Pith. sign in

REVIEW 1 major objections 49 references

A bidirectional framework jointly retrieves atmospheric profiles and reconstructs hyperspectral radiances from FY-4A GIIRS data by enforcing cycle consistency.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.3

2026-06-26 15:38 UTC pith:5ZL4OFKS

load-bearing objection SIMBA pairs bidirectional Mamba with cycle consistency for joint GIIRS profile retrieval and radiance simulation, showing gains over baselines on ERA5-collocated data. the 1 major comments →

arxiv 2606.19943 v1 pith:5ZL4OFKS submitted 2026-06-18 eess.IV cs.AIphysics.ao-ph

SIMBA: ABidirectional Retrieval Forward Simulation Framework for Modeling FY-4A GIIRS Hyperspectral Infrared Radiances Toward NWP Applications

Jingdong Shen , Fu Wang* , Qifeng Lu , Hao Huang , Chunqiang Wu , Chi Yang , Xiaofang Liu This is my paper
classification eess.IV cs.AIphysics.ao-ph
keywords bidirectional retrievalforward simulationhyperspectral infraredcycle consistencyMambaatmospheric profilesradiance reconstructionnumerical weather prediction
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces SIMBA as a unified bidirectional framework that performs both atmospheric profile retrieval from hyperspectral radiances and the reverse process of radiance simulation from profiles. It adds a cycle-consistency constraint to ensure the two directions reinforce each other and uses a bidirectional Mamba module to model dependencies across atmospheric layers. The goal is to overcome the limitations of one-directional deep learning methods that ignore the consistency between atmospheric states and observations. Results on real FY-4A GIIRS data with ERA5 references show gains over standard baselines in both retrieval accuracy and reconstruction fidelity.

Core claim

SIMBA is a bidirectional retrieval-forward simulation framework that jointly retrieves temperature and humidity profiles and reconstructs long-wave and medium-wave radiances from FY-4A GIIRS observations. It enforces cycle consistency between the retrieval and simulation paths and employs a bidirectional Mamba state-space module to capture long-range dependencies along pressure levels. When tested against collocated ERA5 data, the method exceeds representative deep learning baselines on retrieval and reconstruction metrics, and ablations verify the value of the bidirectional and cycle-consistency components.

What carries the argument

The cycle-consistency constraint that couples the retrieval network and the forward simulation network, implemented with a bidirectional Mamba state-space module.

Load-bearing premise

Collocated FY-4A GIIRS observations paired with ERA5 reanalysis profiles serve as accurate ground truth for measuring both retrieval errors and radiance reconstruction errors.

What would settle it

An independent validation set of in-situ or aircraft measurements shows that SIMBA retrieval root-mean-square errors for temperature and humidity are not smaller than those from the compared deep learning baselines.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Higher accuracy in temperature retrieval from GIIRS radiances compared to one-way baselines.
  • Improved specific humidity retrieval performance across pressure levels.
  • Higher fidelity reconstruction of both long-wave and medium-wave radiances.
  • The bidirectional design and cycle-consistency mechanism each contribute measurable gains in the reported experiments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same cycle-consistency structure could be tested on other hyperspectral sensors to check whether the coupling benefit generalizes beyond FY-4A GIIRS.
  • If the retrieved profiles and reconstructed radiances remain consistent, the outputs could be used directly in observation operators for data assimilation without additional bias correction steps.
  • Training the framework on a different reanalysis product would reveal how sensitive the reported improvements are to the choice of reference atmospheric states.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

1 major / 0 minor

Summary. The paper proposes SIMBA, a bidirectional retrieval-forward simulation framework for FY-4A GIIRS hyperspectral IR radiances. It jointly performs atmospheric profile retrieval (temperature and specific humidity) and radiance reconstruction (LW and MW) using a bidirectional Mamba state-space module to capture vertical dependencies and a cycle-consistency constraint to couple the retrieval and forward branches. Evaluated on collocated GIIRS observations and ERA5 reanalysis, it claims outperformance over representative deep learning baselines on both tasks, with ablations confirming the value of the bidirectional design and cycle-consistency mechanism.

