All-sky ATMS radiance data assimilation with MPAS-JEDI

Ban, Junmei; Liu, Zhiquan; Jung, Byoung-Joo; Banos, Ivette Hernandez; Ruston, Benjamin; Collard, Andrew

doi:10.5194/egusphere-2026-1047

Preprints

https://doi.org/10.5194/egusphere-2026-1047

Preprints

06 Mar 2026

| 06 Mar 2026

All-sky ATMS radiance data assimilation with MPAS-JEDI

Junmei Ban, Zhiquan Liu, Byoung-Joo Jung, Ivette Hernandez Banos, Benjamin Ruston, and Andrew Collard

Abstract. This study extends the all-sky radiance data assimilation capability in MPAS-JEDI (data assimilation system for the Model for Prediction Across Scales-Atmosphere based upon the Joint Effort for Data assimilation Integration), previously implemented for the Advanced Microwave Sounding Unit-A (AMSU-A), to the Advanced Technology Microwave Sounder (ATMS). Compared with AMSU-A, ATMS covers a broad frequency range, including high-frequency humidity-sounding channels, in addition to the temperature-sounding and low-frequency imager channels. In contrast to the previous AMSU-A implementation, which assimilated only imaging channels under all-sky conditions, this work assimilates all ATMS channels using the all-sky approach. A situation-dependent observation error model is employed, with distinct cloud predictors over land and ocean surfaces for both temperature- and humidity-sounding channels. The analysis variables, radiance observation operator, and bias correction method are inherited from the AMSU-A all-sky assimilation. The impact of assimilating all-sky ATMS radiances is evaluated with three month-long global hybrid three-dimensional ensemble-variational (hybrid-3DEnVar) experiments: a benchmark experiment without ATMS data, an experiment assimilating only ATMS temperature-sounding channels, and an experiment assimilating all ATMS channels. The 6-hour background forecasts during the assimilation cycling and extended 5-day forecasts are verified against conventional observations, satellite radiances, and Global Forecast System (GFS) analysis. The results show that the background fits to radiosonde observations, satellite radiances, and GFS analyses have improved. Forecast verifications against GFS analyses and independent radiance observations demonstrate statistically significant improvements relative to the benchmark for up to 3 days in both ATMS experiments, across dynamic, thermodynamic, moisture, and cloud fields.

Received: 24 Feb 2026 – Discussion started: 06 Mar 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Junmei Ban, Zhiquan Liu, Byoung-Joo Jung, Ivette Hernandez Banos, Benjamin Ruston, and Andrew Collard

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-1047', Anonymous Referee #1, 07 May 2026
This study extended the all-sky radiance DA capability in MPAS-JEDI to ATMS based on previously AMSU-A all-sky DA work. ATMS temperature-sounding channels, humidity-sounding, and window channels are assimilated using all-sky approach. Three month-long cycling hybrid-3DEnVar experiments were conducted to evaluate the impact for all-sky ATMS DA. The all-sky ATMS DA has been shown to improve the 6-h forecast background fits to radiosonde observations, satellite radiances, and GFS analyses. Forecast verification against GFS analyses and independent radiance observations also demonstrated statistically significant improvement. It’s encouraging to see the positive impact from all-sky ATMS DA in JEDI-MPAS. Overall, this manuscript is well written and well structured. The manuscript also fits well with the special issue related with the application of JEDI.
Comments:
ATMS has different field of view sizes and separations. Did you process the ATMS data with remapping and spatial averaging to a common FOV (if the original data are not remapped)? As the quality control procedures are based on the assumption that all channels have the same beamwidth.

I understand that the experiments are all-sky ATMS DA runs and most data screening related with cloud/precipitation are not necessary, but have you removed any observations affected by strong convective situations, observations having a very large CLW values, observations in the cold air outbreak area where/if the forecast model tends to produce too much ice cloud? Have you performed any data screening specially for channels 9/10 as the all-sky error model is not applicable to these two channels? If you have QC producers for the above situations, please provide more information. If not, please explain why? If possible, please also provide some information (e.g., how much etc.) about the data removed by your QC procedures.

Could the authors show a few plots about the performance (e.g., BT difference for different channels before and after bias correction) of the bias correction from all-sky ATMS DA?

Have you removed observations when/if some window channels are missing (e.g., channels 1, 2, 5-7, 16? Any inter-channel QC is applied?

Can the authors give more details on how to obtain the values on table 2?

Do you have any numbers for the used channels on how much more observations are being assimilated for the all-sky ATMS DA compared with clear-sky ATMS DA? Have you conducted any comparison between clear-sky and all-sky ATMS DA experiments?

Do the authors have plots similar to Fig. 2 but in the averaged vertical profile (similar to Fig. 3 in Liu et al. 2022)? Also, do you have any forecast verification against radiosonde data?

Are there any reasons that you did not assimilate temperature sounding channels from AMSU-A in your benchmark experiment? Are they related to lines 36-40, where the authors listed some challenges about the all-sky DA for AMSU-A temperate sounding channels? In your ATMS all-sky DA experiments, you included some temperature channels (e.g., channels 6-10) using the all-sky approach. Are there any additional procedures (in addition to humidity-sounding/surface channels) to address the challenges listed in lines 36-40 for all-sky DA for the temperature channels in this study? Can you assimilate temperature channels (e.g., channels 5-9, understanding that the channels may vary for different satellites) from AMSU-A using the same/similar all-sky approach for ATMS temperature channels?

