Pune City 

1. AERMOD Model Case Study Mohit C. Dalvi Computational Atmospheric Sciences Team Centre for Development of Advanced Computing (C-DAC) Pune University Campus, Pune Pune City 

2. Overview • About C-DAC • Air Pollution overview • Air Quality Management Components • Air Quality Modeling overview • AERMOD Model • Case study using Linux AERMOD • Use of AQ Model for scenario analysis

3. About C-DAC High Performance Computing Hardware solutions GIS Solutions Scientific Computing Advanced Computing Training Artificial Intelligent Language Technology Medical Informatics Evolutionary Computing

4. Computational Atmospheric Sciences Activities • Computational Research • Workflow Environment Development • Technology Development • Parallel Programming • Model Porting, Optimisation & Simulations • Grid Computing • Joint Collaborative Research • Turnkey solutions • Contract Projects • Consultancy

5. UKMO: PUM Model Output (JJAS 2005) Average Daily Precipitation (mm/day) Computational Atmospheric Sciences • Global Forecast Models • NCEP's T170/T254/T382/PUM • Multi-institutional ERMP program • Regional Weather Research • MM5 / WRF / MM5 Climate / RegCM / RSM • Real Time Weather System (RTWS) • Coupled system development (IITM Collaboration) • Climate Models • CCSM • Climate Change Studies • Ocean Models • MOM4 / POM / ROMS / HYCOM • Coupled system development (IITM collaboration) • Ocean response studies • Air Quality/Environmental Computing • GIS based emissions modeling with IITM • Offline WRFChem with NOAA/FSL • WRF+AERMOD for Pune AQM with USEPA • Aerosol studies using LMDzT – Off-line version with IIT-B

6. Air Pollution Air quality-------- degree of purity of the air to which people and natural resources are exposed at any given moment. Definitions : Air (Prevention & Control of Pollution) Act, 1981 “Air pollutant" means any solid, liquid or gaseous substance 2[(including noise)] present in the atmosphere in such concentration as may be or tend to be injurious to human beings or other living creatures or plants or property or environment; “Air pollution" means the presence in the atmosphere of any air pollutant Primary air pollutants = chemicals that enter directly into the atmosphere. E.g carbon oxides, nitrogen oxides, sulfur oxides, particulate matter, hydrocarbons Secondary air pollutants = chemicals that form from other already present in the atmosphere. E.g ozone, sulfurous acid, PAN

7. Air Pollution Pollutants- Sources & Effects

8. Air Pollution Particulate Matter

9. Air Pollution Global Warming Pune City 

10. Air Pollution ATMOSPHERIC CHEMISTRY • Interactions of Pollutants • Primary Pollutant + Prim. Pollutant  Sec Pollutant • Prim. Pollutant + Existing component  Sec Pollutant • Primary/ Secondary Pollutant  Decay/ Removal - Photolysis - Dry Deposition (on soil, vegetation) - Wet Deposition (washout by rain, on fog, cloud droplet) - Radioactive decay - Absorption/ uptake by plants/ animals - Dissolution in water body/ ocean

11. Air Pollution Air Pollution Legislations– Brief History • Some reference in Factories Act, 1860s/ 1948 • 1952 – London smog – Inversion conditions for 4 days – smoke from coal (fireplaces, boilers) stagnated - ~4000 deaths • Clean Air Act (UK) – 1956 & 1968 • Clean Air Act (USA) – 1970 • Air (Prevention & Control of Pollution) Act, 1981 • Bhopal Gas Tragedy, 1984 • Environmental Protection Act, 1986

12. Air Pollution National Ambient Air Quality Standards ** 24 hourly values should be met 98% of the time in a year. However, 2% of the time it may exceed but not on two consecutive days. * Annual average = annual arithmetic mean of minimum 104 measurements in a year taken twice a week 24 hourly at uniform interval

13. Air Quality Monitoring Source Apportionment Emission Inventory Meteorological Data Strategies, Planning, Development Air Pollution Modeling GIS based Emission gridding Impacts Assessment Air Quality Management Air Quality Management - Components

14. Air Quality Management Air Quality monitoring methods Passive Methods: • Remote Sensing – Satellite Imageries – cloud/ haze • Satellite mapping (TOMS – NASA for Aerosol & Ozone) • LIDAR – Light Detection & Ranging

15. Air Quality Management Emission Inventory “Is a comprehensive listing of the sources of air pollution and an estimate of their emissions within a specific geographic area for a specific time interval.” Inventories can be used to: • Identify sources of pollution • Identify pollutants of concern • Amount, distribution, trends • Identify and track control strategies • Input to air quality modeling

