the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Feb 2025
Optimizing cloud optical parameterizations in RTTOV for data assimilation of satellite visible reflectance data: an assessment using observed and synthetic images
Abstract. The Radiative Transfer for TOVS (RTTOV) is a commonly used forward operator software package for the Data Assimilation (DA) of satellite visible reflectance data. However, a wide choice of Cloud Optical Parameterizations (COPs) in RTTOV poses challenges in discerning the optimal configuration. In this study, the performance of different COPs was evaluated by comparing the observed and synthetic visible satellite images. Observed images (O) were provided by Fengyun (FY)-4B and Himawari-9, two operational geostationary meteorological satellites covering East Asia. Synthetic images (B) were generated by RTTOV with the Discrete Ordinate Method (DOM) and the Method for FAst Satellite Image Simulation (MFASIS). The inputs to RTTOV were provided by the 3-h forecasts of the China Meteorological Administration Mesoscale (CMA-MESO) model and the fifth generation European Centre for Medium-Range Weather Forecasts reanalysis (ERA5) data. On the domain average, B was smaller than O, especially in cloudy situations. The minimum O-B bias was revealed for the COP of liquid water clouds in terms of effective diameter (Deff) in combination with the COP of ice clouds developed by the Space Science and Engineering Center (SSEC), with the Deff for ice clouds parameterized in terms of ice water content and temperature. Compared with the O-B biases, the standard deviations of the O-B departure were less sensitive to COPs. In addition, histogram analysis of reflectance indicated that the synthetic images with the minimum O-B bias resembled best with the observed images. Therefore, the optimal cloud optical parameterization was proposed to be the “Deff” + “SSEC” suite.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2285 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2285 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-242', Anonymous Referee #1, 12 Mar 2025
Review of the manuscript titled "Optimizing cloud optical parameterizations in RTTOV for data assimilation of satellite visible reflectance data: an assessment using observed and synthetic images" by Yongbo Zhou, Tianrui Cao, and Lijian Zhu.
This paper evaluates parametrizations of visible reflectance simulated from CMA-MESO and ERA5 using RTTOV with observations from instruments onboard the satellites FY-4B and Himawari-9.
Results show that the observed reflectance is higher than modelled reflectance, on average, and that the choice of an optimal configuration yields the lowest bias while only slightly increasing unbiased error (standard deviation of O-B). In clear-sky pixels, the atmospheric contribution to visible reflectance is near zero. Thus, the choice of parametrization had no effect. The bias is induced by the land surface representation in the NWP model. In cloudy sky, bias is more relevant than in clear sky cases, because the bias is responsible for a larger fraction of error, as shown in Figures 10, 11, and 12. In clear sky, bias was mostly lower than unbiased error, except of experiment CM-FY-DM in April (Figure 3).
The authors already nicely addressed my comments to an earlier version of the manuscript in review for another journal. For example, the authors extended the analysis by the standard deviation of O-B. The manuscript is now an extensive evaluation and in a very good condition. I recommend publication after minor comments are addressed.
Minor comments
L243-244: "the increased B for C213 would amplify the O-B differences, leading to the increased standard deviations". The standard deviation (SD) is not influenced by the bias (average O-B). Thus, an change in average B does not change the SD. The explanation for an increased SD must be larger O-B differences, due to B being too high and too low (increase in |O-B|).
Figure 4: The caption seems to have an error, as the x-axis label is "reflectance" not "O-B departure".
Equation 6 seems to be incorrect, not all terms are in the exponent, compared to Equation 2 of McFarquar et al. (2003).
L237, L275: Language: Instead of "was expected", you probably mean "was found"? Or was there an expectation before seeing the results?
L243: You state "opposite circumstance" but the same sign (O-B>0) as in the line above. Probably you mean O-B < 0?
L248: "at the low- or high-reflectance ends": Maybe you can quantify this, e.g. "at low reflectance (<0.1)" or similar.
L271: "The results in Figure 4 suggested that the Baran 2018 ice scheme should be used with caution ...".
I am confused, because Figure 4 shows results for ice_scheme=1 (SSEC) and not for "Baran" parametrization. Maybe you meant Figure 5?Figure 7 caption (c) missing.
L367: Language, "B was consistently underestimated compared with O", I think this should be "B consistently underestimated O".
L393: "The smallest O-B bias". Good, but can you quantify it? For example "the optimal configuration reduced the bias by 0.04 on average, while standard deviation increased by less than 0.005.