Significance. If the empirical claims hold under independent validation, the framework offers a principled way to enforce consistency between state and observation spaces, which is directly relevant to NWP data assimilation. The bidirectional Mamba architecture and cycle-consistency loss constitute a clear methodological contribution over one-way retrieval networks; the ablation results provide concrete evidence for these design elements.

major comments (1)
  1. [Abstract / Evaluation] Abstract and evaluation description: the central performance claims rest on retrieval and reconstruction metrics scored against collocated ERA5 profiles and GIIRS observations. ERA5 specific-humidity fields carry substantially larger analysis errors than temperature (particularly in the lower troposphere and tropics); because the cycle-consistency loss couples the two branches, the model can learn to reproduce ERA5-specific biases rather than the true radiance-to-state mapping. This directly affects the interpretation of both the outperformance results and the ablation experiments.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the thoughtful and constructive review. The concern regarding potential ERA5 biases is well-taken and we address it directly below.

read point-by-point responses

  1. Referee: [Abstract / Evaluation] Abstract and evaluation description: the central performance claims rest on retrieval and reconstruction metrics scored against collocated ERA5 profiles and GIIRS observations. ERA5 specific-humidity fields carry substantially larger analysis errors than temperature (particularly in the lower troposphere and tropics); because the cycle-consistency loss couples the two branches, the model can learn to reproduce ERA5-specific biases rather than the true radiance-to-state mapping. This directly affects the interpretation of both the outperformance results and the ablation experiments.

    Authors: We agree that ERA5 reanalysis carries larger uncertainties for specific humidity than for temperature, particularly in the lower troposphere and tropics, and that this constitutes a known limitation when reanalysis serves as the reference for both training and evaluation. Because the cycle-consistency loss operates on the same ERA5-GIIRS pairs, it is possible that some component of the learned mapping reflects ERA5-specific biases rather than purely observational truth. All baselines are trained and scored on identical collocated data, so the relative gains reported for SIMBA remain valid within this experimental setting; the bidirectional Mamba and cycle-consistency still demonstrably improve consistency between the two branches on the given dataset. Nevertheless, the referee’s point correctly identifies a caveat for absolute interpretation. In the revised manuscript we will add an explicit limitations paragraph in the discussion section acknowledging the analysis-error characteristics of ERA5 humidity fields and their potential influence on the cycle-consistency results. We will also note that future work could incorporate independent validation (e.g., radiosonde or other satellite products) to further isolate true mapping performance. This constitutes a partial revision. revision: partial

Circularity Check

0 steps flagged

No significant circularity in empirical ML framework

full rationale

The paper proposes SIMBA, a bidirectional deep learning framework for atmospheric profile retrieval and radiance reconstruction, and supports its claims solely through empirical comparisons to baselines plus ablation studies on collocated FY-4A GIIRS and ERA5 data. No derivation chain, equations, or self-citations appear that reduce any prediction or result to its own inputs by construction; the evaluation remains self-contained against external benchmarks without fitted-input-as-prediction or self-definitional steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no explicit free parameters, axioms, or invented entities beyond standard neural network training assumptions; ERA5 reanalysis is treated as external reference data.

pith-pipeline@v0.9.1-grok · 5798 in / 1073 out tokens · 23843 ms · 2026-06-26T15:38:24.457577+00:00 · methodology

0 comments
read the original abstract

Hyperspectral infrared observations are an important data source for numerical weather prediction (NWP) because they provide rich information on the vertical structure of atmospheric temperature and humidity. However, most existing deep learning methods mainly focus on one-way retrieval from radiances to atmospheric profiles, while the reverse radiance simulation process and the consistency between atmospheric state space and radiance observation space are insufficiently considered. In this study, we propose SIMBA, a unified bidirectional retrieval-forward simulation framework for FY-4A GIIRS hyperspectral infrared radiance modeling toward NWP applications. The framework jointly performs atmospheric profile retrieval and radiance reconstruction, introduces a cycle-consistency constraint to strengthen the coupling between the two processes, and employs a bidirectional Mamba state-space module to capture long-range dependencies along pressure levels. Using collocated FY-4A GIIRS observations and ERA5 reanalysis data, the proposed method is evaluated for temperature retrieval, specific humidity retrieval, long-wave radiance reconstruction, and medium-wave radiance reconstruction. Experimental results show that SIMBA outperforms several representative deep learning baselines across both retrieval and reconstruction tasks, while ablation experiments confirm the contribution of the bidirectional design and cycle-consistency mechanism. These results demonstrate that the proposed framework is effective for joint atmospheric profile retrieval and hyperspectral infrared radiance modeling, and suggest potential for future Jacobian-related analysis and NWP-oriented extensions.