16. Pollutant from 1 car of type A (gm/km or gm/lt fuel) x Avg distance travelled (or lts of fuel consumed) X No of cars of type A in given area = Total emissions from car type A in given region/ street Air Quality Management Emission Inventory Steps: - Identify sources of pollution - Measure/ estimate pollutant release from single unit - Extrapolate to expected no. of sources of same type

17. Inversion layer wind Deposition, washout turbulence Buoyancy Air Quality Management Meteorological Data • Main driver for movement of pollutants (and interactions)

18. Air Quality Management Meteorological Data Parameters of importance : • Wind components – driving force for advection. • Temperature, Surface Heat, lapse rate – for buoyancy, plume rise, stability, vertical transport • Rainfall, humidity – removal by wet deposition • Cloud cover – wet deposition, light intensity (for photochemistry), radiation balance • Landuse, albedo – for biogenic/ geogenic emissions, chemistry, dry deposition • Terrain – impact on wind, obstacle to movement • Source : Weather stations, balloons, SODAR, satellites For forecasting/ projections – numerical weather prediction models

19. Air Quality Modeling TYPES OF AIR QUALITY MODELS Physical Models – Laboratory representations of real life phenomenon Mathematical Models – Set of analytical/ numerical algorithms representing physical and chemical aspects of the behaviour of pollutant in atmosphere. Can be broadly divided into : - Statistical Model– Semiempirical statistical relations among available data & measurements - Determinisitic Models - Fundamental mathematical descriptions of atmospheric processes. Include the analytical and numerical models.

20. Air Quality Modeling PHYSICAL MODELS • Scaled Down version of real phenomenon • Attempt to replicate phenomenon under controlled conditions • E.g Wind Tunnel, Fluid Tanks

21. Air Quality Modeling STATISTICAL MODELS Statistical models are based on the time series (or any other trend) analysis of meteorological, emission and air quality data. These models are useful for real time analysis and short term forecasting. Eg. Air Quality Monitoring and Modeling for Coimbatore City - P.Meenakshi and R.Elangovan (CIT) Use of "least squares" method to analyse how a single dependent variable is affected by the values of one or more independent variables. - The monitored data in Coimbatore City are analyzed by multi regression : SPM= -82.0703 T - 80.5704 P - 0.76381 WD - 2.03456 WV + 64531.68; R = 0.5 SO2= 2.397 T - 1.1481 P + 0.016 WD + 1.173 WV + 831.5413; R = 0.2 NOx= 5.728 T + 3.2582 P - 0.0636 WD + 2.1923 WD + 2.192 WV - 2601.85; R=0.36 Where, T- temperature, P - pressure, WD - wind direction and WV - wind velocity.

22. Air Quality Modeling RECEPTOR MODELS Receptor Models use the chemical & physical characteristics of measured concentrations of pollutants at source as well as receptor to identify the presence and contribution of the source to the pollutant level at receptor. e.g Chemical Mass Balance Equation Ci = Fi1S1 + Fi2 S2 + …. FiJ SJ Ci : Concentration of ith species Fij : Fraction of species i from source j Sj : Sources contribution from sources 1 – J = Dj * Ej Ej = Emission rate Dj = 0 T d [u(t),s(t),x] dt u = wind velocity s =stability parameter x = distance of source from receptor

23. Air Quality Modeling DETERMINISTIC MODELS • Calculate/ predict the concentration field based on mathematical manipulations of the inputs : - source & emission characteristics - atmospheric processes affecting transport - chemical processes affecting mass balance • Eg - Diffusion models – Gaussian models - Numerical models : - Eulerian Models - Lagrangian models

24. Air Quality Modeling Gaussian Plume Model

25. Air Quality Modeling Gaussian Plume Model

26. Air Quality Modeling Gaussian Plume Model - Assumptions

27. Air Quality Modeling Gaussian Plume Model • Simplified form c = concentration (x,y,z,H) , Q emission rate (g/s) , u-wind speed@z y – standard deviation of conc. in y direction (stability dependant) z -standard deviation of conc. in z direction Standard deviations determined by using Briggs/ Pasquill-Gifford formaulas as a function of x (downwind distance) and stability class

28. Air Quality Modeling PLUME RISE • Initial vertical dispersion of the plume emitted from stack due to momentum (exhaust velocity) and buoyancy (higher temperature than surroundings. Briggs Buoyancy Flux parameter : Fb Fb = v2*r2*g*(Ts-Ta)/Ts v = velocity at exit, r = radius Ta = air temp, Ts = stack temp Distance to final plume rise xf = 49(Fb)5/8 for Fb >= 55 119(Fb)2/5 for Fb < 55 Plume rise – unstable/ neutral conditions : ▲h = (1.6 * (Fb)1/3 * (xf)2/3)/u Plume rise – stable conditions : ▲h = 2.4*( (Fb / us)1/3 ) s = stability parameter (g/Ta) (/z) Effective stack height : Ht = hs + ▲h