Citation: https://doi.org/10.5194/egusphere-2025-242-RC1 -
AC2: 'Reply on RC1', Yongbo Zhou, 02 May 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-242/egusphere-2025-242-AC2-supplement.pdf
-
AC2: 'Reply on RC1', Yongbo Zhou, 02 May 2025
-
RC2: 'Comment on egusphere-2025-242', Anonymous Referee #3, 06 Apr 2025
Overall evaluations
The research is highly relevant as it focuses on optimizing cloud optical parameterizations in RTTOV for satellite visible reflectance data assimilation. With the increasing importance of satellite - based data in weather forecasting and climate studies, improving the accuracy of radiative transfer models like RTTOV is crucial. The use of data from recent geostationary meteorological satellites (FY - 4B and Himawari - 9) makes the study timely. The authors conducted four different experiments, covering two observing systems, two radiative solvers, and two atmospheric fields. A wide range of data was used, including forecasts from the CMA - MESO model, ERA5 reanalysis data, and observations from FY - 4B and Himawari - 9 satellites. The data was carefully processed, and multiple statistical methods were employed, such as analyzing O - B biases, standard deviations, and probability density functions. These analyses provided in - depth insights into the performance of different cloud optical parameterizations. The study identified the optimal cloud optical parameterization suite (“Deff” + “SSEC” + B02), which can significantly reduce the O - B biases. This finding has practical implications for improving the accuracy of radiative transfer simulations and data assimilation in weather prediction models.
Major issues:
- The number concentration in Eq. (1) was set to 300 cm-3 Is this generally applicable for all types of clouds? If not, what is the sensitivity of the simulated reflectivity to the other number concentration?
- For cloud ice, is the effective diameter in Eq. (2) independent of cloud ice content and number concentration? What is the number concentration assumed for cloud ice?
- In CMA-MESO, there are six types of hydrometeors. How is the scattering from rain, snow and graupel is treated in this study?
- Since FY-4B AGRI has an aerosol AOD product, it is better to use AOD to quantify the contribution from aerosol? Eq. (8) and (9) seems to be too objective.
- Captions for Figure 7b do not make sense. What is Fig 7c? Also, it is strange to show O-B bias from one month data.
- The DOM solver simplified the scattering interactions between clouds and gaseous molecules into single - scattering processes in non - cloudy layers. This simplification may lead to inaccuracies in the simulated reflectance, especially in areas with complex cloud - gas interactions. Although the MFASIS solver has some improvements in this regard, the overall treatment of scattering in the study could be more refined.
Citation: https://doi.org/10.5194/egusphere-2025-242-RC2 -
AC1: 'Reply on RC2', Yongbo Zhou, 02 May 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-242/egusphere-2025-242-AC1-supplement.pdf
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-242', Anonymous Referee #1, 12 Mar 2025
Review of the manuscript titled "Optimizing cloud optical parameterizations in RTTOV for data assimilation of satellite visible reflectance data: an assessment using observed and synthetic images" by Yongbo Zhou, Tianrui Cao, and Lijian Zhu.
This paper evaluates parametrizations of visible reflectance simulated from CMA-MESO and ERA5 using RTTOV with observations from instruments onboard the satellites FY-4B and Himawari-9.
Results show that the observed reflectance is higher than modelled reflectance, on average, and that the choice of an optimal configuration yields the lowest bias while only slightly increasing unbiased error (standard deviation of O-B). In clear-sky pixels, the atmospheric contribution to visible reflectance is near zero. Thus, the choice of parametrization had no effect. The bias is induced by the land surface representation in the NWP model. In cloudy sky, bias is more relevant than in clear sky cases, because the bias is responsible for a larger fraction of error, as shown in Figures 10, 11, and 12. In clear sky, bias was mostly lower than unbiased error, except of experiment CM-FY-DM in April (Figure 3).
The authors already nicely addressed my comments to an earlier version of the manuscript in review for another journal. For example, the authors extended the analysis by the standard deviation of O-B. The manuscript is now an extensive evaluation and in a very good condition. I recommend publication after minor comments are addressed.
Minor comments
L243-244: "the increased B for C213 would amplify the O-B differences, leading to the increased standard deviations". The standard deviation (SD) is not influenced by the bias (average O-B). Thus, an change in average B does not change the SD. The explanation for an increased SD must be larger O-B differences, due to B being too high and too low (increase in |O-B|).
