Are crop-residue fires upwind turning my city’s air toxic?

What you can answer

• Daily count of active fires and total FRP over the upwind crop belt through the Oct–Nov burning window
• Downwind pollutant load (TROPOMI NO2, CO; MODIS AOD) over the receptor city for the same days
• Source-to-receptor coupling on days when winds blow from the burning belt toward the city
• Burning-season vs baseline contrast in both fire activity and downwind columns
• Spatial map of fire clusters and the smoke plume orientation relative to prevailing winds

What you can NOT answer with these alone

• Surface PM2.5 that people actually breathe — satellites measure vertical columns / AOD, not ground concentration; you need CPCB/OpenAQ monitors and a column-to-surface model.
• Formal source apportionment (what % of the city’s pollution is stubble vs traffic vs industry) — that needs chemical transport modelling or receptor models, not correlation.
• Night-time burning completeness — overpass timing and cloud cover miss fires; FRP is a snapshot, not a daily integral.
• Health outcomes — exposure-response requires epidemiological data outside these products.

Code template

import earthaccess
import xarray as xr
import numpy as np

earthaccess.login(strategy="netrc")

# Source belt (upwind) and receptor city (downwind)
source_aoi = (74, 28, 78, 32)        # Punjab–Haryana crop belt
receptor_aoi = (76.8, 28.4, 77.4, 28.9)  # Delhi NCR (downwind)
season = ("2024-10-01", "2024-11-30")     # post-monsoon burning window

# 1. VIIRS active fires + FRP over the source belt
fires = earthaccess.search_data(short_name="VNP14IMG",
                                bounding_box=source_aoi,
                                temporal=season)
# Open granules; keep fire-mask confidence >= nominal, read FRP per pixel.
# Aggregate per day: fire_count = N detections, frp_total = sum(FRP_MW).

# 2. Downwind pollutant columns over the receptor city
#    TROPOMI NO2 / CO (Sentinel-5P) and/or MODIS AOD
no2 = earthaccess.search_data(short_name="S5P_L2__NO2___",
                              bounding_box=receptor_aoi,
                              temporal=season)
# Apply qa_value >= 0.75 mask; area-mean tropospheric NO2 column per day.

# 3. MERRA-2 low-level winds to gate transport (source -> receptor)
wind = earthaccess.search_data(short_name="M2I3NVASM",
                               bounding_box=source_aoi,
                               temporal=season)
# Compute mean U,V near surface; keep days where wind vector points
# from source_aoi toward receptor_aoi (e.g. NW'ly for Punjab -> Delhi).

# 4. On transport days: scatter frp_total vs downwind NO2/AOD;
#    compare burning-season mean vs a non-burning baseline month.

Expected output

• Time series: daily fire count and summed FRP over the source belt, Oct–Nov
• Overlaid downwind series: city-mean TROPOMI NO2 / CO and MODIS AOD
• Scatter (transport days only): summed FRP vs downwind pollutant column, with correlation noted as suggestive
• Map: VIIRS fire clusters in Punjab–Haryana with a MERRA-2 wind-vector overlay toward the receptor city
• Bar contrast: burning-season vs baseline mean for fires and for downwind columns

Caveats

• FRP ≠ surface PM2.5 — it indexes combustion energy, not breathable concentration; column ≠ surface without boundary-layer correction.
• Overpass and cloud gaps undercount fires; smoke itself and post-monsoon haze can block both fire and pollutant retrievals (use QA flags).
• Correlation, not attribution — traffic, industry, firecrackers (Diwali) and regional dust co-occur with the burning season; wind-gating reduces but does not remove confounding.
• Sensor differences: VIIRS (375 m) detects smaller/cooler fires than MODIS (1 km); don’t merge raw counts across sensors without harmonizing detection limits.
• Geolocation/saturation on intense fires can bias FRP; treat very high values cautiously.

Cross-DAAC composition

LAADS/FIRMS (VIIRS, MODIS fires + AOD) + GES DISC / Copernicus (TROPOMI, MERRA-2) — multiple providers; expect separate auth for Sentinel-5P if pulled from Copernicus rather than the GES DISC mirror.

Sources

• VIIRS VNP14IMG active fire: https://www.earthdata.nasa.gov/data/instruments/viirs
• NASA FIRMS (near-real-time fire): https://firms.modaps.eosdis.nasa.gov/
• Sentinel-5P TROPOMI: https://sentinel.esa.int/web/sentinel/missions/sentinel-5p
• MODIS aerosol (MOD04): https://ladsweb.modaps.eosdis.nasa.gov/missions-and-measurements/products/MOD04_L2
• MERRA-2 (GES DISC): https://disc.gsfc.nasa.gov/datasets?project=MERRA-2