Improving Ozone Simulations in Asia via Multisource Data Assimilation: Results from an Observing System Simulation Experiment with GEMS Geostationary Satellite Observations
- 1School of Geographical Sciences, Fujian Normal University, Fuzhou, Fujian 350007, China
- 2School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen, Guangdong 518055, China
- 3Guangdong Provincial Observation and Research Station for Coastal Atmosphere and Climate of the Greater Bay Area, Shenzhen, Guangdong 518055, China
- 4Institute of Environmental Studies, Pusan National University, Busan 46241, South Korea
- 5Harvard-Smithsonian Center for Astrophysics, Cambridge, Massachusetts 02138, United States
- 1School of Geographical Sciences, Fujian Normal University, Fuzhou, Fujian 350007, China
- 2School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen, Guangdong 518055, China
- 3Guangdong Provincial Observation and Research Station for Coastal Atmosphere and Climate of the Greater Bay Area, Shenzhen, Guangdong 518055, China
- 4Institute of Environmental Studies, Pusan National University, Busan 46241, South Korea
- 5Harvard-Smithsonian Center for Astrophysics, Cambridge, Massachusetts 02138, United States
Abstract. The applications of geostationary (GEO) satellite measurements at an unprecedented spatial and temporal resolution from the Geostationary Environment Monitoring Spectrometer (GEMS) for monitoring and forecasting the alarming ozone pollution in Asia through data assimilation remain at the early stage. Here we investigate the benefit of multiple ozone observations from GEMS geostationary satellite, low Earth orbit (LEO) satellite, and surface networks on summertime ozone simulations through individual or joint data assimilation, built on our previous Observing System Simulation Experiment (OSSE) framework (Shu et al., 2022). We find that data assimilation better represents the exceedance, spatial patterns, and diurnal variations of surface ozone, with a regional mean negative bias reduction from 2.1 to 0.2–1.2 ppbv in ozone simulations as well as precision improvements of a root-mean-square error (RMSE) of by 5–69 % in most Asian countries. Furthermore, the joint assimilation of GEMS and surface observations performs the best. GEMS also brings direct added value for better reproducing ozone vertical distributions, especially in the middle to upper troposphere at low latitudes, but may mask the added value of LEO measurements, which are crucial to constrain surface and upper tropospheric ozone simulations when observations from other platforms are inadequate. Our study provides a valuable reference for ozone data assimilation as multisource observations become gradually available in the era of GEO satellites.
- Preprint
(11136 KB) -
Supplement
(5627 KB) - BibTeX
- EndNote
Lei Shu et al.
Status: final response (author comments only)
-
RC1: 'Comment on acp-2022-744', Anonymous Referee #2, 31 Dec 2022
General Comments:
The authors provided an OSSE-based analysis to investigate the impacts of satellite and surface O3 observations on the assimilated O3 concentrations over Asia. They found that joint assimilation by assimilating both satellite and surface O3 observations has the best performance. I agree with the authors that joint assimilation is helpful. However, the added value to the assimilated surface O3 by assimilating satellite observations is expected to be limited over areas with a high density of surface observations because of weaker sensitivity to surface O3 and larger observation errors in satellite observations. Additional analysis is suggested to demonstrate the difference between areas with and without a high density of surface observations, as it can clarify the advantage of satellite observations with respect to surface O3 observations. I recommend the paper for publication after consideration of the points below.
Speical Comments:
Lines 141-147: I understand that the fast ozone profile retrieval simulation (FOR) is necessary for GEMS because GEMS scientific products have not been released. I suggest more analysis to demonstrate the consistency between FOR and OMI such as their averaging kernels and observation errors, as the conclusion of this work is based on that FOR is good enough to simulate satellite observations.
Figure 1 and Figure 2: I assume they are simulated retrievals rather than GEMS and OMI retrievals. In addition, the sensitivity of the 839 hPa level (GEMS, Figure 1f) is uniform from the surface to 600 hPa; the sensitivity of the 842 hPa level (OMI, Figure 2c) is uniform from the surface to 300 hPa. Consequently, the contributions of assimilating GEMS and OMI on surface O3 concentration are expected to be limited with respect to surface O3 observations.
Lines 254-256: Can the authors perform a new experiment by only assimilating OMI? I am curious about the improvement which we can obtain by assimilating GEMS instead of OMI.
Figure 5: What is the major added value of assimilating satellite measurements over areas with a high density of surface stations, such as E. China? I noticed that the spatial correlation in China is almost ZERO by assimilating GEMS and is about 0.55 by assimilating surface observations.
Figure 6: Please provide more description for panel d. There are different colors and numbers shown in this panel and I don’t understand what they represent. In addition, because only one-month assimilation is performed in this work, it could be helpful to show the time series of daily or houly O3 concentrations. It can better demonstrate the effect of various assimilations on various temporal scales.
-
RC2: 'Comment on acp-2022-744', Anonymous Referee #1, 02 Jan 2023
The paper presents the OSSE experiments results to demonstrate the benefit of GEMS ozone observations in future applications. Both the methods and the data assimilation results with additional OMI and surface synthetic observations are well presented. However, improvement can be made if the authors can address the following concerns.
General:
When the influence of assimilation frequency is investigate, it is not clear how data assimilation experiments with a longer assimilation time window of 3-hr are carried out. Are the hourly surface station observations averaged inside the 3-hr time window? Are the satellite data inside the 3-hr time window assumed to be valid at one particular instance? If so, when are they supposed to be valid?
The authors found that sometimes the data assimilation has negative effects. For instance, “In Japan and Mongolia, the assimilation of GEMS data generally contributes to a deterioration of simulated ozone and even counteracts the positive impact of surface observations when performing the joint assimilation.” It is not impossible to encounter such cases. When this happens, it is probably worth to investigate the reason for such a behavior. With the current OSSE setting, it is probably not too hard to investigate the underlying causes.
Specific:
Line 20: It is probably better to replace “data assimilation better represents” to “data assimilation improves”
Line 22: RMSE is a accuracy metric rather than a precision measure.
Line 113: Is “optimal estimation” the same as “optimal interpolation”?
Equation 5: It would be better to use "y" for variables in observation space.
Line 196: Xap should be in a vector in observation space, but it appears as a state vector. It is better to clearly differentiate state and observation vectors.
Line 211: It is reasonable to assume no correlation between surface station observations. But it is probably questionable to assume no correlation for satellite observations.
Figure 6d: What do the two different shades of color represent in the lower two panels?
Lei Shu et al.
Lei Shu et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
287 | 94 | 9 | 390 | 27 | 3 | 5 |
- HTML: 287
- PDF: 94
- XML: 9
- Total: 390
- Supplement: 27
- BibTeX: 3
- EndNote: 5
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1