1. Earth&Atmospheric Sciences, Georgia Tech Modeling the impacts of convective transport and lightning NOx production over North America: Dependence on cumulus parameterizations Chun Zhao Advisor: Yuhang Wang

2. Earth&Atmospheric Sciences, Georgia Tech Introduction Convection and associated lightning are two important meteorological processes affecting the production and distribution of tropospheric chemical tracers.

3. Earth&Atmospheric Sciences, Georgia Tech Subgrid cloud convection redistributes chemical tracers vertically and then horizontally. It also deplete the soluble tracers by scavenging process [Wang et al., 2001; Doherty et al., 2005; Choi et al., 2005; Folkins et al., 2006]. http://www.metoffice.gov.uk/research/nwp/numerical/physics/convection_image.html

4. Earth&Atmospheric Sciences, Georgia Tech Lightning NOx (NO 2 +NO) production is the main source of NOx in the upper troposphere. It affects other species (e.g. O 3, PAN, and HNO 4 ) by NOx in the free troposphere. [Hudman et al., 2006]. http://someonewhocares.org/photos/lightning/

5. Earth&Atmospheric Sciences, Georgia Tech Introduction Convection and associated lightning are two important meteorological processes affecting the production and distribution of tropospheric chemical tracers. The parameterizations of sub-grid scale convection and lightning NOx production result in large uncertainties in chemical transport models (CTMs) [Doherty et al., 2005; Choi et al., 2005].

6. Earth&Atmospheric Sciences, Georgia Tech Introduction Two parameterizations of cloud convection (KF-eta and Grell) are implemented in a Regional chEmical trAnsport Model (REAM). Lightning NOx production is parameterized differently in each scheme. MM5 with Grell convective scheme is used to provide meteorological inputs for REAM with Grell scheme (MM5- REAM) [Choi et al., 2005]. WRF with KF-eta convective scheme is used to provide meteorological inputs for REAM with KF-eta scheme (WRF- REAM) [this work].

7. Earth&Atmospheric Sciences, Georgia Tech Data description Data used in the study: Intercontinental Chemical Transport Experiment – North America (INTEX-NA) in summer 2004. CO, NMHCs, NOx, O 3, and HNO 3 …… SCIAMACHY satellite retrieved tropospheric columns of NO 2.

8. Earth&Atmospheric Sciences, Georgia Tech Cloud convective parameterization Ground surface Downdraft End layer Downdraft Original layer Cloud bottom Cloud top Updraft flux entrainmentdetrainment entrainmentdetrainment Downdraft flux Grell Scheme Cloud top Cloud bottom Ground surface Downdraft flux Updraft flux Downdraft original layer KF-eta Scheme

9. Earth&Atmospheric Sciences, Georgia Tech Cloud convective parameterization Cloud information affecting convective transport: Mass flux Cloud top height Cloud bottom height Downdraft original level Entrainment and detrainment

10. Earth&Atmospheric Sciences, Georgia Tech Cloud convective parameterization Cloud information affecting convective transport: Mass flux Cloud top height Cloud bottom height Downdraft original level Entrainment and detrainment

11. Earth&Atmospheric Sciences, Georgia Tech Mean deep convective updraft mass fluxes from WRF and MM5 simulations during summer (June, July, and August) 2004.

12. Earth&Atmospheric Sciences, Georgia Tech Vertical profiles of mean deep convective mass fluxes from WRF and MM5 simulations, and entrainment detrainment rate from WRF simulation during summer 2004.

13. Earth&Atmospheric Sciences, Georgia Tech Cloud convective impact Tracers sensitive to convective transport: CO, C 4 H 10, and C 3 H 8 Tracers scavenged by convective updrafts: Highly soluble tracer HNO 3

14. Earth&Atmospheric Sciences, Georgia Tech Cloud convective vertical transport Convection affects non-soluble tracers through the turnover of their concentrations in the troposphere [Wang et al., 2001]. Convection lifts the non-soluble tracers with elevated concentrations from the low altitude to the high altitude and subsides them with low concentrations from the high altitude to the low altitude, then increasing the concentrations in the upper troposphere and decreasing the concentrations in the lower and middle troposphere. Entrainment and detrainment provide a pathway for mixing convective updrafts and downdrafts with background atmosphere in the middle troposphere.

15. Earth&Atmospheric Sciences, Georgia Tech Convective transport enhances CO, C 4 H 10, and C 3 H 8 concentrations by 15, 150, and 100% respectively in the upper troposphere. Entrainment and detrainment drive the enhancement of the concentrations of tracers in the mid- troposphere (3-7km) in WRF-REAM simulation. Vertical profiles of CO, C 4 H 10, and C 3 H 8 from aircraft measurements, REAM simulations with and without convection.

