1. EAS 4480 – Data Analysis James Hite

2.  Motivation: Census Growth  Monitoring Sites (5)  Sdekalb, Yorkville  Bledsoe,Douglasville, Conyers  O 3, CO, NO x, PM 2.5  Previous Study on VOCs 2000-2010 Population Change DeKalb: (43,737, 16.7%) Douglas: (16,847, 48.4%) Spaulding: (3,776, 16.4%) Rockdale: (8,190, 32.7%) Paulding: (22,856, 78.1%) Polk: -- (1,849, 14.6%) Haralson: -- (1,568, 12.3%) (expressed in housing units)

3.  Initial Results  Overall Negative Trends - in line with VOC study  Interesting points ▪ Full positive: Conyers, Sdklb O 3 ▪ 00-10: none/negative  Motivation for reanalysis (averaging) –diurnal cycle Method ======= LS Regression, Bootstrap resampling to determine 95% normal confidence interval

5.  Overall: increase in negative slopes  Daily: Most retained negative trends except  Full&00-10 Conyers O3 – no trend  Neg/none: Douglasville O3; pos/neg Sdekalb O3  Monthly: Less noise  less apparent trends  (in addition) Sdekalb O3: no full trend (00-10 less)  No-trends: Bledsoe O3, Sdkelab PMfine  00-10 Yorkville O3  no trend (full still neg.)

6. 10-year loss: -7.20E-03 ppmv (daily) vs -6.77E-03 ppmv (monthly

7.  Linear Trends identified across indv. months  Interesting results:  JFM, ND off-season (Bedsole): positive trends  (other stations don’t really measure at this time)  Conyers Apr. (+), Aug. (-)  Sdekalb Jul./Aug. (-) Trends in Individual Monthly Averages: OzoneSlopes: ppmv/year MonthBedsoleConyersDouglasvilleSdekalbYorkville Jan.0.000539no data Feb.0.0014580.000724371no data Mar.0.0007840.0004454660.0006979190.0004550.000549 Apr0.00030.0005704810.0005724932.65E-050.000388 May-0.00054-0.000167528-0.000340723-0.00044-0.00057 Jun-8.43E-050.0005902197.82E-05-0.00032-0.00039 Jul-3.79E-05-0.000521273-0.000813447-0.00113-0.00087 Aug-0.00066-0.000965664-0.000811305-0.00126-0.0008 Sep9.63E-05-5.76E-050.0005018-0.000380.000118 Oct-0.00044-2.96E-050.000127452-0.00038-0.00018 Nov0.000632no data 0.001812-0.00134 Dec0.000905no data

8. Mar., Nov. off-season (Bedsole): positive trends Conyers : Aug. (-) Sdekalb : Aug. (-) Note: other sites have similar slopes, did not pass resampling test for 95% confidence interval Interesting: less urban areas have higher [March]

9.  Yorkville / sdekalb pollutants: all significant trends were negative  PSD/CPSD Analysis  Bedsole plot:  (Lombscargle method on 1-hr avgs)  … took ~20 minutes

11.  Separate psd analysis @ each station  Cpsd analysis & phase lag  Daily Nox cpsd shows a shared weekly cycle with a lag of ~.5 day (york  sdekalb), would expect this since prevailing wind is easterly (coherence ~.7)

12.  Data Source: http://georgiaair.org/amp/export.php Georgia Department of Natural Resources, Environmental Protection Division, Air Protection Branch Ambient Monitoring Program - Susan Zimmer-Dauphinee, Program Manager http://georgiaair.org/amp/export.php  Blanchard et. al. 2010  2009 Ambient Air Surveillance Report