2. Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec
Assuming
July was this
years’ Peak
Fire Month4 Months
before
4 Months
After
9 Month Window
3. Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec
M2 M5 M8 M11
Taking the Mean
of every 3 month
Span
3-Month Mean (Moving Window Average)
4. The sum of the 9 month window is
considered to be the FSS for that
given year.
Select one value from
M1 to M12 to
represent the Sea
Surface Temperature
Anomaly of the
given year.
0
2
4
6
8
10
12
14
2001 2002 2003 2004
FSS
ONI
AMO
Find the
Correlations
5. Driver vector corresponding to peak lead time
was bootstrapped and the correlation between
this annually driver vector and FSS was
computed for each of 10,000 iterations.
Experiment 1
6. Driver vector was bootstrapped preserving
seasonality before undergoing any
transformations.
The driver vector is bootstrapped and supplied
as input for the newly constructed Annual
driver vector.
Correlate
Experiment 2
7. After a point is clicked, pick a random
neighbor for each year.
Store that target value into a vector length
10,000.
Repeat for 10,000 iterations
Experiment 3