16. Earth&Atmospheric Sciences, Georgia Tech Spatial distributions of the average convection driven relative changes of the concentrations of CO, C 4 H 10, and C 3 H 8 from REAM simulation during summer 2004.

17. Earth&Atmospheric Sciences, Georgia Tech Cloud convective scavenging Convection affects highly soluble tracers through the scavenging process in convective updrafts [Folkings et al., 2006]. Entrainment and detrainment provide additional channels to diminish the soluble tracers in the free troposphere (above 3 km) through grabbing the soluble tracers in the background atmosphere into convective updrafts and downdrafts.

18. Earth&Atmospheric Sciences, Georgia Tech Vertical profiles of HNO 3 from aircraft measurements and REAM standard simulations.

19. Earth&Atmospheric Sciences, Georgia Tech Vertical profiles of the average convection driven relative changes of the concentrations of HNO3 from REAM simulations during summer 2004.

20. Earth&Atmospheric Sciences, Georgia Tech Cloud convective outflow Convective transport was found to be an important pathway for inflow and outflow of pollution over high-polluted continents such as North America [Wang et al., 2001; Choi et al., 2005; Li et al., 2005; Liang et al., 2007]. INTEX-NA aircraft campaign was conducted with a major goal to understand the mechanisms by which pollutants are lofted and transported during the warm season [Singh et al., 2006].

21. Earth&Atmospheric Sciences, Georgia Tech Concentration of C 3 H 8 along the aircraft tracks in July 28 th and August 6 th. Flight tracks in INTEX-NA campaign over the Atlantic Ocean in July 28 th and August 6 th.

22. Earth&Atmospheric Sciences, Georgia Tech C 3 H 8 concentrations and winds at 300 hpa at 18:00 GMT on July 28 th and August 6 th in 2004 from REAM simulation with and without convection. Flight tracks in INTEX-NA campaign over the Atlantic Ocean in July 28 th and August 6 th.

23. Earth&Atmospheric Sciences, Georgia Tech Cloud convective outflow The higher concentration of C 3 H 8 at 300 hpa in WRF-REAM simulation totally results from the sub-grid convective transport, on the contrary, that in MM5-REAM simulation just partly result from the sub-grid convective transport. So in these two cases, besides sub-grid convective transport, MM5-REAM simulation should include some other vertical transport mechanisms, possibly large-scale convection or convection related warm conveyor belts (WCBs). So it is likely that the vertical transport in MM5-REAM has roots in large-scale convection, which is too large to be parameterized as a sub-grid convection.

24. Earth&Atmospheric Sciences, Georgia Tech Cloud convective impact Through the investigation of convective impact: Vertical transport of non-soluble tracers: The convective impacts on non-soluble tracers are significantly affected by the cumulus cloud top height. The maximum convective impact on non- soluble tracers appears at different altitudes related with cloud top height. The entrainment and detrainment effects are shown in the comparison. They are important for the mixing in the middle troposphere. Convective scavenging of soluble tracers: The model (WRF-REAM) with entrainment and detrainment in convective parameterization simulates more effective scavenging process for highly soluble pollutants (e.g. HNO 3 ) in the free troposphere.

25. Earth&Atmospheric Sciences, Georgia Tech Cloud convective impact Through the investigation of convective impact: Vertical transport of non-soluble tracers: The convective impacts on non-soluble tracers are significantly affected by the cumulus cloud top height. The maximum convective impact on non- soluble tracers appears at different altitudes related with cloud top height. The entrainment and detrainment effects are shown in the comparison. They are important for the mixing in the middle troposphere. Convective scavenging of soluble tracers: The model (WRF-REAM) with entrainment and detrainment in convective parameterization simulates more effective scavenging process for highly soluble pollutants (e.g. HNO 3 ) in the free troposphere.

26. Earth&Atmospheric Sciences, Georgia Tech Cloud convective impact Through the investigation of convective impact: Convective outflow The two days outflow during INTEX-NA were driven by the vertical transport originating from the convection. However, the convection system was parameterized as a sub-grid convection in WRF with KF-eta scheme simulation but was too large to be parameterized as a sub-grid convection in MM5 with Grell scheme simulation, indicating that MM5 with Grell scheme involves less sub-grid convections.

27. Earth&Atmospheric Sciences, Georgia Tech Acknowledgments I would like to thank Dr. Wang for his advisements. I thank Dr. Tao Zeng, Dr. Yunsoo Choi and all my officemates for their helps and suggestions.