Figures

Figures reproduced from arXiv: 2606.19943 by Chi Yang, Chunqiang Wu, Fu Wang*, Hao Huang, Jingdong Shen, Qifeng Lu, Xiaofang Liu.

Figure 1
Figure 1. Figure 1: Overall architecture of SIMBA. The framework consists of a retrieval branch and a forward simulation branch for bidirectional radiance–profile modeling. A bidirectional Mamba module is used for vertical sequence modeling, while FiLM conditioning injects radiative information into the profile generation process. The two branches are coupled through a cycle-consistency constraint. SIMBA uses GIIRS LW and MW … view at source ↗
Figure 2
Figure 2. Figure 2: presents the overall density scatter distributions of temperature retrievals on the test set for different models. In general, the predictions of all models follow the 1:1 reference line, indicating that they can reproduce the overall temperature distribution reasonably well. To quantitatively characterize the scatter dispersion, an expected error (EE) envelope of ±1 K was introduced around the 1:1 line, a… view at source ↗
Figure 3
Figure 3. Figure 3: Scatter plot comparison of temperature retrieval at six representative pressure levels: (a) 100 hPa; (b) 300 hPa; (c) 500 hPa; (d) 700 hPa; (e) 850 hPa; (f) 1000 hPa. Columns correspond to BiCNN, BiMLP, BiTransformer, BiLSTM, and SIMBA, while the rightmost column shows the kernel density estimate (True KDE) of ERA5 temperature at each pressure level. Figures 4 and 5 show the vertical distributions of RMSE … view at source ↗
Figure 4
Figure 4. Figure 4: Vertical distribution of RMSE for temperature retrieval across different models: (a) full pressure range; (b) enlarged view of 50–200 hPa; (c) enlarged view of 700–1100 hPa [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Vertical distribution of relative bias for temperature retrieval across different models: (a) full pressure range; (b) enlarged view of 30–200 hPa; (c) enlarged view of 700–1100 hPa. 3.2.2. Humidity Profiles To further evaluate specific humidity retrieval performance, the overall scatter distri￾butions, layered scatter characteristics, and vertical error profiles of different models are analyzed in this se… view at source ↗
Figure 6
Figure 6. Figure 6: Overall scatter plot comparison of specific humidity retrieval on the test set for different models: (a) BiCNN; (b) BiMLP; (c) BiTransformer; (d) BiLSTM; (e) SIMBA. The layered scatter plots in [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Scatter plot comparison of specific humidity retrieval at six representative pressure levels: (a) 200 hPa; (b) 500 hPa; (c) 700 hPa; (d) 850 hPa; (e) 925 hPa; (f) 1000 hPa. Columns correspond to BiCNN, BiMLP, BiTransformer, BiLSTM, and SIMBA, while the rightmost column shows the kernel density estimate (True KDE) of ERA5 specific humidity at each pressure level [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Vertical distribution of RMSE for specific humidity retrieval across different models: (a) full pressure range; (b) enlarged view of 500–700 hPa; (c) enlarged view of 700–1100 hPa. https://doi.org/10.3390/rs1010000 [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Vertical distribution of relative bias for specific humidity retrieval across different models: (a) full pressure range; (b) enlarged view of 200–500 hPa; (c) enlarged view of 700–1100 hPa. 3.3. Evaluation for Forward Models 3.3.1. Radiation of Long-Wave Bands From the perspective of NWP data assimilation, forward consistency in observation space is important for assessing whether the retrieved atmospheric… view at source ↗
Figure 10
Figure 10. Figure 10: Overall scatter plot comparison of long-wave radiance reconstruction on the test set for different models: (a) BiCNN; (b) BiMLP; (c) BiTransformer; (d) BiLSTM; (e) SIMBA [PITH_FULL_IMAGE:figures/full_fig_p015_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Scatter plot comparison of long-wave radiance reconstruction for six representative channels: (a) CH1; (b) CH20; (c) CH35; (d) CH60; (e) CH80; (f) CH100. Columns correspond to BiCNN, BiMLP, BiTransformer, BiLSTM, and SIMBA, while the rightmost column shows the kernel density estimate (True KDE) of the observed radiance at each channel. https://doi.org/10.3390/rs1010000 [PITH_FULL_IMAGE:figures/full_fig_p… view at source ↗
Figure 12
Figure 12. Figure 12: Distribution of RMSE across 106 long-wave channels for different models: (a) full channel range; (b) enlarged view of CH41–CH61; (c) enlarged view of CH71–CH91 [PITH_FULL_IMAGE:figures/full_fig_p016_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Distribution of relative bias across 106 long-wave channels for different models: (a) full channel range; (b) enlarged view of CH1–CH53; (c) enlarged view of CH54–CH106. Overall, the LW radiance reconstruction results demonstrate that SIMBA provides better observation-space consistency and more stable channel-wise performance than the comparison models, indicating a more reliable radiance–profile mapping … view at source ↗
Figure 14
Figure 14. Figure 14: Overall scatter plot comparison of medium-wave radiance reconstruction on the test set for different models: (a) BiCNN; (b) BiMLP; (c) BiTransformer; (d) BiLSTM; (e) SIMBA. To further analyze reconstruction differences across spectral positions, six representa￾tive channels (CH2, CH20, CH37, CH42, CH70, and CH88) are selected from the 99 MW channels for detailed analysis in [PITH_FULL_IMAGE:figures/full_… view at source ↗
Figure 15
Figure 15. Figure 15: Scatter plot comparison of medium-wave radiance reconstruction for six representative channels: (a) CH2; (b) CH20; (c) CH37; (d) CH42; (e) CH70; (f) CH88. Columns correspond to BiCNN, BiMLP, BiTransformer, BiLSTM, and SIMBA, while the rightmost column shows the kernel density estimate (True KDE) of the observed radiance at each channel [PITH_FULL_IMAGE:figures/full_fig_p018_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Distribution of RMSE across 99 medium-wave channels for different models: (a) full channel range; (b) enlarged view of CH30–CH44; (c) enlarged view of CH70–CH85. https://doi.org/10.3390/rs1010000 [PITH_FULL_IMAGE:figures/full_fig_p018_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Distribution of relative bias across 99 medium-wave channels for different models: (a) full channel range; (b) enlarged view of CH1–CH50; (c) enlarged view of CH51–CH99. 3.4. Ablation Study To isolate the contribution of the bidirectional closed-loop design, ablation experiments compare three variants under the same Mamba backbone, data split, preprocessing, training protocol, and evaluation metrics: Retr… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

