Geoscience Reference
In-Depth Information
a
35
30
T 2m Squared Bias
25
24h
48 h
20
15
10
5
0
A
B
C
D
E
F
G
H
I
b
30
T 2m MSE
24h
48h
20
10
0
A
B
C
D
E
F
G
H
I
c
30
T 2m Squared error SD
20
24h
48h
10
0
A
B
C
D
E
F
G
H
I
100
Fraction of Squared Bias to MSE
24h
48h
80
60
40
20
0
A
B
C
D
E
F
G
H
I
Location
Fig. 25.2
Squared bias
, mean square error (
MSE
), squared standard deviation (
SD
) error and
fraction of Squared bias to MSE (
Er
) for 2-m temperature (
ı
C) forecasts from 1 to 31 May 2006.
Results are plotted for averaged 24- and 48-h forecasts with respect to the regions (
A
,
B
,
C
,
D
,
defined in Fig.
25.1
. Unit: square of temperature (
ı
C) in (
a
), (
b
)and(
c
)(Adapted
from
Xu et al.
(
2009
))
E
,
F
,
G
,
H
,
I/
In the following section, WRF-ARW model forecast error characteristics for
24- and 48-h (e.g. second 24-h) forecasts and diurnal variation are described.
Forcasts results are produced during the 30-day period starting from May 1 through
May 30, 2006.
25.5.1
24-h and 48-h Forecasts
25.5.1.1
2-Meter Temperature
.T
2
-
M
/
Squared biases (WRF forecasts - GFS analysis) in 2-m temperature forecasts vary
with terrain elevation (Fig.
25.2
a). Biases are larger over high terrain areas (E, B, D)
for the 24- and 48-h forecasts. While, the biases are significantly smaller in low
terrain regions (A, C, F, G, I) or water areas (H).
Search WWH ::
Custom Search