals ceases to be accurate when COT > 5. Therefore comparison of night COT was limited to semi-transparent clouds of COT < 5. The performance estimates for these clouds were obtained with Calipso COT products and MODIS COT inverted from the cloud emissivity products.

Performance assessment for night EPS is difficult due to the lack of operational night EPS products and field campaign data. One of the approaches we had adopted was by making comparison with the day time EPS retrievals. Their night counterparts were generated by running a modified version of the night COP algorithms with IR radiances input from the day SDR. Finally we will describe several notable updates made to the cloud algorithms and present the resulting performance improvements to the cloud products.