Clare Flynn, Melanie Follette-Cook, Kenneth Pickering, Christopher Loughner, James Crawford, Andrew Weinheimer, Glenn Diskin October 6, 2015 Evaluation of Vertical Mixing in WRFChem during DISCOVER-AQ July 2011 and Impacts on Pollutant Profiles 1

Deriving Information on Surface Conditions from Column and VERtically Resolved Observations Relevant to Air Quality and VERtically Resolved Observations Relevant to Air Quality A NASA Earth Venture campaign intended to improve the interpretation of satellite observations to diagnose near-surface conditions relating to air quality Objectives: 1. Relate column observations to surface conditions for aerosols and key trace gases O 3, NO 2, and CH 2 O 2. Characterize differences in diurnal variation of surface and column observations for key trace gases and aerosols 3. Examine horizontal scales of variability affecting satellites and model calculations NASA P-3B NASA UC-12 NATIVE, EPA AQS, and associated Ground sites Investigation Overview Deployments and key collaborators Maryland, July 2011 (EPA, MDE, UMd, and Howard U.) SJV, California, January/February 2013 (EPA and CARB) Texas, September 2013 (EPA, TCEQ, and U. of Houston) Colorado, Summer

Deployment Strategy Systematic and concurrent observation of column-integrated, surface, and vertically-resolved distributions of aerosols and trace gases relevant to air quality as they evolve throughout the day. 3 NASA UC-12 (Remote sensing) Continuous mapping of aerosols with HSRL and trace gas columns with ACAM NASA P-3B (in situ meas.) In situ profiling of aerosols and trace gases over surface measurement sites Ground sites In situ trace gases and aerosols Remote sensing of trace gas and aerosol columns (Pandora) Ozonesondes Aerosol lidar observations Three major observational components:

Maryland Observing Strategy

Motivation  Boundary layer mixing plays an important role in the connection between column and surface data  Mixing impacts the vertical distribution of pollutants  importance for profile shapes  Profile shape determines which altitude layers contribute most to the column  Impacts how well column measurements relate to surface quantities  Ultimately, how well can satellite column observations represent surface air quality? 5

Motivation  Can a regional, coupled meteorology-air quality model be effectively used to understand the interplay between vertical mixing and pollutant profiles?  Objective of this study to evaluate the representation of boundary layer mixing within the WRFChem model  Important to note that WRFChem is a coupled meteorology- chemistry model!  No MCIP time averaging  Chemistry and meteorology computed in same time step 6

D ij accounts for differences between magnitude of mixing ratios and profile shapes Reference: Hains, J. C., Taubman, B. F., Thompson, A. M., Stehr, J. W., Marufu, L. T., Doddridge, B. G., Dickerson, R. R. (2008), Origins of chemical pollution derived from Mid-Atlantic aircraft profiles using a clustering technique, Atmos. Env., 42, km12 km 4 km

8 WRFChem Simulation Options Maryland D-AQ Campaign Time PeriodJune 27 through July 31, 2011 Chemical mechanismCBM-Z AerosolsMOSAIC with 8 aerosol bins RadiationLongwave-RRTM; Shortwave-Goddard Meteorology and Chemical Inputs NARR; MOZART-4 CTM PBL SchemeYSU (non-local PBL scheme) Surface Layer Scheme; LSM Monin-Obukhov scheme; unified Noah LSM PhotolysisFast-J Follette-Cook, M. B., K. Pickering, J. Crawford, B. Duncan, C. Loughner, G. Diskin, A. Fried, A. Weinheimer (2015), Spatial and temporal variability of trace gas columns derived from WRF/Chem regional model output: Planning for geostationary observations of atmospheric composition, Atmos. Environ., 118, 28-44, doi: /j.atmosenv

Evaluation of Model PBLH  Bias of YSU scheme computed relative to several observational data sets  Bias = WRFChem PBLH – Observational PBLH  All comparisons during daytime (mostly 8am-5pm EDT)  Meteorological estimates of PBLH – based on potential temperature profile  P-3B (available at all 6 spiral sites)  Ozonesonde (available at 2 spiral sites)  Aerosol estimates of PBLH – based on aerosol backscatter profile  MPL (MicroPulse Lidar; available at 3 spiral sites) D ij accounts for differences between magnitude of mixing ratios and profile shapes Reference: Hains, J. C., Taubman, B. F., Thompson, A. M., Stehr, J. W., Marufu, L. T., Doddridge, B. G., Dickerson, R. R. (2008), Origins of chemical pollution derived from Mid-Atlantic aircraft profiles using a clustering technique, Atmos. Env., 42,

Comparison of PBLH Values 10 Small sonde sample size!

Comparison of PBLH Values 11

Average Model PBLH Biases 12 Observational Dataset Model Resolution Mean Bias (m) (Model-Obs.) ± 1σ (m) P3B12km P3B4km Ozonesonde12km Ozonesonde4km MPL12km MPL4km Only MPL demonstrates a statistically significant difference between the 12km and 4km simulations!!

13 PBLH Diurnal Average Behavior

14 PBLH Diurnal Average Behavior Too deepGood relative to sonde—due to fewer samples?

15 PBLH Diurnal Average Behavior

16 PBLH Diurnal Average Behavior PBL too deep and collapses too early relative to MPL mixed layer heights—differences between PBLH based on stability parameters and aerosol backscatter

17 Potential Temperature Profiles Both resolutions reproduce the diurnal variation in ozonesonde theta profiles. However, some struggle with collapse of CBL during evening for both resolutions

18 Potential Temperature Profiles Same story relative to P3B as for the ozonesondes at both resolutions

19 Mixing and Pollutant Median Profiles - CO Simulated and observed profiles compare better during early afternoon than for other times of day.

20 Variability of Pollutant Profiles Error bars represent the 25 th and 75 th percentile values for observed median profile and simulated median profile. Model reproduces the range of the distributions during the afternoon hours within the CBL.

21 Variability of Pollutant Profiles Error bars represent the 25 th and 75 th percentile values for observed median profile and simulated median profile. Model struggles to reproduce median profile and distributions— model not extreme enough.

22 Variability of Pollutant Profiles Error bars represent the 25 th and 75 th percentile values for observed median profile and simulated median profile. Model distributions not extreme enough.

Conclusions  YSU PBL scheme performs differently relative to different types of observational PBLH estimates at both resolutions  Too deep relative to P3B meteorological estimates on average  Too shallow relative to MPL aerosol estimates on average  Reasonably well simulates average diurnal behavior of PBLH relative to meteorological estimates!  Both resolutions also reasonably well capture the diurnal variation in theta profiles relative to the P3B and ozonesondes  Some struggle to capture CBL collapse  Best captures potential temperature and CO median profile shapes during early afternoon when CBL is fully developed 23

Future Work  Run WRFChem with other PBL schemes, such as ACM2, MYJ and MYNN (local schemes), to further investigate vertical mixing issues  Which scheme best captures PBL mixing and height?  How does vertical mixing impact pollutant profiles?  Compare observational PBLH estimate data sets among each other  Also compare against the airborne High Spectral Resolution Lidar (HSRL) PBLH data set  Investigate spatial and temporal variability in the model bias  Investigate impacts on column-surface correlation for O 3 and NO 2 for each PBL scheme evaluated  Which scheme best captures the observed relationship? 24