Baseline Ozone in Western North America: Measurements and Models David Parrish CIRES University of Colorado NOAA/ESRL Chemical Sciences Division Boulder, Colorado USA Consultant with David.D.Parrish, LLC
Acknowledgements: Many people have provided O 3 data sets ~ 575 years of data from ~ 29 sites! Results from 3 CCMs: J.-F. Lamarque NCAR CAM-chem V. Naik, L. Horowitz NOAA GFDL-CM3 D. T. Shindell - GISS-E2-R Free running meteorology with similar emissions Used for latest IPCC Report AR5
Modeled daily maximum O 3 (ppbv) Importance of CCMs: Calculate Background O 3 Is it possible to meet NAAQS through U.S. controls? Proposed NAAQS Green bars = background O 3 Assimilated meteorology Measured daily maximum O 3 (ppbv) M. Lin et al., JGR, 2014 Green bars show policy-relevant background O 3 for April June 2010 at polluted sites in Southern California, California s Central Valley, and Las Vegas, Nevada (minimum, 25 th, 50 th, 75 th percentiles and maximum).
Critical Questions: What is the magnitude of baseline ozone and how is it changing? How accurate are model calculations of baseline ozone? Today s Approach to answers: Quantify and compare measured and modeled: Long-term O 3 changes Brief look at global emission inventories Seasonal cycles of O 3 in the marine troposphere
Approximately baseline sites: Northern mid-latitudes: Japan and western North America Continental Asia outflow Combine 5 NA Pacific MBL data sets Guiding ideas: Combine 3 Japanese MBL data sets O 3 lifetime in free troposphere is long enough that similar trends are expected at all of these sites. Continuous, slowly varying trends are expected. Include as many data as possible to improve precision seasonal means at many sites Pay close attention to confidence limits I will give 95% confidence intervals Avoid continental influence 2 mountain sites Trinidad Head Cooper et al., 2010 NA free troposphere data set
Quantify and Compare measurements and models Long-term changes In my opinion, this includes all available data as of 2011 Could be updated with more recent data and other sites (Mt. Bachelor, Cheeka Peak) Parrish et al., JGR, 2014
Quantify and Compare measurements and models Long-term changes Similar long-term changes for all sites when normalized (true in all seasons) Spring Measurements Normalize by dividing by year 2000 intercept Parrish et al., JGR, 2014
Quantify and Compare measurements and models Long-term changes Similar long-term changes for all sites (in all seasons) when normalized Polynomial fits give Shape Factors that define longterm changes for all sites (different between seasons) (Power Series expansion with year 2000 as origin; keep terms with statistically significant coefficients.) Spring Measurements Parrish et al., JGR, 2014
Quantify and Compare measurements and models Long-term changes Similar long-term changes for all sites (in all seasons) when normalized Polynomial fits give Shape Factors that define longterm changes for all sites (different between seasons) Spring Measurements O 3 = a + bt + ct 2 (t = year -2000) a = 100.5 ± 1.5, b = 0.93 ± 0.16, c = -0.021 ± 0.022 Parrish et al., JGR, 2014
Quantify and Compare measurements and models Long-term changes Similar long-term changes for all sites (in all seasons) when normalized Polynomial fits give Shape Factors that define longterm changes for all sites (different between seasons) For comparing measured and modeled trends, exact colocation is not important! OMeasurements 3 = a + bt + ct 2 and Models: Normalization (t = year -2000) gives similar a long-term = 100.5 ± changes 1.5, b = 0.93 for ± all sites 0.16, (in c = all -0.021 seasons). ± 0.022 Parrish et al., JGR, 2014
Quantify and Compare measurements and models Long-term changes Measurements: O 3 continues to increase in spring Models: O 3 maximum reached, now decreasing Parrish et al., JGR, 2014
Quantify and Compare measurements and models Long-term changes Measurements: O 3 now decreasing in summer Models: O 3 maximum reached earlier We have talked about increasing background O 3. This is no longer the case during the high O 3 season! Parrish et al., JGR, 2014
Quantify and Compare measurements and models Long-term changes Measurements: O 3 has increased less in autumn than other seasons Models: O 3 maximum reached earlier Parrish et al., JGR, 2014
Quantify and Compare measurements and models Long-term changes Measurements: O 3 still increasing in winter. Models: O 3 maximum appears to be coming earlier Parrish et al., JGR, 2014
NOx to CO molar ratio Quantify and Compare measurements and models Brief look at global emission inventories 0.50 Inventories for Los Angeles Basin ACCMIP MACCity Global emission inventories require improvement: If Emission input is not accurate, Model output cannot be accurate. 0.10 0.05 Measurements Los Angeles [Pollack et al.,2013] 0.01 1960 1980 2000 (NOx to CO ratio is proxy for NOx to VOC ratio, which controls pollution photochemistry) Hassler et al., 2015, in preparation
