Homogeneous verification data samples

1
Homogeneous verification data samples

• Cases from the 2004 and 2005 seasons where there
  was an operational
• GFDL forecast, a 2006 GFDL forecast, and an
  observation from the best-track
• record with which to verify the forecasts

Forecasts are from the following TCs 2004
03, 04, 06, 07, 09, 11, 12, 13 2005 09, 12,
14, 15, 18
2
Joint distribution comparison 2006 GFDL vs.
operational GFDL
0 h lead time
3
Joint distribution comparison 2006 GFDL vs.
operational GFDL
12 h lead time
4
Joint distribution comparison 2006 GFDL vs.
operational GFDL
24 h lead time
5
Joint distribution comparison 2006 GFDL vs.
operational GFDL
36 h lead time
6
Joint distribution comparison 2006 GFDL vs.
operational GFDL
48 h lead time
7
Joint distribution comparison 2006 GFDL vs.
operational GFDL
72 h lead time
8
Joint distribution comparison 2006 GFDL vs.
operational GFDL
96 h lead time
9
Joint distribution comparison 2006 GFDL vs.
operational GFDL
120 h lead time
10
Marginal distribution comparison Obs., 2006
GFDL, op. GFDL,
0 h lead time
11
Marginal distribution comparison Obs., 2006
GFDL, op. GFDL,
12 h lead time
12
Marginal distribution comparison Obs., 2006
GFDL, op. GFDL,
24 h lead time
13
Marginal distribution comparison Obs., 2006
GFDL, op. GFDL,
36 h lead time
14
Marginal distribution comparison Obs., 2006
GFDL, op. GFDL,
48 h lead time
15
Marginal distribution comparison Obs., 2006
GFDL, op. GFDL,
72 h lead time
16
Marginal distribution comparison Obs., 2006
GFDL, op. GFDL,
96 h lead time
17
Marginal distribution comparison Obs., 2006
GFDL, op. GFDL,
120 h lead time
18
Summary measures 2006 GFDL vs. operational GFDL
Mean absolute error (accuracy)
19
Summary measures 2006 GFDL vs. operational GFDL
Mean error (unconditional bias)
20
Summary measures 2006 GFDL vs. operational GFDL
Normalized mutual information (information
content)