Line 59 “clear-sky ATMS DA” suggested change to “clear-sky ATMS data”

Lines 318-319, “undergo cloud detection at each pixel”, do you remove cloudy pixels and only using clear pixels for verification or what does that mean? Any QC procedure has been conducted for the ABI radiance data used for verification?
Citation: https://doi.org/10.5194/egusphere-2026-1047-RC1
- AC1: 'Reply on RC1', Junmei Ban, 04 Jun 2026
  
  Authors thank two reviewers for taking time in reviewing this paper and providing valuable comments and suggestions. We addressed all comments and provide responses here in blue txt in the attached file.
  
  Citation: https://doi.org/10.5194/egusphere-2026-1047-AC1
RC2:
'Comment on egusphere-2026-1047', Anonymous Referee #2, 07 May 2026

This manuscript demonstrates the extension of the all-sky assimilation to ATMS radiances in the MPAS-JEDI system. A couple of month-long data assimilation experiments were conducted to evaluate the impact of assimilation of additional ATMS radiances starting from temperature sounding channels and then addition of window and humidity sounding channels. The background forecast and extended forecast up to 5 days are verified against other observations and GFS analysis. The results are encouraging. The manuscript is generally well written. However, some clarifications about the all-sky approach, the choices of the data sources and the verifications are needed to help readers to better understand this work and what can be learned from it.
Section 2.1 and 2.2: The assimilated observations are from different sources. Could you provide the rationale behind the choice of data sources. The ATMS data in BUFR format should also be available. Why did you choose to use the ATMS observations from GES DISC? How did you do quality control and bias correction GMI and ABI radiances used for verification?
Section 3.4: In Zhu et al. (2019) and Tong et al. (2020), radiances affected by strong scattering are excluded. In this study, although precipitation hydrometeors are included, the lookup table based on Mie scattering theory would introduce large biases for radiances affected by strong scattering. Did you also do QC based on scattering index as in Zhu et al. (2019)?
Section 4: Please explain the considerations of assimilating some channels over both ocean and land, and some channels over ocean only for ATMS experiments.
Table 3: The description of ATMS_THSch is a bit confusing. It reads like there are two subset of experiments for ATMS_THSch as indicated by (1) and (2).
Figure 2: The statistical significance level or confidence interval is missing in the caption.
Line 270-275: How about the fit to other AMSU-A channels?
Figure 5: Please add ‘with respect to GFS analysis’ after ‘RMSE’
Line 330-335: To help readers to better understand the impact, please add why you choose GMI channel 5 for verification.
Line 354-355: You need to be careful when making this comment. It’s not a fair comparison between operational GFS and this under development MPAS-JEDI system. First, more observations are assimilated in operational GFS than this study. So the impact from adding ATMS radiances could be different. Second, Tong et al. 2020 assimilate precipitation affected radiances and marginal improvement was found in the vector wind forecast over the Southern Hemisphere. The all-sky approach used in this study is similar to early studies. So, here it’s too quick to jump to the conclusion.
Line 378: ‘In Liu et al. (2022), AMSU-A temperature sounding channels were assimilated under clear-sky conditions.’ Since AMSU-A all-sky assimilation has already been implemented in the MPAS-JEDI system, this sentence doesn’t seem to be necessary.

Citation: https://doi.org/10.5194/egusphere-2026-1047-RC2
- AC2: 'Reply on RC2', Junmei Ban, 04 Jun 2026
  
  Authors thank two reviewers for taking time in reviewing this paper and providing valuable comments and suggestions. We addressed all comments and provide responses here in blue txt in the attached file.
  
  Citation: https://doi.org/10.5194/egusphere-2026-1047-AC2

Junmei Ban, Zhiquan Liu, Byoung-Joo Jung, Ivette Hernandez Banos, Benjamin Ruston, and Andrew Collard

Viewed

Total article views: 1,319 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
774	494	51	1,319	73	111

HTML: 774
PDF: 494
XML: 51
Total: 1,319
BibTeX: 73
EndNote: 111

Views and downloads (calculated since 06 Mar 2026)

Month	HTML	PDF	XML	Total
Mar 2026	600	385	44	1,029
Apr 2026	100	64	2	166
May 2026	64	41	3	108
Jun 2026	10	4	2	16

Cumulative views and downloads (calculated since 06 Mar 2026)

Month	HTML	PDF	XML	Total
Mar 2026	600	385	44	1,029
Apr 2026	100	64	2	166
May 2026	64	41	3	108
Jun 2026	10	4	2	16

Viewed (geographical distribution)

Total article views: 1,306 (including HTML, PDF, and XML) Thereof 1,306 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 06 Jun 2026

Short summary

The assimilation of all-sky radiances in MPAS–JEDI, the data assimilation system for the Model for Prediction Across Scales–Atmosphere (MPAS-A) based on the Joint Effort for Data Assimilation Integration (JEDI), has been extended in this study to incorporate ATMS observations. Month-long cycling experiments demonstrate consistent and encouraging improvements in dynamical, thermodynamic, and moisture-related fields resulting from the assimilation of all-sky ATMS radiances.


Total:	0
HTML:	0
PDF:	0
XML:	0