Same steady-state Gaussian plume as the 3D version, but this 2D view shows exactly what a satellite instrument actually measures: a top-down column density (mass per unit ground area), obtained by integrating the 3D concentration field vertically:
Ccol(x,y) = ∫ C(x,y,z) dz = Q / (u σy(x) √(2π)) · exp(-y²/2σy²)
Because ground reflection conserves the vertical integral exactly, this closed form needs no vertical grid at all — the stack height H cancels out of the column total. σy(x) is the Briggs rural crosswind-spread coefficient, which grows downwind and depends on the Pasquill–Gifford stability class (A = strong daytime mixing, F = calm stable night). Mass balance across any downwind line then gives:
Q = u · ∫ Ccol(x,y) dy (exact, noise-free)
This is the Integrated Mass Enhancement (IME) principle behind real satellite methane retrievals (GHGSat, Sentinel-5P TROPOMI, MethaneSAT): image the plume's column density along a downwind line, sum the enhancement, divide out the wind speed — without ever seeing the source directly.
- Sensor noise σ — Gaussian noise added to every simulated column pixel before integration, modelling instrument and retrieval noise in a real spectral inversion.
- Detection threshold — the AI classifier zeroes out any pixel below this column density, exactly as a real detection algorithm suppresses sub-threshold noise; set it too high and the plume's spreading tail — and the estimate — gets clipped.
- Detection confidence — a logistic score of the noisy peak column density against the threshold, standing in for a CNN plume-classifier's output probability.
- Resample scene — draws a fresh noise realisation, the way each satellite overpass gives an independent noisy retrieval of the same true source.
- Drag / scroll — pan and zoom the satellite view; the dashed line marks the downwind cross-section used for the mass-balance retrieval.