49 extracted references · 3 canonical work pages · 1 internal anchor

  1. [1]

    Challenges and opportunities in numerical weather prediction.Bull

    Brotzge, J.A.; Berchoff, D.; Carlis, D.L.; Carr, F.H.; Carr, R.H.; Gerth, J.J.; Gross, B.D.; Hamill, T.M.; Haupt, S.E.; Jacobs, N.; et al. Challenges and opportunities in numerical weather prediction.Bull. Am. Meteorol. Soc.2023,104, E698–E705

    2023

  2. [2]

    Overview and prospect of data assimilation in numerical weather prediction.J

    Lei, L.; Weng, F.; Duan, W.; Chen, Y.; Zhang, L.; Wang, R.; Yang, J.; Qin, X.; Han, W.; Li, J.; et al. Overview and prospect of data assimilation in numerical weather prediction.J. Meteorol. Res.2025,39, 559–592

    2025

  3. [3]

    Assessment of the hyperspectral infrared atmospheric sounder (HIRAS)

    Carminati, F.; Xiao, X.; Lu, Q.; Atkinson, N.; Hocking, J. Assessment of the hyperspectral infrared atmospheric sounder (HIRAS). Remote Sens.2019,11, 2950

    2019

  4. [4]

    An overview of the applications of earth observation satellite data: impacts and future trends.Remote Sens.2022,14, 1863

    Zhao, Q.; Yu, L.; Du, Z.; Peng, D.; Hao, P .; Zhang, Y.; Gong, P . An overview of the applications of earth observation satellite data: impacts and future trends.Remote Sens.2022,14, 1863

    2022

  5. [5]

    Assessment of the potential impact of a hyperspectral infrared sounder on the Himawari follow-on geostationary satellite.SOLA 2020,16, 162–168

    Okamoto, K.; Owada, H.; Fujita, T.; Kazumori, M.; Otsuka, M.; Seko, H.; Ota, Y.; Uekiyo, N.; Ishimoto, H.; Hayashi, M.; et al. Assessment of the potential impact of a hyperspectral infrared sounder on the Himawari follow-on geostationary satellite.SOLA 2020,16, 162–168

