NOAA/NESDIS has provided suites of derived atmospheric sounding products for polar orbiting satellites since the onset of TOVS back in 1979. These legacy products have evolved and now include separate suites of operational ATOVS and MIRS sounding products from the current suite of ATOVS sensors onboard NOAA-18 and MetOp satellites. Each of these product suites is based on various retrieval approaches and assumptions, particularly with respect to the computation of the first guess and sensor measurement adjustments to optimize the retrieval solution. The following paper presents a revised approach for computing first guess profiles, sensor measurements adjustments, and retrieval solutions for ATOVS operational soundings that are based on a carefully compiled dataset of collocated radiosonde and satellite observations. Results are evaluated and serve as the basis for the design of a merged MIRS / ATOVS derived products system which integrates the scientific advantages of each system.

A focal point of the report is the series of results which compare the respective performances of the three (3) candidate derived satellite product systems based on the Noaa PROducts (integrated) Validation System (NPROVS). NPROVS currently provides a centralized NOAA function which compiles collocations of multiple satellite, ground truth, and NWP observations on a daily basis and which includes a variety of quality control and sampling options for analyzing results. The revised ATOVS “test” sounding system uses the NPROVS collocations as the basis for computing respective first guess, radiative transfer (RT) bias adjustment, and retrieval solution coefficients. Results include specific comparisons of first guess, RT adjusted sensor data, and final retrieval as available from each system. These results are used to construct a model for a merged MIRS/ATOVS scientific approach which integrates the scientific advantages of each method. Results from the merged system are reported as available.

Strategies and quality control procedures for compiling the global sample of collocations and for generating the respective first guess, radiative transfer (RT) bias (sensor) adjustment, background error, and noise covariance matrices for the “test” ATOVS are included. Special cases of anomalous sensor data that are difficult to retrieve are presented, for example, as observed for Microwave Humidity Sounder (MHS) in comparatively hot and dry Saudi Arabia versus the cold and dry Antarctic regions. The use of NPROVS to analyze and develop strategies for the RT bias adjustment of the sensor measurements, for example, in conjunction with the retrieval of upper tropospheric moisture is presented.

The goal is the development of a single operational ATOVS derived product system which utilizes the complete suite of HIRS, AMSU, AVHRR and MHS sensors comprising ATOVS. A key ingredient is the independence from NWP and (thereby) increased potential for “meaningful” impact in real-time weather and climate applications. The integrated MIRS/ATOVS products are considered a viable candidate for use as a first guess approach in planned next-generation hyper-spectral infrared soundings for NPOESS.