2.2. The AIRS and CrIS Retrieved CH4 Profiles
The AIRS science team retrieval algorithm is described in Susskind et al. (2003) [
41], including the sensor calibration, microwave (MW) first guess retrieval, cloud-clearing, initial first guess IR retrieval, and a final IR physical retrieval. The cloud-clearing step enables the use of partially cloudy measurements [
42,
43]. Validation and improvements of AIRS retrievals are described in Fetzer et al. [
44], Tobin et al. [
45], Divakarla et al. [
46], Chahine et al. [
38], Susskind et al. [
47], Nalli et al. [
48], and many other studies. Because of the AMSU-A2 instrument failure, AIRS products using both TIR and MW were discontinued after 24 September 2016. Thus, for a consistent data record, we use AIRS Version 7 IR-only retrievals for the entire record from the NASA Goddard Earth Sciences (GES) Data and Information Services Center (DISC) (
https://disc.gsfc.nasa.gov) [
49].
The NOAA Unique Combined Atmospheric Processing System (NUCAPS) is the NOAA operational retrieval algorithm for the operational hyperspectral thermal IR sounders [
50]. The NUCAPS algorithm is based on the AIRS science team retrieval algorithm Version 5 [
30,
41,
47] which differs primarily in the first guess methodology (v5 uses a linear Eigenvector regression, whereas v7 uses a neural network nonlinear regression) [
41]. The NUCAPS algorithm runs operationally at NOAA/NESDIS, with the operational products being publicly available from the Comprehensive Large Array-data Stewardship System (CLASS).
Both the AIRS and NUCAPS CrIS retrieval algorithms use spectral channels near 7.66 μm to retrieve CH
4 [
28,
29]. The CH
4 channel selection is based on the sensitivity to CH
4 as indicated by the kernel functions [
51,
52]. To reduce interference from other absorbing species, channels with overlapping absorption bands (e.g., water vapor and HNO
3), are minimized. The AIRS and NUCAPS CH
4 retrievals are in general sensitive in the range between 650 hPa and the lower stratosphere, with peak sensitivity around 300-400 hPa, depending on surface and atmospheric conditions [
53]. The CH
4 weighting function, or Jacobian matrix, describes the portion of the CH
4 profile represented by each radiance measurement [
53]. An example of the CrIS CH
4 Jacobian matrix, calculated from the Standalone AIRS Radiative Transfer Algorithm (SARTA, see algorithm description in Strow et al., 2003), is shown in
Figure 1. From
Figure 1, it is clear that the most CH
4-sensitive region is the middle to upper troposphere (i.e., ~200 - 600 hPa), as depicted in the non-red color shades in
Figure 1. The peak of the sensitivity can be higher or lower depending on the local scenario of CH
4 profiles.
Prior to the CH
4 retrieval step in the algorithm, the atmospheric vertical temperature and moisture profiles (AVTP and AVMP, respectively), surface skin temperatures, and land emissivity are retrieved using different channels than those used for CH
4 retrievals. This data, as well as the a priori profiles of CH
4, are used as inputs to SARTA forward calculation [
56]. The difference between the observed and calculated radiances (obs minus calc) is minimized to construct an error covariance matrix, from which the change of CH
4 is derived using an eigenvector transformation and damping. The final retrieved profile of CH
4 is usually obtained through several iterations. For more details on the NUCAPS and AIRS algorithm, see Susskind et al. [
41,
58], the NUCAPS Algorithm Theoretical Basis Document (ATBD) (2021), and Smith and Barnet [
30]; for the details of the CH
4 first guess and channel selections, see Warner et al. [
51], and Xiong et al. [
28], and Gambacorta et al. [
52].
The NUCAPS CH
4 retrieval has been significantly improved over the last 2 years. The major changes include updates of the CH
4 and N
2O a priori, quality control criteria, and refinements in the CH
4 channel selection. There were also other NUCAPS algorithm improvements, such as the updated TIR spectral tuning (i.e., an empirical radiance bias correction), which improved the performance of the NUCAPS temperature and water vapor retrievals, which in turn had a positive impact on the downstream CH
4 products. The CH
4 retrievals from CrIS on SNPP and NOAA-20 have gone through a thorough validation process by comparison to in-situ data and similar products from other satellite sensors [
59,
60,
61]. For the AIRS and CrIS CH
4 products, we estimated a precision of 1%. While the same retrieval approach was used for AIRS and NUCAPS algorithms, the differences in the sensors, channels, a priori, tuning, and quality control, can contribute to the differences in CH
4 retrievals between AIRS and CrIS.