    2020

  6. [6]

    Yin, R.; Han, W.; Gao, Z.; Li, J. Impact of high temporal resolution FY-4A Geostationary Interferometric Infrared Sounder (GIIRS) radiance measurements on Typhoon forecasts: Maria (2018) case with GRAPES global 4D-Var assimilation system.Geophys. Res. Lett.2021,48, e2021GL093672

    2018

  7. [8]

    AIRS: Improving weather forecasting and providing new data on greenhouse gases.Bull

    Chahine, M.T.; Pagano, T.S.; Aumann, H.H.; Atlas, R.; Barnet, C.; Blaisdell, J.; Chen, L.; Divakarla, M.; Fetzer, E.J.; Goldberg, M.; et al. AIRS: Improving weather forecasting and providing new data on greenhouse gases.Bull. Am. Meteorol. Soc.2006, 87, 911–926

    2006

  8. [9]

    A closer look at the ABI on the GOES-R series

    Schmit, T.J.; Griffith, P .; Gunshor, M.M.; Daniels, J.M.; Goodman, S.J.; Lebair, W.J. A closer look at the ABI on the GOES-R series. Bull. Am. Meteorol. Soc.2017,98, 681–698

    2017

  9. [10]

    Rodgers, C.D.Inverse Methods for Atmospheric Sounding: Theory and Practice; World Scientific: Singapore, 2000; Volume 2

    2000

  10. [11]

    Advanced radiative transfer modeling system developed for satellite data assimilation and remote sensing applications.J

    Yang, J.; Ding, S.; Dong, P .; Bi, L.; Yi, B. Advanced radiative transfer modeling system developed for satellite data assimilation and remote sensing applications.J. Quant. Spectrosc. Radiat. Transf.2020,251, 107043

    2020

  11. [12]

    Monitoring the performance of the Fengyun satellite instruments using radiative transfer models and NWP fields.J

    Lu, Q.; Hu, J.; Wu, C.; Qi, C.; Wu, S.; Xu, N.; Sun, L.; Li, X.; Liu, H.; Guo, Y.; et al. Monitoring the performance of the Fengyun satellite instruments using radiative transfer models and NWP fields.J. Quant. Spectrosc. Radiat. Transf.2020,255, 107239

    2020

  12. [13]

    Assimilation of satellite data in numerical weather prediction

    Eyre, J.; Bell, W.; Cotton, J.; English, S.; Forsythe, M.; Healy, S.; Pavelin, E. Assimilation of satellite data in numerical weather prediction. Part II: Recent years.Q. J. R. Meteorol. Soc.2022,148, 521–556

    2022

  13. [14]

    The ECMWF operational implementation of four-dimensional variational assimilation

    Rabier, F.; Järvinen, H.; Klinker, E.; Mahfouf, J.F.; Simmons, A. The ECMWF operational implementation of four-dimensional variational assimilation. I: Experimental results with simplified physics.Q. J. R. Meteorol. Soc.2000,126, 1143–1170

    2000

  14. [15]

    Coupled data assimilation at ECMWF: current status, challenges and future developments.Q

    de Rosnay, P .; Browne, P .; de Boisséson, E.; Fairbairn, D.; Hirahara, Y.; Ochi, K.; Schepers, D.; Weston, P .; Zuo, H.; Alonso- Balmaseda, M.; et al. Coupled data assimilation at ECMWF: current status, challenges and future developments.Q. J. R. Meteorol. Soc.2022,148, 2672–2702

    2022

  15. [16]

    The application of neural networks in the earth system sciences.Neural Netw

    Krasnopolsky, V .M. The application of neural networks in the earth system sciences.Neural Netw. Emul. Complex Multidimens. Mapp.2013,46

    2013

  16. [17]

    An artificial neural network based fast radiative transfer model for simulating infrared sounder radiances.J

    Krishnan, P .; Srinivasa Ramanujam, K.; Balaji, C. An artificial neural network based fast radiative transfer model for simulating infrared sounder radiances.J. Earth Syst. Sci.2012,121, 891–901

    2012

  17. [18]

    Deep learning and process understanding for data-driven Earth system science.Nature2019,566, 195–204

