1. ATom GRIPs project (M. Prather, Mar16) – NEW SLIDES (M. Prather Jul12)
These slides looks at the statistics of the Reactivity of the air parcels as will be measured by ATom. For the Representativeness, see earlier notes and later examples. For now, reactivity is measured as P-O3 (using only the ROO+NO & HO2+NO rates) and L-O3 (using only the three loss rates: O(1D)+H2O and O3+HOx) and L-CH4 (tbd exactly how we best standardize this – currently UCI is doing troposphere only, and using the OH loss freq with a constant 1800 ppb CH4).
Results are taken from an ATom slice at 180W of all the air parcels in the UCI CTM run at T319L60. The initial conditions are taken from 0000GMT Jan (exact date and year do not matter) and then integrated for 24 hours without transport or emissions (discussed earlier). Our results are in absolute units (e.g., kg/day or kg) and then scaled to correspond to a 1.0-degree wide slice. Statistics here are for latitude-pressure ‘Boxes’, e.g., 60S to 60N and 0-12 km(approx). The clouds, temperatures and water vapor change as per our met fields for the following 24 hours.
For an ATom slice with Box = 60S-60N x 0-12km, we have about 27 x 212 = 5724 air parcels. These are now sorted by NOx (or H2O2 or...) abundance in ppt and put into 10 bins per decade (i.e., 30 bins from 1 to 1000 ppt covers all NOx in our simulation.)
We then recalculate the Reactivity in response to 10% or absolute ppt increase in NOx and H2O2 (with additional 24-hour simulations). The 10% tells us if additions have a linear impact back to zero, and the absolute increases (+10 ppt) give us an idea of how addtional pollution may impact the integrated reactivity.

2. pair of CO & NOx the scale is normalized such that a uniform
From the models we want a single, core data set for each model for the ~16 Aug simulations with floating point values in 4-D (see list of quantities below):
(I=longitude grid, J=latitude grid, K = model layer from 0 to 13 km or so, N = species index below)
This allows us to sub-sample and test for a range of issues across the models.  This is obviously set for netcdf.  For our high resolution model (~ 0.5 deg) we have about 160 x106 quantities, can cut a bit by 60S-60N limits (probably need some latitude range).  I would suggest all longitudes at this time. This is a manageable/transferable data set for us to start comparisons.  The attached plots are typical     1) we digitize all mole fraction to 20 per decade from 0.1 ppt to 1 ppm (index M=1,140)             values below 0.1 are given M=1 and above ppt are given M=140     2) we plot tracer-tracer correlations (CO_NOx in attached plots) as density plots
where the relative density in each box = amount of air mass (moles) with that
pair of CO & NOx the scale is normalized such that a uniform
1 decade by 1 decade would have values = 1     3) we can pick different pairs of interest     4) we also plot the total P-O3, L-O3, L-CH4 in a parallel density plot
(ppb/day x air mass) to show where the activity lies.     5) we need to come up with a log-scale color bar I think.

3. N = Index for gird-box Quantities:. 1 Air in each cell (moles)
N = Index for gird-box Quantities: 1    Air in each cell (moles) 2   Strat-trop tracer (e90, O3, or PV or diag to split strat-trop cells, also give the edge value) Tracer Species: (all in mole fraction) 3    O3           4    NOx = NO+NO2 5   HONO2        6   HO2NO2 7   C2H3NO5 = PAN 8   RONO2   (if any)      9   H2O2      10   CH3OOH       11   HCHO        12  C2H5OOH     13  CH3CHO  = acetaldehyde 14  CH3CO3   = peroxyacetyl radical (?important)    15  C3H6O = acetone 16  CO 17   C2H6       18   Alkane (>C2)  19  Alkene       20  C5H8 = Isoprene  Derived Products (mole fraction / day) 21 L-CH4 (OH+CH4, but scale to 1800 ppb CH4, so get OH*k) 22  P-O3 (HO2/RO2 + NO only) 23 L-O3 (O1d+H2O, OH+O3, HO2+O3 only) 24 ?? dO3/dt (actual daily change in O3, a checksum for missing terms in P & L)

4. 60S-60N x 0-12km x 1°(@180W) Jan16 sorted by NOx
NOx (and all other species) are sorted by
abundance with 10 bins per decade.
In case, all of the air mass (Peta-moles) in the 1° wide tomographic slice has NOx between 2 and 400 ppt. Most is ppt range, but there is a second peak at 4 ppt.
NB – the low-NOx air mass is probably tropical lower trop, and this could be readily pulled out if we take a different ‘box’, like 24S-24N x surf-800hPa.
Another look plots the total amount NOx in the slice (Kilo-moles) and of course most of it lies at the higher abundances, peaking about ppt.

