The lifetime of tropospheric O₃ is difficult to quantify because we model O₃ as a secondary pollutant, without direct emissions.  For other reactive greenhouse gases like CH₄ and N₂O, we readily model lifetimes and timescales that include chemical feedbacks based on direct emissions.  Here, we devise a set of artificial experiments with a chemistry-transport model where O₃ is directly emitted into the atmosphere at a quantified rate.  We create three primary emission patterns for O₃, mimicking secondary production by surface industrial pollution, that by aviation, and primary injection through stratosphere-troposphere exchange (STE).  The perturbation lifetimes for these O₃ sources includes chemical feedbacks and varies from 6 to 27 days depending on source location and season.  Previous studies derived lifetimes around 24 days estimated from the mean odd-oxygen loss frequency.  The timescales for decay of excess O₃ varies from 10–20 days in NH summer to 30–40 days in NH winter.  For each season, we identify a single O₃ chemical mode applying to all experiments.  Understanding how O₃ sources accumulate (the lifetime) and disperse (decay timescale) provides some insight into how changes in pollution emissions, climate, and stratospheric O₃ depletion over this century will alter tropospheric O₃.  This work incidentally found two distinct mistakes in how we diagnose tropospheric O₃, but not how we model it.  First, the chemical pattern of an O₃ perturbation or decay mode does not resemble our traditional view of the odd-oxygen family of species that includes NO₂.  Instead, a positive O₃ perturbation is accompanied by a decrease in NO₂.  Second, heretofore we diagnosed the importance of STE flux to tropospheric O₃ with a synthetic ‘tagged’ tracer O3S, which had full stratospheric chemistry and linear tropospheric loss based on odd-oxygen loss rates.  These O3S studies predicted that about 40 % of tropospheric O₃ was of stratospheric origin, but our lifetime and decay experiments show clearly that STE fluxes add about 8 % to tropospheric O₃, providing further evidence that tagged tracers do not work when the tracer is a major species with chemical feedbacks on its loss rates, as shown for CH₄. 

Here we present the results from the UC Irvine chemistry-transport model (UCI CTM) used in the study of direct, primary ozone emissions.  There are 16 *.nc data files that provide a 3-D (every modeled grid cell, 320x160x57) snapshot of O3 or other key variables (air mass, e90(tropopause tracer), T and, surface pressure (2D only, 320x160).  The netcdf variables are 4-D, being sampled every 5 days for 5 years (320x160x57x365), for 4 years (320x160x57x292), or for 6 months (320x160x57x36). The 17^th *.nc file gives the relative differences (RD) for the trace chemical species between the control run (CTRL) and the aviation direct-O3 emissions (EO3A) on July 1. 

The direct O3 emission case shorthand is:

• EO3A = uses aviation NOx pattern for O3 emissions
• EO3B = not direct O3 emissions, but a doubling of aviation NOx
• EO3S = surface pollution pattern (NH Asia)
• EO3T = stratosphere-troposphere exchange pattern
• EO3U = revised stratosphere-troposphere exchange pattern

The Matlab scripts that read, process, and plot the results are in the accompanying zenodo archive.  The figures in the publication as well as supplementary figures (in both *.png and *.eps format) generated by these scripts are in the xenodo supplement.  The figure files are labeled in shorthand and match the plotid's in the Matlab plotting script.