28. Earth&Atmospheric Sciences, Georgia Tech Thank you very much!

29. Earth&Atmospheric Sciences, Georgia Tech Lightning NOx production parameterization Lightning NOx production is parameterized as a function of certain meteorological variables: convective updraft mass fluxes (UMF), convective available potential energy (CAPE), cumulus cloud top height, and precipitation [Price et al., 1993; Allen et al., 1999; Choi et al., 2005]. MM5-REAM UMF and CAPE from Grell scheme (Choi et al., 2005) WRF-REAM UMF and CAPE from KF-eta scheme (this work) Cloud-to-ground lightning flash rate is scaled to National Lightning Detection Network (NLDN) observation. Intra-cloud (IC) to cloud-to- ground (CG) flash ratio is calculated following Wang et al. [1998].

30. Earth&Atmospheric Sciences, Georgia Tech Mean deep convective cloud top pressures from the measurements of GOE-10 and GOE-12 satellites and the simulations of WRF and MM5 during July 2004.

31. Earth&Atmospheric Sciences, Georgia Tech Lightning NOx production impact Lightning NOx (NO2+NO) production is a major source of NOx in the upper troposphere and significantly enhances tropospheric NO2 columns over the ocean, where NO2 columns are more sensitive to the lightning NOx [Labrador et al., 2004; Choi et al. 2005; Martin et al., 2006; Hudman et al., 2006]. It provides the best explanation for the highly elevated NOx in the upper troposphere, which is difficult to be explained by deep convective injection of boundary layer pollution and aircraft emissions [Hudmain et al., 2006].

32. Earth&Atmospheric Sciences, Georgia Tech Spatial distributions of average NO 2 concentrations above 8 km, from measurements and REAM simulation.

33. Earth&Atmospheric Sciences, Georgia Tech NO 2 columns from SCIAMACHY, WRF-REAM and MM5-REAM. Lightning enhances tropospheric NO2 columns especially over the ocean by more than a factor of 2. Both WRF-REAM and MM5-REAM reproduce tropospheric NO2 columns well and highly correlate with the measurements by a coefficient 0.73 and 0.63 respectively. MM5-REAM with higher cloud top height (up to 15 km) over-distributes lightning NOx to above 12 km, while WRF-REAM with lower cloud top height (10-12 km) performs better in the vertically distribution of the lightning NOx.

34. Earth&Atmospheric Sciences, Georgia Tech Spatial distributions of average O 3 concentrations above 8 km from measurements and simulations with and without lightning. O3 concentration is increased by up to 20 ppb due to the lightning in the upper troposphere. The upper tropospheric O3 concentration is more sensitive to the lightning over the ocean. Both WRF- REAM and MM5-REAM simulates larger lightning related enhancement of upper troposphere O3 concentration in the regions where lightning NOx production is larger.

35. Earth&Atmospheric Sciences, Georgia Tech Through the investigation of lightning impacts: Lightning effects are found to be important in simulations. Lightning enhances the concentrations of NO 2 by a factor of 2 in the upper troposphere. It also enhances O 3 concentration by up to 20 ppb in the upper troposphere. The lightning NOx production and distribution are sensitive to cumulus cloud top height in the parameterization following Wang et al. [1998] and Allen et al. [1999]. The WRF-REAM model with lower cloud top height (10-12 km) simulates much better lightning NOx production and distribution than the MM5-REAM model with higher cumulus cloud top height (up to 15 km). Lightning NOx production parameterization

36. Earth&Atmospheric Sciences, Georgia Tech Through the investigation of lightning impacts: Lightning effects are found to be important in simulations. Lightning enhances the concentrations of NO 2 by a factor of 2 in the upper troposphere. It also enhances O 3 concentration by up to 20 ppb in the upper troposphere. The lightning NOx production and distribution are sensitive to cumulus cloud top height in the parameterization following Wang et al. [1998] and Allen et al. [1999]. The WRF-REAM model with lower cloud top height (10-12 km) simulates much better lightning NOx production and distribution than the MM5-REAM model with higher cumulus cloud top height (up to 15 km). Lightning NOx production parameterization

37. Earth&Atmospheric Sciences, Georgia Tech Summary The cumulus cloud top height (CTH) is a critical factor in determining the convective impacts on chemical tracers. The CTH is also a significant factor controlling lightning NOx production and distribution. WRF-REAM model with lower CTH (10-12 km) simulates better lightning NOx production and distribution than the MM5-REAM model with higher CTH (up to 15 km). The entrainment and detrainment provide effective scavenging processes for soluble pollutants. WRF-REAM model with convective parameterization including entrainment and detrainment simulates better highly soluble pollutants (e.g. HNO 3 ) in the free troposphere. KF-eta scheme are found to provide better simulations for tropospheric trace gases than Grell scheme.

38. Earth&Atmospheric Sciences, Georgia Tech Acknowledgments I would like to thank Dr. Wang for his advisements. I thank Dr. Tao Zeng, Dr. Yunsoo Choi and all my officemates for their helps and suggestions.

39. Earth&Atmospheric Sciences, Georgia Tech Thank you very much!