In this work, the NOAA Unique Combined Atmospheric Processing System (NUCAPS) CrIS/ATMS sounding products from Suomi-NPP are assessed using reference and conventional radiosonde observations. Reference radiosondes are those processed using software developed in coordination with the GCOS Reference Upper Air Network (GRUAN). These includes subsets of JPSS-funded satellite synchronized radiosondes launched at U.S. DOE Atmospheric Radiation Measurements (ARM) sites and in special field campaigns such as NOAA Aerosols and Ocean Expeditions (AEROSE), CalWater 2015 ARM Cloud Aerosol Precipitation Experiment (ACAPEX) over the north Pacific Ocean and the 2016 El Nino Rapid Response Experiment (ENRR). Conventional RAOBs on the other hand are typically launched twice a day at about 850 sites globally and are used for numerical weather prediction (NWP) assimilation and forecasting. Conventional and Reference RAOBs are complementary in support of the JPSS sounding EDR calibration/validation (cal/val), and respectively serve as anchors for compiling multiple satellite product collocations in the NOAA Products Validation System (NPROVS) and its expansion NPROVS+. Validation uncertainties arising from temporal and spatial mismatch between RAOB launch and satellite overpass, potential RAOB accuracy issues, and inconsistencies in radiosonde versus satellite product vertical resolution inconsistencies are included in discussions. Inter-satellite product comparison versus the radiosondes and new procedures to integrate the GRUAN reference radiosonde uncertainty in cal/val are also presented.