    Reichstein, M.; Camps-Valls, G.; Stevens, B.; Jung, M.; Denzler, J.; Carvalhais, N.; Prabhat, F. Deep learning and process understanding for data-driven Earth system science.Nature2019,566, 195–204

  18. [19]

    Machine learning methods in weather and climate applications: A survey

    Chen, L.; Han, B.; Wang, X.; Zhao, J.; Yang, W.; Yang, Z. Machine learning methods in weather and climate applications: A survey. Appl. Sci.2023,13, 12019

    2023

  19. [20]

    Physics-informed machine learning: Case studies for weather and climate modelling.Philos

    Kashinath, K.; Mustafa, M.; Albert, A.; Wu, J.L.; Jiang, C.; Esmaeilzadeh, S.; Azizzadenesheli, K.; Wang, R.; Chattopadhyay, A.; Singh, A.; et al. Physics-informed machine learning: Case studies for weather and climate modelling.Philos. Trans. R. Soc. A Math. Phys. Eng. Sci.2021,379

    2021

  20. [21]

    Retrieving atmospheric gas profiles using FY-3E/HIRAS-II infrared hyperspectral data by neural network approach.Remote Sens.2023,15, 2931

    Li, H.; Gu, M.; Zhang, C.; Xie, M.; Yang, T.; Hu, Y. Retrieving atmospheric gas profiles using FY-3E/HIRAS-II infrared hyperspectral data by neural network approach.Remote Sens.2023,15, 2931

    2023

  21. [22]

    Simultaneous estimation of land surface and atmospheric parameters from thermal hyperspectral data using a LSTM–CNN combined deep neural network.IEEE Geosci

    Ye, X.; Ren, H.; Nie, J.; Hui, J.; Jiang, C.; Zhu, J.; Fan, W.; Qian, Y.; Liang, Y. Simultaneous estimation of land surface and atmospheric parameters from thermal hyperspectral data using a LSTM–CNN combined deep neural network.IEEE Geosci. Remote Sens. Lett.2021,19, 1–5

    2021

  22. [23]

    Hybrid Bayesian-machine learning framework for multi-profile atmospheric retrieval from hyperspectral infrared observations.Adv

    Kong, S.; Bi, L.; Han, W.; Yin, R.; Zhang, H. Hybrid Bayesian-machine learning framework for multi-profile atmospheric retrieval from hyperspectral infrared observations.Adv. Atmos. Sci.2026,43, 373–389

    2026

  23. [24]

    An improved retrieval method of atmospheric parameter profiles based on the BP neural network

    Zhao, Y.; Zhou, D.; Yan, H. An improved retrieval method of atmospheric parameter profiles based on the BP neural network. Atmos. Res.2018,213, 389–397

    2018

  24. [25]

    A Dive into Generative Adversarial Networks in the World of Hyperspectral Imaging: A Survey of the State of the Art.Remote Sens.2026,18, 196

    Ranjan, P .; Nandal, A.; Agarwal, S.; Kumar, R. A Dive into Generative Adversarial Networks in the World of Hyperspectral Imaging: A Survey of the State of the Art.Remote Sens.2026,18, 196

    2026

  25. [26]

    Statistical retrieval of atmospheric profiles with deep convolutional neural networks.ISPRS J

    Malmgren-Hansen, D.; Laparra, V .; Nielsen, A.A.; Camps-Valls, G. Statistical retrieval of atmospheric profiles with deep convolutional neural networks.ISPRS J. Photogramm. Remote Sens.2019,158, 231–240

    2019

  26. [27]

    Data assimilation: making sense of Earth Observation.Front

    Lahoz, W.A.; Schneider, P . Data assimilation: making sense of Earth Observation.Front. Environ. Sci.2014,2, 16

    2014

  27. [28]

    The future of Earth system prediction: Advances in model-data fusion.Sci

    Gettelman, A.; Geer, A.J.; Forbes, R.M.; Carmichael, G.R.; Feingold, G.; Posselt, D.J.; Stephens, G.L.; Van den Heever, S.C.; Varble, A.C.; Zuidema, P . The future of Earth system prediction: Advances in model-data fusion.Sci. Adv.2022,8, eabn3488

    2022

  28. [29]

    Physics-guided deep learning for skillful wind-wave modeling.Sci

    Wang, X.; Jiang, H. Physics-guided deep learning for skillful wind-wave modeling.Sci. Adv.2024,10, eadr3559

    2024

  29. [30]

