Quality control flags

In most of our data files, "flags" are used for quality control, and to indicate whether, or how, the data were processed. The following table provides an explanation for each of our two-digit flags:

First digit	Second digit
1 — Observed data
	1 — Raw data
	2 — Processed data
	3 — Possible problems in data
	8 — Chart data
2 — Use of other instruments to fill in data, or large amount of radio or other interference subtracted
	1 — Raw data, another instrument
	2 — Processed data
	3 — Corrected data
6 — Interpolated data
	9 — 95 to 99.9% of data present
	8 — 90 to 95% of data present
	7 — 85 to 90% of data present
	6 — 80 to 85% of data present
	5 — 70 to 80% of data present
	4 — 60 to 70% of data present
	3 — 50 to 60% of data present
	2 — 40 to 50% of data present
	1 — 30 to 40% of data present
	0 — less than 30% of data present
7 — Calculated data
	1 — Diffuse
	2 — Tilted
9 — Missing or bad data
	9 — Missing or bad data
0 — Chart data (obsolete)
	9 — Chart data (obsolete)

Notes

Flag 13.

Data flagged 13 is not reliable for use in developing models. Often data is flagged 13 if it might have something wrong. For example, it's not always possible to discern the precise time when an Eppley NIP goes out of alignment. If the NIP is known to be out of alignment on a particular day, then the direct normal data for the previous day may be flagged 13.

Flag 23.

Data are corrected only if we can be reasonably sure of the correction, and if the resulting values are within 5% of the actual values. For example, if snow is building up on a pyranometer during the first part of the day, and the direct beam measurements show that there were few—if—any clouds, it is sometimes possible to correct the global values. Any values that are manually changed from the original data are flagged 23.