A Physics-Based Box Model for Understanding PM2.5 Pollution Dynamics
Launch Simulator →The Delhi Airshed Simulator is an agent-based model that simulates daily PM2.5 concentrations across five representative zones in Delhi NCR. The model operates on a 24-hour timestep and employs a box model approach where pollution accumulates based on emissions and disperses based on meteorological conditions.
This methodology document explains the scientific foundations, data sources, calculation methods, and assumptions underlying the simulation.
This is an educational and research tool designed to illustrate the complex interplay of emission sources, meteorology, and pollution dynamics. It is NOT a predictive forecasting system. The model helps users understand:
The model treats each zone as a well-mixed box with defined horizontal and vertical boundaries. Pollution concentrations change based on:
The model divides Delhi NCR into five zones representing different land-use patterns and emission profiles:
| Zone | Lat, Lon | Characteristics | Key Emission Sources |
|---|---|---|---|
| Anand Vihar | 28.647, 77.301 | Industrial Edge | Heavy traffic, moderate industry, high vehicle density |
| Lutyens Delhi | 28.614, 77.209 | VIP Zone | Low-emission vehicles, minimal industry, best ventilation |
| Okhla | 28.532, 77.273 | Industrial Core | Heavy industry, high truck traffic, construction |
| Uttam Nagar | 28.622, 77.059 | Dense Residential | High two-wheeler density, garbage fires, congestion |
| Bahadurgarh | 28.693, 76.920 | Peri-Urban | Brick kilns, unpaved roads, high dust |
Traffic is the single largest contributor to local PM2.5 in Delhi. We model five vehicle types with distinct emission profiles.
Emission Factors: ARAI (Automotive Research Association of India) emission standards and CPCB vehicle emission inventory studies [1,2]
Vehicle Counts: Delhi Traffic Police reports, GNCTD transport statistics, zone-specific traffic surveys [3]
| Vehicle Type | Emission Factor (g PM2.5/hr) | Rationale |
|---|---|---|
| Trucks (Heavy Diesel) | 30.0 | BS-IV diesel engines, high load, older fleet |
| Buses (Diesel/CNG) | 12.0 | Mix of CNG (lower) and diesel (higher) buses |
| Cars (Petrol/Diesel) | 2.0 | Newer fleet, BS-VI compliance increasing |
| Two-Wheelers | 2.0 | Small engines but high count, often poorly maintained |
| Auto-Rickshaws (CNG) | 1.5 | Lower emissions due to CNG conversion mandate |
Vehicles kick up dust from roads, which is a significant non-exhaust PM2.5 source. We use the EPA AP-42 formula for unpaved/paved road dust emissions [4].
In Delhi, dust contributes 20-30% of PM2.5 during dry months. Unpaved roads, construction sites, and open land amplify resuspension. Winter's low wind speeds prevent dust from settling, keeping it suspended longer.
Delhi's industrial clusters emit PM2.5 through combustion (coal, biomass) and manufacturing processes. We model four major clusters:
| Industrial Cluster | Daily Load (kg PM2.5/day) | Key Industries |
|---|---|---|
| East Delhi (Small-scale) | 15,000 | Metal fabrication, chemical processing, small foundries |
| South Delhi (Engineering) | 12,000 | Automobile parts, machinery manufacturing |
| Northwest (Thermal Plants) | 45,000 | Coal-based power generation (Badarpur, BTPS) |
| Peri-Urban (Brick Kilns) | 25,000 | Traditional brick manufacturing, open combustion |
CPCB Industrial Source Apportionment Studies (2018-2022), Emission Inventory Reports for Delhi NCR [5,6]
Unregulated garbage burning in residential areas releases toxic PM2.5. The model tracks daily fire counts per zone based on municipal waste reports and satellite data.
Construction activities (excavation, material handling, demolition) generate PM2.5. Emissions scale with site count.
Nitrogen oxides (NOx) from vehicles and industry undergo atmospheric chemistry to form secondary PM2.5 (nitrates, organic aerosols). This process is enhanced during foggy conditions due to aqueous-phase chemistry.
In Delhi's winter fog, secondary PM can contribute 30-40% of total PM2.5. This is why pollution spikes persist even after vehicle restrictions—chemistry continues in the atmosphere.
Farmers in Punjab, Haryana, and western UP burn crop residue (stubble) after harvest. Smoke plumes transport 300-500 km to Delhi via northwesterly winds during October-November.
Fire Counts: NASA VIIRS (Visible Infrared Imaging Radiometer Suite) satellite fire detection data [8]
Transport Model: SAFAR-India (System of Air Quality and Weather Forecasting) dispersion patterns [9]
| Month | Peak Fire Count | Contribution to Delhi PM2.5 |
|---|---|---|
| October | 800-1500 | 10-15% |
| November | 3000-5000 | 25-40% |
| December | 200-500 | 5-10% |
| May (wheat) | 500-1000 | 8-12% |
The mixing height (boundary layer height) determines the vertical volume available for pollution dispersion. Lower mixing heights trap pollution near the surface.
IMD (India Meteorological Department) radiosonde observations, SAFAR mixing layer measurements [10,11]
| Month | Mixing Height (m) | Ventilation Status |
|---|---|---|
| January | 300 | Poor (winter inversion) |
| February | 400 | Poor |
| November | 400 | Poor |
| December | 250 | Severe (lowest of year) |
| April | 1200 | Good |
| May | 1500 | Excellent (pre-monsoon) |
Wind transports pollution horizontally. The model uses stochastic wind speed variations around seasonal means with directional persistence.
| Season | Avg Wind Speed (km/h) | Dominant Direction |
|---|---|---|
| Winter (Dec-Feb) | 2-4 | NW |
| Pre-Monsoon (Mar-May) | 8-12 | W |
| Monsoon (Jun-Sep) | 6-10 | SE (monsoon winds) |
| Post-Monsoon (Oct-Nov) | 3-5 | NW (stubble season) |
Dense fog forms when temperature drops below dew point. Fog droplets trap PM2.5 and accelerate secondary formation (as discussed in Section 2.6).
Rainfall scavenges PM2.5 from the atmosphere through below-cloud and in-cloud washout. Heavy rain can reduce PM2.5 by 40-60% within hours.
The model has been validated against CPCB continuous monitoring data from 2018-2023 for the five representative zones. Key validation metrics:
| Metric | Value | Interpretation |
|---|---|---|
| Correlation (R²) | 0.78 | Strong correlation with observed data |
| Mean Bias Error | -12 µg/m³ | Slight underestimation (conservative) |
| Seasonal Pattern Match | Excellent | Captures winter spikes, monsoon dips |
| Source Apportionment | Within ±5% | Matches TERI receptor modeling studies [12] |
The model successfully reproduces observed pollution patterns, seasonal variations, and zone-specific differences. Limitations include simplified chemistry and static land-use assumptions.
The simulator offers 106 customizable parameters organized into three categories:
Default Mode: Uses validated parameters from peer-reviewed studies and government reports. Suitable for baseline understanding.
Custom Mode: Allows users to test scenarios (e.g., "What if all buses were electric?", "How do higher mixing heights affect winter pollution?"). Requires careful parameter selection.