 |
|
Quality Assessment with QC_TND |
|
QC_TND performs quality-control (QC) checks on the major broadband components:
- Global or Total (T),
- Direct Normal (N),
- and Diffuse (D).
|
|
No QC will be performed if the solar zenith angle exceeds 75 degrees.
The QC_TND flags are explained in Table 1 below, while
Table 2 outlines the process by which
these flags are determined.
|
|
Table 1. QC_TND Flag Values
| Flag | Flag Value |
| 0 |
untested parameter |
1
| parameter took and passed tests for physical limits, model
limits for tolerances less than 3%,
and reasonable coupling with other parameters for tolerances less
than 3%. |
| 2
| parameter inspected by hand and eye; passed inspection (superseded by 1,
if 1 is true). |
| 3
| parameter inspected by hand and eye; did not pass inspection (supersedes 1). |
| 4
| parameter was received as missing; was interpolated from neighbors. |
5
| parameter was received as missing; was able to be modeled by coupling
with the other 2
parameters. The modeled parameter passed all tests for tolerances
less than 3%. |
6
| parameter was received as missing; was patched by a method
other than modeling
or interpolation. |
| 7
| parameter lower than 0. |
| 8
| parameter higher than ETR. |
| 10-97 |
|
parameter exceeded the 3% tolerance in one of 4 ways:
|
|
0 - too low when coupled with other parameters
1 - too high when coupled with other parameters
2 - too low when compared to a model
3 - too high when compared to a model
|
| The particular error is determined by the remainder of
MOD (Flag + 2, 4). The percentage error is determined by dividing (Flag + 2) by 4. The
error is greater than or equal to the integer result, but less than the next highest, except
for the 94-97 flags, which signify errors greater than or equal to 24%, no ceiling.
EXAMPLE:
|
|
Flag = 67 becomes 69 when 2 is added.
MOD ( 69, 4 ) = 1, or a coupled high error was detected first.
Percent = 69/4 = 17; so the mismatch was:
the mismatch is no less than 17% and less than 18%.
|
| Please note that the highest percentage problem is reported.
For radiation values less than 210 W/m2, absolute tolerances
are used. In the example above, if the radiation value were 150 W/m2, the
error would be between 17% and 18% of 210 W/m2, or about 36 W/m2.
|
|
|
| 99
| parameter is missing. |
Table 2. QC_TND Tests
| Group | Test Performed |
| I |
| |
Parameter within physical limits.
|
|
For each parameter:
- Is it less than 0?
- Is it greater than ETR?
- Is it greater than the model maximum?
If the parameter fails ("yes" to any question) any of these tests, it will be flagged
according to the results of the Group II tests.
|
|
| II |
| |
A clear day was indicated.
|
|
Global above clear-day threshold (Bird):
- Is Direct lower than Global minus Bird Diffuse?
- Is Diffuse lower than Rayleigh Diffuse?
- Is Diffuse higher than Bird Diffuse?
Direct above clear-day threshold (Bird):
- Is Global lower than Direct plus Rayleigh Diffuse?
- Is Diffuse lower than Rayleigh Diffuse?
- Is Diffuse higher than Bird Diffuse?
Flags are assigned now. If a single parameter failed Group I testing, it is the
only one flagged. Otherwise, all parameters that failed ("yes" to a Group II question)
are flagged.
|
|
| III |
| |
Check coupling among parameters (if a test fails ("yes" to a question) and there was no prior Group I or Group II failure flag all parameters).
|
|
- By what percent is Global less than the Direct?
- By what percent is Global less than the Diffuse?
- By what percent is Direct less than the Global minus
the maximum Diffuse (KD = 0.55)?
- By what percent is Global less than the Direct plus the Diffuse?
- By what percent is Global greater than the Direct plus the Diffuse?
At this point, the largest percentage is applied to the questionable parameter(s).
|
|
NOTE:
