Correlation and Time Correlation and Time Series Analysis of Series Analysis of

Tropospheric Tropospheric Hydroxyl Radicals Hydroxyl Radicals

and Solar Ultraviolet and Solar Ultraviolet RadiationRadiation

Zhijun ZhaoZhijun Zhao

EAS4803 Class ProjectEAS4803 Class Project

April, 2008April, 2008

BackgroundBackground

Hydroxyl radical (OH): Hydroxyl radical (OH): Atmospheric detergentAtmospheric detergent

Extremely reactiveExtremely reactive

Lifetime <1sLifetime <1s

Concentration < 1 x 10Concentration < 1 x 1077 molecule cm molecule cm-3-3 (0.4pptv) (0.4pptv)

Major Source:

O3 + hν → O(1D) + O2 ( λ < 330 nm)

O(1D) + H2O → 2 OH

Major Sinks:

OH + CO + O2 → CO2 + HO2

OH + hydrocarbons, etc → HO2 + …

OH + NO2 + M → HNO3 + M

ObjectiveObjective

Use measured long term OH and Use measured long term OH and J(O1D) data to verify the dependence J(O1D) data to verify the dependence of [OH] and solar ultraviolet radiationof [OH] and solar ultraviolet radiation

Employ time series analysis to Employ time series analysis to illustrate the periodicity of [OH] and illustrate the periodicity of [OH] and solar ultraviolet radiationsolar ultraviolet radiation

Use cross spectral analysis to show Use cross spectral analysis to show their correlation in the frequency their correlation in the frequency domain.domain.

Data sourceData sourceMeteorological Observatory Meteorological Observatory Hohenpeissenberg (MOHp) in GermanyHohenpeissenberg (MOHp) in Germany

1999-20031999-2003

Data converted to Data converted to 1 hour1 hour average. average.

In this project, data have been sorted into In this project, data have been sorted into monthly averagesmonthly averages in some of the in some of the experiment to show the long term behaviorexperiment to show the long term behavior

Acknowledge to Prof. Harald Berresheim in National University of Ireland Galway and Prof. Franz Rohrer in Forschungszentrum Juelich  (FZJ), Institut ICG-2: Troposphare    

Rohrer and Berresheim, Nature, 442, 13, 184-187, 2006      

DataData

1 hour data1 hour data monthly datamonthly data

Regression and Correlation Regression and Correlation AnalysisAnalysis

1 hour data1 hour data monthly datamonthly data

Time Series AnalysisTime Series AnalysisFor the whole 1h data, use Least-square Spectral Analysis (LSSA) For the whole 1h data, use Least-square Spectral Analysis (LSSA) by Lomb-Scargle method because data are not evenly distributedby Lomb-Scargle method because data are not evenly distributed

J(O1D) PSD [OH] PSD

Major frequencies:

~1 (1 day)

~0.0028 (360 day)

Select a small portion of data, interpolate to get evenly Select a small portion of data, interpolate to get evenly distributed data, use periodogram, obtain major long frequencydistributed data, use periodogram, obtain major long frequency

Time Series AnalysisTime Series Analysis

For monthly average dataFor monthly average data

Major period:

12 month

Cross Spectral AnalysisCross Spectral Analysis

Cross-spectral analysis via Welch method

Coherence

ConclusionConclusion

Regression and time series analysis were Regression and time series analysis were used to study the [OH] and J(Oused to study the [OH] and J(O11D).D).

Strong correlation between [OH] and Strong correlation between [OH] and J(OJ(O11D) was found with little annual D) was found with little annual variation.variation.

The daily and annual variation of both The daily and annual variation of both [OH] and J(O[OH] and J(O11D) was demonstrated by D) was demonstrated by periodogram, which are highly periodogram, which are highly correlated shown by cross spectral correlated shown by cross spectral analysisanalysis