I tried to run SMOKE for the ‘ptegu’ sector for 2028 from EPA 2016v1 platform using 2028fh1 version. I used every input files as EPA including scripts.
I successfully simulated ‘ptegu_summer’ and ‘ptegu_winter’ , and cross checked the difference of my run output with the EPA output using ‘inlineto2d’ and ‘m3diff’ tool.
The difference is negligible ( I checked 1 week of May, 1 week of Jan, 1 week of Dec 2016 ) for the Jan-Dec of 2016. The maximum difference is more than a million decimal ( such as 9.53674E-07 for Jan 1).
However, I ran a 10 day SPIN UP run ( 20151222 to 20151231) using this script Dec2015_ADEQ_Annual_ptegu_daily_winter_12US1_2028fh1_16j.csh.txt (7.5 KB) . When I compare the output , the difference is high. I checked for 20151222-20151224, 20151231 . The difference is significantly higher for all of the SPIN UP period.
For example, top three differences for 20151222 , 20151223, and 20151231 are
( 9.99237E+00,9.98373E+00,9.88843E+00), (9.97088E-01, 9.90876E+00,9.89914E-03), and (9.90237E-02, 9.75291E-03, 9.73965E-01) respectively.
Any suggestions why it is happening would be greatly appreciated.
In the EPA platform, we don’t create spinup ptegu emissions in SMOKE by running it for the last 10 days of December of the preceding year, instead we copy the emissions from the same dates for the base year. If you do run spinup ptegu in SMOKE, you will get different results because the 2016 CEMs don’t cover the spinup period at the end of 2015. You can try just copying your last 10 days of 2016 to the last 10 days of 2015 and the numbers should match EPA’s.