29. Air Quality Modeling EULERIAN MODELS • Based on conservation of mass of a given pollutant species (r,t) • Modeling Domain is a fixed 3-Dimensional grid of cells • Atmospheric parameters are homogenous for a given cell at t • Computations for each cell at each timestep u,v wind velocity in x, y direction Kxy, Kz Horizontal, vertical diffusion coeff. Vd = dep velocity, Δz =plume ht W=washout coeff., I=prep. Intensity,H=layer ht Pc=Product matrix, Rc=Reactant matrix Soln : Finite differences, FiniteElement, Parabolic – req initial & boundary conditions

30. Air Quality Modeling LAGRANGIAN MODELS • Lagrangian approach derived from fluid mechanics – simulate fluid elements following instantaneous flow • Frame of reference follows the air mass/ particle • Advection not computed separately as against Eulerian <c(r,t)> = -α t  p(r,t|r’,t’) S(r’,t’) dr’ dt’ c(r,t) = conc. At locus r at time t S(r’,t’) source term (g/m3s) p = probability density function that parcel moves from r’,t’ to r,t (for any r’ & t>t’ p<=1) (solved statistically e.g Monte Carlo) Chemistry/ dry/wet removal handled by change in mass at each step: m (t+Δt) = m (t) exp(-Δt/R) , R: rate of reaction/dry/wet deposition • Preferred method for particle tracking • Puff simulation by simulation at centre of mass of puff

31. Air Quality Modeling Comparison – Eulerian, Lagrangian frames Eulerian approach Lagrangian approach z t t1 t t1 y x Combined models: Eulerian models where individual puff/particle are handled by Langragian module till it attains grid dimensions

32. Air Quality Modeling AERMOD (AERMIC MODEL) Developed by AMS/ EPA Regulatory Model Improvement Committee - 1994 – 2001 till first version - Steady-state Gaussian Plume Dispersion Model • Improvements over traditional Gaussian Models (ISC) - Computes turbulence before dispersion - Separate schemes for Convective & Stable BL - Inbuilt computation of vertical profiles (PDF) - Urban handling- nighttime boundary layer - Specified as Preferred Regulatory Model by USEPA in 2006 Pune City 

33. AERMAP TERRAIN PREPROCESSOR AERMET METEOROLOGICAL PREPROCESSOR AERMOD MAIN MODEL Air Quality Modeling AERMOD Modeling System Receptors DEM Data Surface Obs. Upper Air Data Site Met. Data Concentration Profiles Average, Exceedance, Source Contributions Sources & Emissions Point, Area, Volume Version 02222

34. AERMOD Modeling System AERMET – Meteorological Preprocessor • Extract, Quality check & Preprocess- Raw Met. data • Inputs : Surface Observation Parameters (Hourly)– - Minimum :Ambient Temperature,Wind direction & speed, sky cover - File formats : NWS, CD-144, TD-3280, Samson Upper Air Data • - Supports NWS (twice daily) UA soundings, NOAA-FSL data • - Parameters (Levelwise): Atmospheric Pressure, Height, Temperature (dry bulb),Wind direction, Wind speed Onsite Meteorological Records • - Optional – User specified format - Output • 1. Surface File with PBL parameters • 2. Profile file with levelwise data

35. AERMOD Modeling System AERMOD Model • Inputs : Outputs from AERMAP & AERMET • Source & Emission Information: - Point sources: - Locations, Emission Rate, Stack parameters. Building dimensions - Area Sources : - Location & dimensions, Emission rate - Volume Sources: Location, ‘initial’ dimensions, Emission Rate • Urban Source Option – Population [and Surface Roughness]

36. Pune - Air Quality Modeling WRF-AERMOD coupling for Pune Air Quality Modeling (MOEF-USEPA Program for Urban Air Quality Management) • C-DAC role: Emission inventory, data processing, air quality modeling • Hourly meteorology req. for AERMOD air quality model • First time in the world Development of Preprocessor for coupling WRF and AERMOD • Stakeholders: PMC, NEERI, MPCB, C-DAC ,. . .

