Table 4 Treatment methods for nulls and the relevant official documents.

1

Successive nulls for >5 days

Consider them as downtime for maintenance and omit them in emission.

a. According to the regulation⁴⁰, a power plant should go through at least one long maintenance shutdown per year, with one lasting for at least 5 days.

b. The estimated downtime (corresponding to successive nulls for at least 5 days) accounted for 17.11% of the time for 2015, which are generally consistent with the official statistics (19.41%)⁹ (considering that 3–4% of plants do not have CEMS).

2

Successive nulls for 1–5 days

Assume them around the levels of valid values near the time (in terms of monthly averages).

Chinese government published Specifications for Continuous Emissions Monitoring of Flue Gas Emitted from Stationary Sources (HJ/T 75-2007)²⁷. It suggests no interpolation for successive missing data of emission concentrations lasting for above 24 hours during operation, assuming them at similar levels to the points near the time and not to largely deviate from the average values.

3

Nulls lasting for 1–24 hour(s) (involving non-successive nulls)

Set them to the arithmetic mean of the two nearest valid points before and after them.

The guideline (HJ/T 75-2007)²⁷ suggests setting missing data lasting for 1–24 hour(s) to the arithmetic mean of the two nearest valid points before and after them.