1. Roger W. Brode U.S. EPA/OAQPS/AQAD Air Quality Modeling Group AERMET Training NESCAUM Permit Modeling Committee Annual Meeting New London, Connecticut May 31, 2007

2. Presentation Outline Brief History of AERMOD Basic Physics of Air Dispersion AERMET Scaling Parameters AERMOD Sensitivity Analysis with Example Meteorological Data Sets Meteorological Data Issues Recent AERMET Updates

3. Brief History of AERMOD Developed by AMS/EPA Regulatory Model Improvement Committee (AERMIC) Proposed as Replacement for ISCST3 April 2000 EPRI-sponsored PRIME Downwash Algorithms Incorporated in AERMOD in 2001 Notice of Data Availability (NDA) for AERMOD with PRIME Issued September 2003 Promulgated as EPA’s Preferred Model on December 9, 2005 One-year Grandfather Period Expired on December 9, 2006

4. AERMOD Design Criteria Up-to-date Science Simple – Captures Essential Physical Processes Robust – Applies Over Range of Meteorology Easily Implemented – Simple I/O, User-friendly Can Evolve – Easily Updated

5. Basic Physics of Dispersion Air dispersion is driven by two main forces – buoyancy effects and shear stress effects Buoyancy controlled by solar heating (day) and radiative cooling (night) Shear stress (friction) controlled by surface roughness elements and aerodynamic effects

6. Physics of Dispersion - Daytime Buoyancy caused by daytime solar heating generates large scale convective cells Convection causes rapid vertical spread of plumes and growth of the mixed layer Strength of convection controlled by solar angle (time-of-day and latitude), cloud cover and surface characteristics (albedo and Bowen ratio)

7. Physics of Dispersion - Daytime Albedo – Measure of reflectivity of surface, from 0 to 1 – Typical values ranges from about 0.1 for water to 0.6 or higher for full snow cover Bowen Ratio – Ratio of sensible to latent heat flux – Determines how much solar heating goes to evaporation of surface moisture – Ranges from about 0.1 (very wet) to 10 (very dry)

8. Physics of Dispersion - Nighttime Radiative cooling at night causes stable lapse rate to develop – suppresses propagation of turbulence Generation of turbulence dominated by friction-induced shear stress Shear stress or mechanical turbulence controlled by wind speed and surface roughness

9. Physics of Dispersion - Nighttime Surface Roughness Length (z o ) – Height at which wind speed goes to zero (0), based on theoretical logarithmic profile – Related to the surface roughness elements, but is not = height of elements – Ranges from about 0.001m (1mm) over water to 1.0m or higher for forests and urban areas – May vary by season and wind sector

10. AERMOD Similarity Theory Concepts Wind, temperature and turbulence are scaled with height based on Similarity Theory Mechanical (shear stress) turbulence scaled by friction velocity (u * ) Convective turbulence scaled by convective velocity scale (w * ) Monin-Obukhov length (L) stability parameter – Positive for stable conditions; negative for unstable – ~ Height at which friction and buoyant forces balance

11. Sensible Heat Flux – H If H > 0 PBL is convective; H < 0 stable Daytime – Convective B o = Bowen ratio R n = Net radiation Nighttime – Stable θ * = Temperature scale

12. Friction Velocity – u *

13. Monin-Obukhov Length – L Represents height at which mechanical (friction) and buoyant “forces” balance:

14. Monin-Obukhov Length (L) vs. PG Stability Class Roughness~L (m) PG Stability Class 0.1 m -12.5A -25B -65C --D +65E +30F Roughness~L (m) PG Stability Class 0.5 m -16A - 50B -100C --D +100E +50F

15. AERMOD CBL Treatment Turbulence in the CBL is driven by convection Convective cells grow during the day creating areas of updrafts and downdrafts AERMOD accounts for non-Gaussian vertical structure of dispersion in the CBL – Portions of plume released into updrafts vs. downdrafts are simulated separately – Full or partial penetration of plume through top of mixed layer also simulated

16. CBL Dispersion Comparisons – Crosswind Integration Concentrations ISCST3 AERMOD Tank Study

17. Convective Velocity Scale – w *

18. AERMOD Sensitivity Analysis Varied Surface Characteristics: – Albedo Test: α=0.1-0.6 (B o =1.0, z o =0.1) – Bowen Ratio Test: B o =0.1-10 (α=0.2, z o =0.1) – Surface Roughness Test: z o =0.001-1.3m (α=0.2, B o =1.0) Varied Stack Heights and Buoyancy – 5,10,15,20,25,30,50,75,100,150,200m non-buoyant – 100,150,200m very buoyant (VB)

19. AERMOD Sensitivity Analysis Two meteorology sites: – Pittsburgh – Oklahoma City Rural and Urban (population=2.4 million) 24-hr and Annual averages

20. AERMOD Sensitivity Analysis Changes in albedo and Bowen ratio affect convective turbulence – ↓ Albedo ↑ Convective turbulence – ↑ Bowen ratio ↑ Convective turbulence Changes in surface roughness affect mechanical turbulence – ↑ Surface roughness ↑ Mechanical turbulence