Figure 4: The caption seems to have an error, as the x-axis label is "reflectance" not "O-B departure".
Equation 6 seems to be incorrect, not all terms are in the exponent, compared to Equation 2 of McFarquar et al. (2003).
L237, L275: Language: Instead of "was expected", you probably mean "was found"? Or was there an expectation before seeing the results?
L243: You state "opposite circumstance" but the same sign (O-B>0) as in the line above. Probably you mean O-B < 0?
L248: "at the low- or high-reflectance ends": Maybe you can quantify this, e.g. "at low reflectance (<0.1)" or similar.
L271: "The results in Figure 4 suggested that the Baran 2018 ice scheme should be used with caution ...".
I am confused, because Figure 4 shows results for ice_scheme=1 (SSEC) and not for "Baran" parametrization. Maybe you meant Figure 5?Figure 7 caption (c) missing.
L367: Language, "B was consistently underestimated compared with O", I think this should be "B consistently underestimated O".
L393: "The smallest O-B bias". Good, but can you quantify it? For example "the optimal configuration reduced the bias by 0.04 on average, while standard deviation increased by less than 0.005.
Citation: https://doi.org/10.5194/egusphere-2025-242-RC1 -
AC2: 'Reply on RC1', Yongbo Zhou, 02 May 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-242/egusphere-2025-242-AC2-supplement.pdf
-
AC2: 'Reply on RC1', Yongbo Zhou, 02 May 2025
-
RC2: 'Comment on egusphere-2025-242', Anonymous Referee #3, 06 Apr 2025
Overall evaluations
The research is highly relevant as it focuses on optimizing cloud optical parameterizations in RTTOV for satellite visible reflectance data assimilation. With the increasing importance of satellite - based data in weather forecasting and climate studies, improving the accuracy of radiative transfer models like RTTOV is crucial. The use of data from recent geostationary meteorological satellites (FY - 4B and Himawari - 9) makes the study timely. The authors conducted four different experiments, covering two observing systems, two radiative solvers, and two atmospheric fields. A wide range of data was used, including forecasts from the CMA - MESO model, ERA5 reanalysis data, and observations from FY - 4B and Himawari - 9 satellites. The data was carefully processed, and multiple statistical methods were employed, such as analyzing O - B biases, standard deviations, and probability density functions. These analyses provided in - depth insights into the performance of different cloud optical parameterizations. The study identified the optimal cloud optical parameterization suite (“Deff” + “SSEC” + B02), which can significantly reduce the O - B biases. This finding has practical implications for improving the accuracy of radiative transfer simulations and data assimilation in weather prediction models.
Major issues:
- The number concentration in Eq. (1) was set to 300 cm-3 Is this generally applicable for all types of clouds? If not, what is the sensitivity of the simulated reflectivity to the other number concentration?
- For cloud ice, is the effective diameter in Eq. (2) independent of cloud ice content and number concentration? What is the number concentration assumed for cloud ice?
- In CMA-MESO, there are six types of hydrometeors. How is the scattering from rain, snow and graupel is treated in this study?
- Since FY-4B AGRI has an aerosol AOD product, it is better to use AOD to quantify the contribution from aerosol? Eq. (8) and (9) seems to be too objective.
- Captions for Figure 7b do not make sense. What is Fig 7c? Also, it is strange to show O-B bias from one month data.
- The DOM solver simplified the scattering interactions between clouds and gaseous molecules into single - scattering processes in non - cloudy layers. This simplification may lead to inaccuracies in the simulated reflectance, especially in areas with complex cloud - gas interactions. Although the MFASIS solver has some improvements in this regard, the overall treatment of scattering in the study could be more refined.
Citation: https://doi.org/10.5194/egusphere-2025-242-RC2 -
AC1: 'Reply on RC2', Yongbo Zhou, 02 May 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-242/egusphere-2025-242-AC1-supplement.pdf
Peer review completion
Journal article(s) based on this preprint
Data sets
The processed data for evaluating cloud optical parameterizations in RTTOV Yongbo Zhou, Tianrui Cao, and Lijian Zhu https://doi.org/10.5281/zenodo.14642334
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 212 | 90 | 17 | 319 | 14 | 28 |
- HTML: 212
- PDF: 90
- XML: 17
- Total: 319
- BibTeX: 14
- EndNote: 28
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Tianrui Cao
Lijian Zhu
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2285 KB) - Metadata XML