37. Case Study Pune City  • Rural Area – One processing plant, two clusters of households

38. Case Study Emission Inventory Industry: Manufacturing plant using coal. Requires 10 tonnes coal/ day with ash 36%. Pollution control equipment : scrubber with 90% efficiency Particulate matter emissions: 10 tonnes/day coal x 0.36 tonnes/ton ash x 0.8 (percent flyash) = 2.88 tonnes/day fly ash Scrub : 2.88 x (100-90)/100 = 0.288 tn/day (0.288 tn/day x 1,000,000 gm/tn )/ 86400 sec/day = 3.33 gm/sec Stack details : ht = 25 m , top dia = 0.5 m, exit velocity = 5 m/s, exit temp = 453. 0K Pune City 

39. Case Study Emission Inventory Household cooking: Stoves using firewood and kerosene in 65:35 usage ratio. Consumption : firewood - 175 kg/p/yr; kerosene – 56 kg/p/yr (PMC) Emission factors : firewood – 1.7 g/kg ; kerosene = 0.6 g/kg (URBAIR) Population – cluster1 – 500. cluster2 – 245. Area : cluster1 – 800 sq.m ; cluster2 – 550 sq.m Amount of firewood : Cluster1 : 500 persons x 0.65 x 175 = 56875 kg/yr = 155 kg/day Cluster2 : 245 persons x 0.65 x 175 = 27878 kg/yr = 76.3 kg/day Kerosene : Cluster1 : 500 persons x 0.35 x 56 = 9800 kg/yr = 26.84 kg/day Cluster2 : 245 persons x 0.35 x 56 = 4802 kg/yr = 13.15 kg/day Emissions: Cluster1 : (155 x 1.7) + (26.84 x 0.6) = 279.6 g/day = 0.0032 gm/sec / 800 = 4.0E-6 g/sec-m2 Cluster2 : (76.3 x 1.7) + (13.15 x 0.6) = 137.1 g/day = 0.0016 gm/sec / 550 = 2.91E-6 g/sec-m2 Pune City 

40. AERMOD Modeling System GUI for AERMOD model on Linux Platform • AERMOD designated by USEPA as replacement for ISC model. • AERMOD set-up (sources, receptors, options) cumbersome • Linux based graphical user interface for ease of use • Features: • Drawing tools to specify the source/ receptors • Simplified forms to specify options. • Online validation of parameters • Automatic generation of the input file. • Actual AERMOD runs through the GUI • Post-processing for contour plots Pune City 

41. Case Study Demo

42. AERMOD Modeling System Pune Air Quality Modeling – Scenario Analysis • Feasibility of using Pune AQM system for Control Scenarios • Simplifying the process : Inventory Model input • Scenarios – Planned Development/ Controls (PMC) Probable/ Likely situations/ measures • Sourcewise controls and emissions impacts • Projected – 2010, 2015 • Currently – Relative impacts on contribution from specified sources

43. Average Contribution of Sources to PM10 over Pune – Base Case run AERMOD Scenario Analysis Pune Air Quality Modeling – Scenario Analysis • Base Case Run : 2006-07

44. AERMOD Scenario Analysis Pune Air Quality Modeling – Scenario Analysis • Vehicular Sources – BAU 2010/ 2015 • Increase in Vehicle population as per RTO/ PMC- AQM Cell survey • Results

45. AERMOD Scenario Analysis Pune Air Quality Modeling – Scenario Analysis • Vehicular Sources – CNG 2010/ 2015 • 3-Wheelers – 40% conversion by 2010; 100% by 2015 • Passenger Cars – 5% by 2010, 10% by 2015 • Results

46. AERMOD Scenario Analysis Pune Air Quality Modeling – Scenario Analysis • Vehicular Sources – PMT 2010/ 2015 • Improvement in PMT bus service – increased no/ frequency:- Expected to benefit about 20000 passengers daily Reduction in personal vehicle trips by these passengers • Results

47. AERMOD Scenario Analysis Pune Air Quality Modeling – Scenario Analysis • Vehicular Sources – Bus Shifting 2007-08 • Shifting of Interstate Bus stations to outskirts Reduction in Heavy vehicle traffic (~ 2000 state, 120 private) thru city Increase in personal (2/4W) and public (3/W) trips to new Bus stands Current / Immediate future only Results

48. AERMOD Scenario Analysis Pune Air Quality Modeling – Scenario Analysis • Slum Fuel Use – SLUM 2010/ 2015 • Traditionally : biofuels  kerosene  LPG • As per AQM Cell survey, faster shift from biofuel to LPG Expected ratio 50% LPG; 35% kerosene; 15% biofuel - Increase in slum population – 6% / yr (AQM Cell) Results

49. AERMOD Scenario Analysis Pune Air Quality Modeling – Scenario Analysis • Combined Scenarion – CNG + Slum Fuel Use – SLMCNG 2010/ 2015 • Most likely scenarios • Contribution from Vehicular + Slum fuel use Results

50. AERMOD Scenario Analysis Pune Air Quality Modeling – Scenario Analysis • Scenarios At A Glance