Abstract
A 16-year (2003–2018) evaluation of CO, NO2, SO2, O3, and PM10 from three reanalysis products (CAMS, TCR-2, and MERRA-2) against independent observations and intercomparison among reanalyses are performed over South Korea. All three reanalyses show significant and persistent biases in the five pollutants, but the reanalyses partly capture decreasing trends in CO, NO2, SO2, and PM10 and increasing trends in O3 over the 16-year period. CAMS outperforms TCR-2 and MERRA-2 in terms of area-averaged climatology and trends of all species except for SO2. SO2 is best reproduced in TCR-2. The long-term averaged spatial distributions of the pollutants in CAMS reveal that their spatial distributions are largely influenced by a priori emissions. All reanalysis products show larger biases in CO, NO2, and SO2 in cold months than in warm months. The average daily correlations between bias-corrected reanalysis data and observations are found to be reasonably high for some species: 0.74–0.82 for CO, NO2, and PM10 in CAMS, 0.66 for SO2 in TCR-2, and 0.7 for PM10 in MERRA-2. Although TCR-2 generally shows lower daily correlations than CAMS, it shows the potential that correction of emissions through data assimilation can improve its performance at seasonal and interannual time scales. It is recommended that a bias correction be applied to the reanalysis products for use in regional air quality modeling as initial/boundary conditions or in other relevant studies because each reanalysis product has persistent biases throughout season/year. The evaluation of vertical profiles during KORUS-AQ campaign indicates that CAMS reasonably captures CO, SO2, O3, and aerosols but underestimates NO2.
| Original language | English |
|---|---|
| Article number | 118062 |
| Journal | Atmospheric Environment |
| Volume | 246 |
| DOIs | |
| Publication status | Published - 2021 Feb 1 |
Bibliographical note
Funding Information:We thank two anonymous reviewers for providing helpful comments and suggestions. This study is supported by the Korea Meteorological Administration Research and Development Program under Grant KMI2020-01413.
Funding Information:
Despite the short term availability of PM2.5 observations, we compare the PM2.5 performance of CAMS and MERRA-2 during 2015?2018 (Figs. S5?S8, supplementary). It is found that the PM2.5 performance is similar to PM10 one both for CAMS and MERRA-2. One notable result is that the PM2.5 performance does not deteriorate in springtime unlike PM10 (Figs. S7?S8). Because PM2.5 is much less affected by dust particles than PM10, this result also supports the implication that both CAMS and MERRA-2 need to improve their skills in capturing dust in springtime over South Korea and likely over East Asia.When assessing the bias-corrected daily concentrations, CAMS generally shows the best performance with the lowest daily RMSEs and highest daily correlation coefficients among the reanalyses. In terms of area-averaged PM10, the performance of MERRA-2 is comparable to that of CAMS, except for underestimated PM10 in winter in MERRA-2. Both CAMS and MERRA-2 show large errors in capturing Asian dust in springtime. Even though daily variations of area-averaged concentrations (PM10) are satisfactorily captured in CAMS (MERRA-2), the discrepancies in long-term averaged spatial distributions between the reanalyses and observations suggest that the distributions of increments are constrained by a priori emissions. The large overestimations in CAMS CO, SO2, aerosols (OC and BC) over Seoul also support that more accurate and better spatially resolved emissions are utmost required.We thank two anonymous reviewers for providing helpful comments and suggestions. This study is supported by the Korea Meteorological Administration Research and Development Program under Grant KMI2020-01413.
Publisher Copyright:
© 2020
All Science Journal Classification (ASJC) codes
- Environmental Science(all)
- Atmospheric Science