1. ALL the following plots are subject to the filtering :
Wind direction at AWS# < θ < 285,
* also 3 m/s < wind speed < 25 m/s
(for cut-in and cut-out speeds)
also only use 12PM-2AM data.
Gives: ON, approx 16 hours of data (188 samples);
OFF, approx 8 hours of data (97 samples)

2. Average temps for “on” and “off”

3. Analysis I. Temperature differences.
Look at temperature differences at sites up-and downwind of a selected turbine,
for “on” and “off” conditions A B C D E F

4. Analysis I. Times series of variance:
every hobo included
OFF ?
Mean wind (on) = 4.2 m/s
Mean wind (off) = 4 m/s

5. Times Series for hobos A, B, C

6. PCA analysis.
An objective method for determining underlying
patterns in data.
Many meteorological (usually climatological) applications.
Very simple matter to determine the underlying
structures…
…interpreting the structures is the difficult part;
often the results have no obvious physical significance.

7. PCA produces three types of analysis:
The empirical orthogonal functions (EOFs): the patterns, or structures, in the data;
The principal components (PCs): a time series, reflecting the relative contribution of each EOF at a given time
The eigenvalues: give the overall importance of each EOF

8. PCA analysis.
Time series of principal
components
OFF
EOF1: 98.9%
EOF2: 0.5%
EOF3: 0.4%

9. EOF1
Average
EOF2
EOF3

10. Reconstructing the data from the EOFsα +β +δ

11. PC2 x1 +
PC3 x 0.5
PC2 x -1 +
PC3 x 1.9

12. Summary of initial analysis
There appears to be a signal.
Much more scatter when “on” than “off” for adjacent sites.
Variance for the whole domain increases when “on”.
Signal seen in PCA analysis also - although more work is needed on
the interpretation of underlying structures.
Underlying structures appear to imply a warming across the site
(although not very large in magnitude)
Signal present during latter stage of “off” phase. Why?

13. 1. Form the data matrix X containing your data;
X is of size K x N (K stations, measurement
points, grid points, etc; N samples)
2. Calculate the covariance matrix S, based on X;
3. Solve Se = le for the eigenvectors e and
eigenvalues l (K EOFs and eigenvalues)
4. Solve P = Xe to calculate the principal
components (N PCs)
Many off-the-shelf packages, e.g. IDL, have
PCA routines.