    Surrogate modeling for the climate sciences dynamics with machine learning and data assimilation.Front

    Bocquet, M. Surrogate modeling for the climate sciences dynamics with machine learning and data assimilation.Front. Appl. Math. Stat.2023,9, 1133226

    2023

  30. [31]

    Atmospheric boundary layer height from ground-based remote sensing: A review of capabilities and limitations.Atmos

    Kotthaus, S.; Bravo-Aranda, J.A.; Collaud Coen, M.; Guerrero-Rascado, J.L.; Costa, M.J.; Cimini, D.; O’Connor, E.J.; Hervo, M.; Alados-Arboledas, L.; Jiménez-Portaz, M.; et al. Atmospheric boundary layer height from ground-based remote sensing: A review of capabilities and limitations.Atmos. Meas. Tech. 2023, 16, 433–479

    2023

  31. [32]

    One-dimensional variational retrieval of temperature and humidity profiles from the FY4A GIIRS.Adv

    Xue, Q.; Guan, L.; Shi, X. One-dimensional variational retrieval of temperature and humidity profiles from the FY4A GIIRS.Adv. Atmos. Sci.2022,39, 471–486

    2022

  32. [34]

    Region-aware sequence-to-sequence learning for hyperspectral denoising

    Xiao, J.; Liu, Y.; Wei, X. Region-aware sequence-to-sequence learning for hyperspectral denoising. InProceedings of the European Conference on Computer Vision; Springer: Berlin/Heidelberg, Germany, 2024; pp. 218–235

    2024

  33. [35]

    Efficiently Modeling Long Sequences with Structured State Spaces

    Gu, A.; Goel, K.; Ré, C. Efficiently modeling long sequences with structured state spaces.arXiv2021, arXiv:2111.00396

    work page internal anchor Pith review Pith/arXiv arXiv

  34. [36]

    A Mamba-Based Hierarchical Partitioning Framework for Upper-Level Wind Field Reconstruction.Aerospace2025,12, 842

    Chen, W.; Zhang, Y.; Liu, R.; Sun, S.; Feng, Q. A Mamba-Based Hierarchical Partitioning Framework for Upper-Level Wind Field Reconstruction.Aerospace2025,12, 842

  35. [37]

    Improving typhoon predictions by assimilating the retrieval of atmospheric temperature profiles from the FengYun-4A’s Geostationary Interferometric Infrared Sounder (GIIRS).Atmos

    Feng, J.; Qin, X.; Wu, C.; Zhang, P .; Yang, L.; Shen, X.; Han, W.; Liu, Y. Improving typhoon predictions by assimilating the retrieval of atmospheric temperature profiles from the FengYun-4A’s Geostationary Interferometric Infrared Sounder (GIIRS).Atmos. Res. 2022,280, 106391

    2022

  36. [38]

    Land Surface Temperature end-to-end retrieval considering the topographic effect using radiative transfer model-driven convolutional neural network.IEEE Trans

    Ye, X.; Wang, P .; Zhu, J.; Duan, Y.; Yang, B. Land Surface Temperature end-to-end retrieval considering the topographic effect using radiative transfer model-driven convolutional neural network.IEEE Trans. Geosci. Remote Sens.2025,63, 5001010

    2025

  37. [39]

    Emulation of Forward Modeled Top-of-Atmosphere MODIS-Based Spectral Channels Using Machine Learning.IEEE J

    Gonzalez, J.; Dipu, S.; Sourdeval, O.; Siméon, A.; Camps-Valls, G.; Quaas, J. Emulation of Forward Modeled Top-of-Atmosphere MODIS-Based Spectral Channels Using Machine Learning.IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens.2024,18, 1896–1911

    2024

  38. [40]

    Unpaired image-to-image translation using cycle-consistent adversarial networks

    Zhu, J.Y.; Park, T.; Isola, P .; Efros, A.A. Unpaired image-to-image translation using cycle-consistent adversarial networks. In Proceedings of the IEEE International Conference on Computer Vision, Venice, Italy, 22–29 October 2017; pp. 2223–2232

    2017

  39. [41]

    A graph clustering approach to localization for adaptive covariance tuning in data assimilation based on state-observation mapping.Math

    Cheng, S.; Argaud, J.P .; Iooss, B.; Ponçot, A.; Lucor, D. A graph clustering approach to localization for adaptive covariance tuning in data assimilation based on state-observation mapping.Math. Geosci.2021,53, 1751–1780

