Readme 1/14/16

Fluxes were processed with 3 different correction schemes applied to the psd Windmaster sonic

1) Standard psd motion/flux processing to the raw wind data
2) Application of the Gill W correction factors to compensate for an error in Windmaster
     firmware: +W multiplied by 1.166, -W multiplied by 1.289  (_Wcorr files)
3) Application of the Nakai(2012) correction (aoa2012.m) to raw Windmaster uvw. (_AoA files)

There appears to be no significant difference between 2) and 3) with respect to flux results.
Suggest using the _AoA results for now.

The folder Processed_v3 uses corrections 1 & 3 above for the psd wind data. PSD DAS
occasionally looses 10Hz data.  The da file has data on missing points in each
10min interval for the sonic, motionPak and licor.  This information may be used to filter
10min results.

The Windmaster shows high frequency rolloff above 1Hz compared to the R2.  This issue was
reported by Nakai(2014).  However, mean, selected wu cospectra for the R2 and Windmaster
are almost identical, so it's not clear the rolloff issue affects the stress result.
At this time we are not applying a spectral correction as described by Nakai(2014).

Mean, filtered Ts spectra from the Windmaster look better than the R2.  But hsc and hsib
from the R2 seem to correlate better with COARE, although both datasets have issues.  Neither
dataset looks good for negative hs.  Windmaster hsib has greater scatter than the R2 hsib.
Windmaster hsc it high compared to COARE.  R2 hsc has quite a bit of bad data - lower or negative
compared to COARE.

Inertial latent heat results compare very well with COARE.  Covariance hlc is quite
scattered.