21. Albedo – Normalized Percent Differences

22. Bowen Ratio – Normalized Percent Differences

23. Surface Roughness – Normalized Percent Differences

24. Sensitivity Analysis Conclusions AERMOD is most sensitive to changes in z o for low-level releases – This is due to the impact of changes in z o on mechanical turbulence: ↑ Surface roughness will ↑ Mechanical turbulence Some sensitivity to changes in albedo and Bowen ratio, especially for taller, more buoyant stacks – Lower albedo or higher Bowen ratio will increase convective turbulence and bring plume down quicker

25. AERMOD Sensitivity Example Ground-level volume source based on “haul road” example Meteorological data based on two sets of surface characteristics: – With snow cover: z o =0.01m, α = 0.6, B o = 1.5 – Without snow cover: z o =0.1m, α =0.16, B o = 0.8 Snow cover characteristics based on literal interpretation of look-up tables for winter, which assume continuous snow cover – sometimes overlooked by users

26. Sensitivity Example Results – 1hr

27. Sensitivity Example Results – 24hr

28. Identification/Selection of Met Data NWS surface data formats – CD144 (original data format – stands for “card deck” 144) – SCRAM (1984-1992) – SAMSON (through 1990) – HUSWO (1990-1995) – TD-3280 (24-hour records by element – NCDC may not support) – ISHD (TD-3505) – full archival format, not abbreviated format Note that ISHD surface data is reported in GMT, not local time! NWS upper air formats – TD-6201 – FSL

29. AERMET On-site Met Data Surface (single value) measurements – Functional: sky cover net radiation Solar radiation (insolation) temperature differences between levels mixing height surface roughness length surface friction velocity and others. – Not yet functional: surface heat flux & others.

30. AERMET On-site Met Data Tower (multi-level) measurements – Functional Height,  a,  w, temp., wind speed and direction Relative humidity – Not yet functional Vertical wind component  u,  e,  v Dew point

31. AERMET Meteorological Data Issues Automated Surface Observing System (ASOS) – NWS began replacing observer-based system with ASOS in 1992 – Hourly ASOS data now available for over 900 stations within U.S. – Ceilometer cloud cover limited to 12,000 feet – Cloud cover (total only – no opaque) reported by category (CLR, SCT, BKN, OVC) – Increased incidence of calms

32. ASOS vs. Observer-based Calms

33. AERMET Meteorological Data Issues Adoption of METAR standard for reporting weather observations – Began on July 1, 1996 – Introduced variable wind code (VRB) for wind speeds up to 6 knots – Variable winds now coded as missing WD with non- missing WS – previous versions of AERMET coded variable winds as calms – Added new cloud cover code (CLR, FEW, SCT, BKN, OVC)

34. AERMET Meteorological Data Issues Surface NWS data format issues – ASOS cloud cover data reported as four categories: CLR, SCT, BKN, OVC – METAR introduced fifth category: CLR, FEW, SCT, BKN, OVC – Abbreviated ISHD data does not include “FEW” cloud cover code – combined with “SCT” cloud code – TD-3280 cloud cover codes mapped to upper bound of category in tenths – Numerous additional cloud cover codes available in TD-3280 format not supported by AERMET – no plans to include these – Full archival ISHD format appears to be the best resource

35. AERMET Meteorological Data Issues ASOS station location uncertainties – Excel file with ASOS commission dates and station locations available on NCDC website appears to be unreliable for location information – Additional data available for about 200 ASOS stations as part of tropical cyclone wind study appears to be reliable – Many station locations appear to be off by several hundred meters (median value of about 500m) – Use of erroneous station locations in AERSURFACE could invalidate results

36. ASOS Met Station Locations – Cyclone Wind Study

37. ASOS Met Station Locations

38. Recent AERMET Update: Bug Fixes Significant changes to processing of NWS Integrated Surface Hourly Data (ISHD) surface data (TD-3505) – Selection of which record to process for hours with multiple records – Processing of cloud cover codes – Initialization of “additional” character variable – caused data from previous hours to be used – Identify “variable” winds as missing WD and non-missing WS rather than calm – Corrected problem that allowed observation hour to be incremented prematurely

39. Recent AERMET Update: Bug Fixes Corrected treatment of “variable” winds for TD-3280 format – coded as missing WD with non-missing WS Corrected treatment of missing data codes for WS/WD for HUSWO format Modified upper and lower bounds for surface pressure for FSL upper air data to avoid skipping valid soundings at high or low elevations Corrected bugs in calculation of critical solar angle – affecting transition hours

40. Recent AERMET Update: Enhancements Single AERMET executable – still needs to be run in 3 stages Optional station elevation for ISHD data to substitute for missing elevation in estimating surface pressure

41. Recent AERMET Update: Miscellaneous Impose lower limit of 0.001 meters for user- specified surface roughness, for consistency with AERMOD model User-specified time window for ISHD data removed – default window of 30 minutes preceding the hour used Given range of changes affecting results, AERMOD modified to require reprocessing of met data with 06341

42. Unresolved AERMET Issues/ Planned Updates Format problem with some ISHD data may cause AERMET to crash – FIXISHD utility program – interim fix released in April Data (record-period) gap in ISHD data Inconsistencies between NWS surface data formats – TD-3280 sky cover codes – Abbreviated ISHD (TD-3505) sky cover

43. Questions