1. 1 Surface O 3 over Beijing: Constraints from New Surface Observations Yuxuan Wang, Mike B. McElroy, J. William Munger School of Engineering and Applied Sciences, Harvard University Jiming Hao, Hong Ma, Yaosheng Chen Tsinghua University, Beijing, China Tsinghua University, Beijing, China GC 3 rd Users Meeting, April 15, 2007

2. 2 http://www.fallingrain.com/world/CH/22/Miyun.html Beijing Miyun Station

3. 3 Overview of Miyun Station 1.Established by the China Project; started in Nov 2004, continuously to date 2.Miyun ( 40.3N, 116.8E ): Location chosen to observe both clean continental air and pollution in the Beijng plume 3.Directly observe trace-gas covariances near the source; 4.Species measured: CO, O 3, CO 2 ( NO x, NO y, SO 2 ) 5.Data used here: 30-min averages prepared by J.W. Munger

4. 4 Sometimes It’s Clear Courtesy of Bill Munger

5. 5 Mean Flow Patterns at the surface East Asian Monsoon JanJuly

6. 6 Seasonal variations of other factors Qiu et al., Atmos. Environ., 2000 Beijing AOD (1984-1994) AOD Model sulfate AOD peaks in July (0.5) Precipitation obs model

7. 7 O 3 observations: monthly statistics distributions of O 3 skewed toward high values in summer months O 3 peak in early summer and early fall Summertime ozone pollution daytime daytime O 3 80ppbv Legend variations in summer

8. 8 Seasonality of Ozone at Lin An Wang et al., 2002 obs: daytime mean (1999-2000) model

9. 9 CO observations : monthly statistics small seasonality in CO clean continental air and local pollution distributions of CO close to bimodal: clean continental air and local pollution Legend daytime CO variations in summer >3ppm CO!

10. 10 GEOS-Chem Model Set-up  GEOS-4 global simulation; 2 o x 2.5 o resolution; version 7-04-09  Updated Chinese emissions of CO and NO x with monthly variability  Simulation period: Nov 2004 – Dec 2006. Focus on 2006  1-hr model outputs sampled by observations.

11. 11 Chinese emissions of CO and NO x NO x (2004) CO (2001) Zhang, Streets, et al., 2007 China total: 17.8 TgNO 2 (up from 11.3 for TRACE-P) Streets, et al., 2006 China total: 146 Tg/yr (up from 100)

12. 12 Monthly variations in Chinese emissions NO x (1.3: 1) Zhang, Streets, et al., 2007 CO (1.5:1) heating and residential Seasonality not yet in the standard model

13. 13 Soil NO x emissions Wang et al., 2007 a priori soil a posteriori soil derived from 3-year GOME observations over east China a posteriori 0.85 TgN/yr vs a priori 0.24 TgN/yr peak in summer, significant in spring and fall

14. 14 Simulated Seasonal Variations (O 3 ) model captures the overall seasonal patterns model fails to simulate the decrease of O 3 in July model cannot reproduce very high ozone levels Afternoon model model high bias in summer

15. 15 Simulated Seasonal Variations (CO) model tends to underestimate mean CO concentrations model captures the overall seasonal patterns model

16. 16 CO-O 3 relationship Parrish et al., 1998 Miyun O 3 /CO ratio is much lower at Miyun seasonal variability are generally consistent between the two sites Sable Island small correlation in July

17. 17 Average diurnal pattern of summertime ozone obs June obs July Model July very low ozone in the afternoon! (< 20 ppb!) not captured by model!

18. 18 Low Ozone Events in July obs model 9-10 18 24-26 21 28 61 rain or thunderstorm!

19. 19 Flow patterns between June and July July June

20. 20 frequency distribution of CO O 3 levels as a function of CO R.H. as a function of CO June July June July June July 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 4000 O3O3 RHRH <200ppb

21. 21 July CO-O 3 Relationship model Sensitivity: increase cloud optical depth Impact of precipitation and cloudiness on local ozone production!

22. 22 Summary  Observations reveal seasonal variations in O 3, CO, and their correlations in north China region  Model capture the general characteristics in observations of CO and O 3  Major discrepancies in the model :  Underestimate CO concentrations  Not able to represent small-scale dynamics (in transition seasons)  Not able to capture low-ozone events in July, likely due to underestimate of cloudiness  still work in progress….  Need for a finer-resolution model -- nested-grid capability in GEOS-5