5. 60S-60N x 0-12km x 1°(@180W) Jan16 sorted by NOx
The 3 Reactivities (Mega-moles/day) are then plotted as a function of NOx.
Note that LCH4 is multiplied by 2 so that its similarity with LO3 is more apparent.
PO3 peaks with the high-NOx air masses. The low-NOx (4 ppt) is important for LCH4 and LO3.
should read:
LCH4 x 2 (Mmol/d)

6. 60S-60N x 0-12km x 1°(@180W) Jan16 sorted by NOx
Test for NOx and H2O2 linearity:
Recalculate reactivity with 1.1*NOx and 1.1*H2O2. Scale these up assuming linearity.
PO3(lin-NOx) =
[PO3(1.1*NOx) – PO3(std)] x 10
i.e., how well does dPO3/dNOx ~ PO3/NOx ?
NOx almost linearly drives PO3, has no impact on LO3, but some impact on LCH4

7. 60S-60N x 0-12km x 1°(@180W) Jan16 sorted by NOx
Test for NOx and H2O2 absolute errors:
Recalculate with +10 ppt NOx in all air parcels and separately with +100 ppt H2O2.
dPO3(+10 ppt-NOx) =
PO3(+10 ppt-NOx) – PO3(std)
Which air parcels are most sensitive to additional NOx or H2O2 or ???. Also allows for estimate of the source of model-model differences.

8. 24S-24N x 2-8km x 1°(@180W) Jan16 sorted by H2O2
H2O2 sorted in smaller domain:
mid-trop in tropics.
mode peak mass = 550 ppt, max = 2 ppb

9. 24S-24N x 2-8km x 1°(@180W) Jan16 sorted by H2O2
H2O2 sorted in smaller domain:
mid-trop in tropics.

10. 24S-24N x 2-8km x 1°(@180W) Jan16 sorted by H2O2
H2O2 sorted in smaller domain:
mid-trop in tropics.

11. Reactivity summary tables for our 1°-wide Box:
60S-60N x 0-12km x Jan sorted by NOx
std
lin-NOx
lin-H2O2
+10 ppt NOx
+20ppt / 2
+100ppt H2O2
+200ppt / 2
air (Pmol)
336.3
NOx (Mmol)
13.7
PO3 (Mmol/d)
464.8
276.4
21.9
103.9
100.1
9.8
9.2
LO3 (Mmol/d)
575.7
25.8
17.3
11.9
11.2
4.5
4.4
LCH4 (Mmol/d)
254.2
36.3
14.2
15.4
15.1
3.2
3.1
24S-24N x 2-8km x Jan sorted by H2O2
std
lin-NOx
lin-H2O2
+10 ppt NOx
+20ppt / 2
+100ppt H2O2
+200ppt / 2
air (Pmol)
90.1
H2O2 (Mmol)
58.9
PO3 (Mmol/d)
175.5
7.4
1.7
LO3 (Mmol/d)
228.2
7.0
1.3
LCH4 (Mmol/d)
94.6
5.3
1.9

12. Joint PDFs for global (60-60, 0-12km) vs tropical Pacific (20-20, 0-12km)
Example of NOx x CO, plotting relative amount of total in log-log pixels
½ - 12

13. Joint PDFs for global (60-60, 0-12km) vs tropical Pacific (20-20, 0-12km)
Example of NOx x CO, plotting relative amount of total in log-log pixels
overflow, >1,000 ppb
½ - 12

14. Joint PDFs for global (60-60, 0-12km) vs tropical Pacific (20-20, 0-12km)
Example of NOx x CO, plotting relative amount of total in log-log pixels
½ - 12

15. Joint PDFs for global (60-60, 0-12km) vs tropical Pacific (20-20, 0-12km)
Example of NOx x CO, plotting relative amount of total in log-log pixels
½ - 12
0.78 ppb/day
1.01 ppb/day

16. Joint PDFs for L-O3 vs L-CH4
Example of NOx x CO, plotting relative amount of total in log-log pixels
½ - 12

17. Joint PDFs for L-O3 vs L-CH4
Example of NOx x CO, plotting relative amount of total in log-log pixels
½ - 12
1.01 ppb/day

18. One critical ATom measurement is the production of O3 (P-O3) in the 30-sec air parcels. Below is a model sampling of the ATom Pacific flights in January.
We select a NOx measurement threshold of 14 ppt since 90% of the O3 is produced in air parcels with NOx > 14 ppt.
> 14 ppt = 90% of P-O3
Note that in terms of total air mass, about 28% has NOx < 14 ppt, but the reactivity in these parcels is not sensitive to the NOx levels, and hence a measured upper limit is adequate.