Quantify and Compare measurements and models O 3 seasonal cycles in marine troposphere Trinidad Head Storofdi, Iceland Mace Head 7 marine boundary layer sites: 3 northern mid-latitudes 1 tropical Samoa 3 southern mid-latitudes Ushuaia, Argentina Cape Point Cape Grim Approximately baseline sites
Quantify and Compare measurements and models Seasonal cycles 29 years of monthly averages Ian Galbally - CSIRO Detrend, Calculate Fourier Transform
Quantify and Compare measurements and models Seasonal cycles 29 years of monthly averages Only fundamental and 2 nd harmonic significant. Fundamental Two, and only two, terms are significant in measured and all modeled seasonal cycles at all 7 sites. 2 nd Harmonic Detrend, Calculate Fourier Transform
Quantify and Compare measurements and models Seasonal cycles 29 years of monthly averages 5 parameters define average seasonal cycle: Annual average (Y 0 ), 2 magnitudes (A 1, A 2 ), 2 phases (f 1, f 2 ) Provide basis for quantitative comparisons Fundamental Cape Grim 2 nd Harmonic Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Seasonal cycles 29 years of monthly averages All sites have a late winter to early spring maximum and a summer minimum Fundamental Pacific MBL 2 nd Harmonic Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Seasonal cycles All sites have a late winter to early spring maximum and a summer minimum Highest ozone at northern mid-latitudes, lowest in tropics Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Seasonal cycles All sites have a late winter to early spring maximum and a summer minimum Highest ozone at northern mid-latitudes, lowest in tropics Models reproduce seasonal cycles reasonably well in the marine boundary layer Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Seasonal cycles Trinidad Head: 2 nd harmonic is large relative to fundamental; secondary maximum in fall Models overestimate MBL baseline O 3 by 10-17 ppb (30-52%) Relative contributions of fundamental and second harmonic differ widely Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Altitude dependence of seasonal cycles Hypothetical Picture: Stratospheric influence dominates in upper FT spring seasonal max Photochemical production dominates in lower FT May-June seasonal max Photochemical destruction dominates in MBL summer minimum, late winter seasonal maximum Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Altitude dependence of seasonal cycles Hypothetical Picture: Model results do not fit this hypothetical picture: No strong shift in seasonal cycle above MBL Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Altitude dependence of seasonal cycles Hypothetical Picture: Model results do not fit this hypothetical picture: No strong shift in seasonal cycle above MBL O 3 sharply reduced in MBL Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Altitude dependence of seasonal cycles Hypothetical Picture: Model results do not fit this hypothetical picture: No strong shift in seasonal cycle above MBL No sharp reduction in O 3 within MBL O 3 sharply reduced in MBL Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Altitude dependence of seasonal cycles Hypothetical Picture: Model results do not fit this hypothetical picture No strong shift in seasonal cycle above MBL No sharp reduction in O 3 within MBL 2 nd harmonic confined to MBL Fit sine functions to fundamental and 2 nd harmonic
Quantify and Compare measurements and models Altitude dependence of seasonal cycles Hypothetical Picture: Model results do not fit this hypothetical picture No strong shift in seasonal cycle above MBL No sharp reduction in O 3 within MBL 2 nd harmonic term of seasonal cycle present throughout troposphere MBL structure and dynamics? Fit sine functions to fundamental and 2 nd harmonic
Conclusions: Baseline O 3 transported from Pacific to North America increased rapidly before 2000, but decrease has stopped in summer and autumn. Models (at least these 3 CCMs) cannot quantitatively calculate baseline O 3 : Give incorrect trends Overestimate magnitude by 30-50% in MBL Give incorrect altitude gradients Why? Time evolution of emission inventories are poorly defined. Marine boundary layer dynamics appear to be poorly described. Others? Note: Hess et al., ACP, 2015 find that stratosphere to troposphere flux and methane chemistry play major roles in tropospheric O 3 trends.
Quantify and Compare measurements and models Seasonal cycles 29 years of monthly averages 5 parameters define average seasonal cycle: Annual average (Y 0 ), 2 magnitudes (A 1, A 2 ), 2 phases (f 1, f 2 ) Fundamental Provide basis for quantitative comparisons 2 nd Harmonic Fit sine functions to fundamental and 2 nd harmonic