The Analysis of Ozone Dependence on Synoptic Weather Patterns

1
The Analysis of Ozone Dependence on Synoptic
Weather Patterns
7th Annual CMAS Conference, October 6th , 2008

• Fong (Fantine) Ngan and DaeWon Byun
• IMAQS, Department of Earth Sciences,
• University of Houston

2
Contents

• Introduction
• Methodology
• Cluster Characteristics
• Clusters .vs. Ozone Episodes
• Clusters .vs. CMAQ Forecasting
• Summary and Future Work

3
Introduction

• Houston/Galveston area is a unique environment
  for O3 development.
• (Various emissions complex geophysical
  characteristics)
• Banta et al. (2005), Darby (2005) Met. factors
  are a key on determining the production, dilution
  and transport of O3.
• Muller Jackson (1985), Eder et al. (1995),
  Davis et al. (1998) The development of O3
  levels depends greatly on meteorological
  conditions which affect photochemical reaction
  and transport of O3 precursors.
• Appel et al. (2007) model biases in O3
  prediction vary based on the meteorological
  conditions.

Synoptic flow patterns have been classified
during 2005/2006 TexAQS-II. Understanding the
dependence of O3 on meteorology How high O3
events are associated with certain weather
scenarios Evaluating AQF results according to
the weather clusters
4
Methodology

• The study period encompassed 5-month ozone season
  (May-September) recorded in 2005 and 2006 5 days
  were excluded since they were influenced strongly
  by Hurricanes. (total 301 days)
• The meteorological grid data (100100 in 12-km
  resolution) from UH-RDAS (NAM objective
  analysis) was used for the cluster analysis.
• The weather conditions were categorized using a
  two-stage (average linkage then convergent
  k-means) clustering approach based on the 850 mb
  wind fields (U V components of wind) at 12 UTC.

Principle Component Analysis reduce size of
dataset
850mb U/V from UH-RDAS
12-km domain
8 PCs (80)
5
Weather Clusters (composite maps on 850 mb)
C2 E, 56 days (19)
C3 clam, 69 days (23)
C1 SE/S, 93 days (30)
C5 SW, 40 days (13)
C6 strong SW, 2 days (1)
C4 N, 41 days (14)
6
ClusterCharacteristics
C4
Mean hourly dew point depression associated with
clusters
C2, C3
CAMS sites cluster average of T and T-TD C4
(northerly) Lowest min T and driest C1 C5
(southerly) Relative lower T and more humid C6
(warm section) highest T and most humid
C1, C5, C6
Mean hourly temperature associated with clusters
C6
C4
7
ClusterCharacteristics
Mean hourly wind SPD associated with clusters
C6

• CAMS sites cluster average of wind speed O3
  conc.
• Good correlation of cluster wind speed O3 conc.
• Lower wind, higher O3
• Higher O3 level group
• C2 (easterly) 65 ppb
• C3 (clam condition) 61 ppb
• C4 (northerly) 56 ppb
• Peak happens at 14 CST
• Lower O3 level group
• C1 (southerly) 35 ppb
• C5 (southwesterly) 40 ppb
• Lowest O3 level
• C6 (strong SW) 20 ppb

C1
C5
C2, C3, C4
Mean hourly ozone associated with clusters
C2, C3, C4
C5
C1
C6
8
Data Source and Ozone Episodes
Both met. and chemical data are from CAMS sites
(TCEQ). Around 45 sites are available in HGA
(region 12) during the study period.
CAMS sites map
Criteria for defining O3 Episodes 8 hr average
O3 gt 85 ppb 92 ozone events Criteria for
defining rainy days precip. recorded during 10
16 CST 98 rainy days
9
Black bars number days White bars number of O3
days
Percentages ratio of O3 days and total days in
clusters
10
Cluster Characteristics during TexAQS-II
intensive period
O3 events occurred in C2, C3 C4 Precip.
Prevents O3 events
Sub-tropical high was dominant No C4 was
identified
11
Location of O3 peak
C2

• Observed O3 (CAMS sites, 8-hr average) on event
  days have been sorted. The three sites with the
  highest values were marked on the maps.
• The circle color indicates how many times were
  marked at that CAMS site.
• O3 peak is subjected to the flow pattern.
• C2 SW of downtown
• C3 wide spread over the city
• C4 S of Harris County

C4
C3
Harris
Brazoria
12
Clusters .vs. CMAQ Forecasting (O3)
C2

• Mean max 8hr O3 for clusters
• UH-AQF (F1, shaded)
• CAMS OBS (circle)

C3
Figures are generated by Dr. HyunCheol Kim
13
Clusters .vs. CMAQ Forecasting (O3)
Mean max 8hr O3 for clusters UH-AQF (F1,
shaded), CAMS OBS (circle)
C4
Mean max O3 for cluster 4 in AQF further south
than CAMS observations
by Dr. HyunCheol Kim
14
C4 case September 13, 2006

• Weather condition
• Mostly clear to partly cloudy
• Light northerly winds all day
• Weak sea breeze in late afternoon
• Max T 93F at IAH
• Front passed through HGA
• It is classified as cluster 4.

SFC weather map at 06z 9/13
SFC weather map at 12z 9/13
SFC weather map at 18z 9/13
15
Snap shot of O3 wind on 9/13UH-AQF (shaded)
CAMS (circle)
06 CST

• O3 level started to build up at noon.
• Too strong northerly (northeasterly) wind in the
  morning pushed pollutants southward (or
  southwestward).
• O3 peak ended up in Brazoria.

Figures are from UH-AQF web display
12 CST
18 CST
16
Snap shot of model observed wind on 9/13UH-AQF
(shaded) CAMS (circle)

• Over-prediction of wind speed in the morning

08 CST
Figures are from UH-AQF web display
17
Time Series of O3 on 9/13 forecasting at CAMS
site C1 C619 Red AQF O3 Blue observed O3
Located in Houston downtown Under-prediction of O3
Located south of Houston Good estimation of O3
Figures are from UH-AQF web display
18
Summary Future Work

• Synoptic weather patterns in HGA during summer
  2005/2006 were classified into 6 clusters.
• C1, C5 C6 (southerly) lower mean hourly O3,
  higher average wind speed.
• C2 (easterly), C3 (stagnant condition) and C4
  (northerly) have larger contributions on high O3
  events. Precip., clouds and strong wind prevent
  O3 development in these clusters.
• August 2006 dominated by southerly flow (C1
  C5), less O3 events
• September 2006 frequent frontal passage (C4),
  more O3 events (C2 C3)
• In cluster 4 (northerly), CMAQ forecasting of O3
  peak was placed south of the city due to
  over-prediction of wind speed in the morning.

• Include more years for weather pattern
  classification (2007 2008)
• Develop secondary factor base on weather pattern
  for O3 forecasting