TWP/SKYRAD/C1 - PSPs and NIP: Persistent bias

The NIP/Eppley comparison shows that around solar noon,
the NIP is low relative to the Eppleys by about 4% with
this fraction decreasing slowly toward dawn and dusk. 
At very low sun angles, the NIP irradiance exceeds that
from the Eppleys, presumably because of the roll off in
the Eppley's cosine response at those angles.

Larger discrepancies arise between the NIP and and direct
beam calculated from the Eppleys during periods of sun
following rain events. Significant examples of this
phenomenon occurred on October 10 at 02 GMT and 05 GMT
and on October 13 at 20 GMT. In the latter case, the rain
had actually occurred the previous evening.  

We noticed another rain related peculiarity in the NIP
data.  On several occasions, when a rain event followed
a period of full sun, the NIP actually overshot zero and
went negative by about 10 watts (this is an estimated
value only) and persisted for several minutes.  Examples
occurred on October 16, 17, and 31.

In the morning, the sun is apparently blocked by a line
of trees until about 2030 GMT because large but very brief
discrepancies often occur at that time of day.  In the
afternoon, there is often convection over the main part
of Manus Island to the west which blocks the direct beam.

Another likely explanation is dew on the sensors.  I have
requested that the RESET team examine the sensors after
dawn during their next visit (February 1997).  This phenomenon
occurs on November 4, 12, 13, 15, 17, 19, 20, 22, 27, 28, 30.

Throughout the Manus record, there is a distinct tendency
for the sum of the direct plus diffuse SW to be greater than
the corresponding global PSP. The magnitude of the difference
on average fluctuates around 2-3%. There was an increase in
the average percent difference on about 19980509, roughly
corresponding with the reset visit.  However, the daily
average percent differences exhibit a large variability
around this mean, with a range from 0 - 8% for the period
19980509 - 19980731.  Since the beginning of data collection
rarely has the daily average difference been negative. While
differences of a few percent should be expected, I believe
that having these differences always biased warrants closer
scrutiny.  One would typically expect changes in these daily
average differences to be highly correlated with RESET visits,
which is not obvious in the data record, and to randomly switch
sign given a large enough sample size, which we don't have yet.
I recommend that we monitor this to make sure there's not an
unknown SMALL system grounding problem or Campbell logging
program glitch that is causing this bias.

• Shortwave direct normal irradiance, pyrheliometer(short_direct_normal)
• Downwelling Shortwave Hemispheric Irradiance, Ventilated Pyrgeometer, Standard
  Deviation(down_short_hemisp_std)
• Downwelling shortwave diffuse hemispheric irradiance, pyranometer, maxima(down_short_diffuse_hemisp_max)
• Downwelling Shortwave Diffuse Hemispheric Irradiance, Pyranometer, Standard
  Deviation(down_short_diffuse_hemisp_std)
• Downwelling Shortwave Diffuse Hemispheric Irradiance, Ventilated Pyranometer(down_short_diffuse_hemisp)
• Downwelling Shortwave Hemispheric Irradiance, Pyranometer, Minima(down_short_hemisp_min)
• Downwelling Shortwave Diffuse Hemispheric Irradiance, Pyrgeometer, Minima(down_short_diffuse_hemisp_min)
• Down-welling unshaded pyranometer voltage(down_short_hemisp)
• Downwelling shortwave hemispheric irradiance, pyranometer, maxima(down_short_hemisp_max)
• Shortwave Direct Normal Irradiance, Pyrgeometer, Standard Deviation(short_direct_normal_std)
• Shortwave direct normal irradiance, pyrheliometer, minima(short_direct_normal_min)
• Shortwave Direct Normal Irradiance, Pyrheliometer, Maxima(short_direct_normal_max)