EVALUATION OF RAGWEED FORECASTING IN TULSA

1
EVALUATION OF RAGWEED FORECASTING IN TULSA

• Estelle Levetin, PhD
• The University of Tulsa

2
Ambrosia Pollen

• Most important pollen allergen in N.A.
• In Tulsa area, cumulative Ambrosia pollen is
  first or second in terms of yearly abundance
• The ability to accurately predict day to day
  pollen levels could provide important benefit to
  sensitive individuals either by avoidance or by
  taking prophylactic medication

3
Stand of Ambrosia trifida along the east bank of
the Arkansas River
4
Flowering in Ragweed

• Controlled by photoperiod
• Pollination is the same time each year at a given
  location unless stressful climatic conditions
  influence growth and reproduction in the plants.
• Once pollination begins, pollen release and
  atmospheric pollen concentrations are influenced
  by meteorological conditions.

5
Pollen Forecasts from TU

• Multiple regression models
• Empirical model for mountain cedar pollen release
  coupled with HY-SPLIT dispersion model
• Development of ragweed forecasts
• Empirical Model
• Ragweed Pollen Forecaster (computer software)
  generated by 6 students from Dept of Computer
  Science (Cyber Corp)

6
Air Sampling

• Burkard Spore Trap has been used for air sampling
  in Tulsa since Dec. 1986
• Ragweed data from 1987 to 2001 was used to
  determine pollen season characteristics
• Start date - 5 of season total)
• End date 95 of season total
• Typical peak date

7
(No Transcript)
8
Pollen Season Characteristics

• Mean start date (5 of season total) 27 Aug
• Mean end date (95 of season total) 11 Oct
• Mean peak date 10 Sep

9
Empirical Model

• Pollen concentrations compared with data from the
  National Weather Service to determine the effects
  of meteorological conditions on airborne pollen
  levels
• Empirical model was developed based on phenology
  and the weather forecast
• NGM-MOS 60 hour forecasts were used
• Model was used to generate pollen forecasts for
  the 2002 and 2003 ragweed seasons
• Comparison with the atmospheric ragweed pollen
  concentrations was used to evaluate the model

10
What conditions trigger pollen entrainment?

• No rain
• Sunshine
• Low humidity (below 75)?
• Moderate to high wind speeds
• Afternoon temperatures below 95oF
• Morning temperatures above 65oF
• Phenological phase

11
What are Low, Moderate, High, and Very High
Values?
    Low Moderate High Very High
NAB Percentile 0-50th 50-75th 75-99th gt99th
  NAB Concen 0-9 10-49 50-499 gt500
  Tulsa Concen 0-129 129-284 285-613 gt613
1994 AAAAI Pollen and Spore Report Percentile 0-50th 50-75th 75-90th gt90th
1994 AAAAI Pollen and Spore Report Tulsa Concen 0-129 129-284 285-410 gt411
Tulsa Model Percentile Goal 0-25th 25-50th 50-95th gt95th
  Actual Percentile 0-27th 27-47th 47-95th gt95th
  Concentration 0-49 50-99 100-489 gt490
Burge, H.A. 1992. Monitoring for Airborne
Allergens. Annals of Allergy, 69 9-18
12
(No Transcript)
13
Forecasting Model

• Phenology Factor (PF) based on day in the
  pollen season and 15 year mean concentration
  (Range 1 to 6)
• Metereological forecasts from NGM-MOS 60 hr
  forecast
• R forecast of rain (- variable amount)
• T temp outside optimum range (morning
  temperature lt 65 F or afternoon temperature gt 95
  F) (- variable amount)
• RH forecast of noon relative humidity gt75 (-1)
• W-sp wind speeds gt15 mph (1)
• W-dir wind from N - Aug 15-31 or wind from S -
  Oct 1-31 (1)
• Pre Preseason weather hot, dry July and
  August (-1)

Forecast PF R T RH W-sp W-dir Pre
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
Forecast Pollen Level NAB Pollen Categories NAB Pollen Categories NAB Pollen Categories NAB Pollen Categories
Forecast Pollen Level Low Moderate High Very High
Forecast Pollen Level Number of Days Number of Days Number of Days Number of Days
LOW 28 9 3  
LOW TO MODERATE 11 16 7  
MODERATE 2 3 7  
MODERATE TO HIGH     15  
HIGH     16 1
HIGH TO VERY HIGH     14 2
VERY HIGH     5 1
20
Ragweed Pollen Forecaster
21
(No Transcript)
22
Computer Program Evaluation

• Correct forecast 34 days (49)
• Incorrect forecast 13 days (19)
• No forecast data 23 days (32)
• For the 47 days with data 72 correct

23
Conclusions

• Empirical model accurately predicted the pollen
  level on 84 of the days during the 2002 and 2003
  ragweed seasons (74 using NAB levels)
• Computer program needs more work
• Pollen forecasts are only as accurate as the
  meteorological forecasts
• More research is needed on the
• effects of RH and rain on pollen release and
  dispersal
• influence of pre-season meteorological conditions
  on the seasonal pollen potential

24
Acknowledgment The assistance of Claudia Owens,
Shernell Surratt, and Christen Townsend in
counting pollen is greatly appreciated.