Using Averaging Kernels (AKs) for Validation of IR Sounder EDRs: Application to the NOAA Unique Combined Atmospheric Sounding System (NUCAPS)

This presentation overviews the application of averaging kernels (AKs) toward the validation of IR sounder retrieved environmental data records (EDRs) versus high-resolution correlative datasets. Attention is given to the ongoing validation of operational atmospheric profile EDR retrievals obtained by the NOAA Unique Combined Atmospheric Processing System (NUCAPS) onboard the NOAA-20 and SNPP satellites. NUCAPS is a NOAA operational algorithm designed to retrieve global environmental data records (EDRs) from Cross-track Infrared Sounder (CrIS) and Advanced Technology Microwave Sounder (ATMS) sensor data records (SDRs). NUCAPS provides users with atmospheric vertical profile EDRs including temperature (AVTP) and moisture (H2O or AVMP), ozone (O3), and carbon trace gases (CO, CH4, CO2). Rigorous validation of these products are conducted with respect to various baseline in situ correlative (or "truth") datasets including dedicated radiosonde observations (ARM, GRUAN, AEROSE), ozonesonde ensembles (e.g., SHADOZ and WOUDC), and datasets-of-opportunity (e.g., AirCore). These in situ measurements typically acquire data with a much higher vertical resolution than either the 100 layer forward model calculations (e.g., SARTA or CRTM) or the resolution afforded by hyperspectral sensors themselves (viz., CrIS, IASI, AIRS). Thus, high-resolution truth measurements must first be "reduced" to the 100 layers of the forward model, thereby rendering truly "correlative" data for retrieval assessments. Such assessments then provide estimates of the total retrieval system uncertainty, coupled with the uncertainties in the truth measurement and collocation "mismatch error." This total system error also includes the so-called "null-space" error, that is, the error arising from the limitation of the retrieval vertical resolution. By applying the retrieval AKs to the correlative data, we obtained smoothed profiles that represent what the sounding system theoretically "sees" given its resolution limitations, thereby removing these null-space errors. The presentation will include an overview of the NUCAPS "effective AKs" along with examples in application to NUCAPS T/H2O/O3 and carbon trace retrievals.