    2021

  40. [42]

    Surrogate models of radiative transfer codes for atmospheric trace gas retrievals from satellite observations.Mach

    Brence, J.; Tanevski, J.; Adams, J.; Malina, E.; Džeroski, S. Surrogate models of radiative transfer codes for atmospheric trace gas retrievals from satellite observations.Mach. Learn.2023,112, 1337–1363

    2023

  41. [43]

    Enhancing 3D planetary atmosphere simulations with a surrogate radiative transfer model.Mon

    Tahseen, T.P .; Mendonça, J.M.; Yip, K.H.; Waldmann, I.P . Enhancing 3D planetary atmosphere simulations with a surrogate radiative transfer model.Mon. Not. R. Astron. Soc.2024,535, 2210–2227

    2024

  42. [44]

    The ERA5 global reanalysis.Q

    Hersbach, H.; Bell, B.; Berrisford, P .; Hirahara, S.; Horányi, A.; Muñoz-Sabater, J.; Nicolas, J.; Peubey, C.; Radu, R.; Schepers, D.; et al. The ERA5 global reanalysis.Q. J. R. Meteorol. Soc.2020,146, 1999–2049

    2020

  43. [45]

    A channel selection method for hyperspectral atmospheric infrared sounders based on layering.Atmos

    Chang, S.; Sheng, Z.; Du, H.; Ge, W.; Zhang, W. A channel selection method for hyperspectral atmospheric infrared sounders based on layering.Atmos. Meas. Tech.2020,13, 629–644

    2020

  44. [46]

    Retrieval and fusion of atmospheric temperature and humidity profiles in the East China based on FY-4B/GIIRS.J

    Zhang, L.; Bao, Y.; Liu, H.; Lu, Q.; Wang, Y.; Huang, Y.; Wu, Y. Retrieval and fusion of atmospheric temperature and humidity profiles in the East China based on FY-4B/GIIRS.J. Meteorol. Sci.2026,46, 80–91. https://doi.org/10.12306/2025jms.0003

    work page doi:10.12306/2025jms.0003 2026

  45. [47]

    Film: Visual reasoning with a general conditioning layer

    Perez, E.; Strub, F.; De Vries, H.; Dumoulin, V .; Courville, A. Film: Visual reasoning with a general conditioning layer. In Proceedings of the AAAI Conference on Artificial Intelligence, New Orleans, LA, USA, 2–7 February 2018; Volume 32

    2018

  46. [48]

    A convolutional neural network and attention-based retrieval of temperature profile for a satellite hyperspectral microwave sensor.Atmosphere2024,15, 235

    Tan, X.; Ma, K.; Dou, F. A convolutional neural network and attention-based retrieval of temperature profile for a satellite hyperspectral microwave sensor.Atmosphere2024,15, 235

  47. [49]

    Retrieval of atmospheric temperature profiles from FY-4A/GIIRS hyperspectral data based on TPE-MLP: Analysis of retrieval accuracy and influencing factors.Remote Sens.2024,16, 1976

    Xu, X.; Han, W.; Gao, Z.; Li, J.; Yin, R. Retrieval of atmospheric temperature profiles from FY-4A/GIIRS hyperspectral data based on TPE-MLP: Analysis of retrieval accuracy and influencing factors.Remote Sens.2024,16, 1976

    2024

  48. [50]

    A transformer network air temperature and humidity inversion method based on ATMS brightness temperature data.IEEE Geosci

    Xiao, C.; Dong, J.; Dou, H.; Li, Y.; Wang, W.; Ren, F. A transformer network air temperature and humidity inversion method based on ATMS brightness temperature data.IEEE Geosci. Remote Sens. Lett.2024,22, 4500205

    2024

  49. [51]

    A deep learning framework for enhanced retrieval of atmospheric temperature and humidity profiles across China: Unifying inversion algorithms across multiple stations.Atmos

    Jiang, S.; Ma, Y.; Deng, F.; Lei, L. A deep learning framework for enhanced retrieval of atmospheric temperature and humidity profiles across China: Unifying inversion algorithms across multiple stations.Atmos. Res.2025,315, 107793. Disclaimer/Publisher’s Note:The statements, opinions and data contained in all publications are solely those of the individu...

    work page doi:10.3390/rs1010000 2025