Environmental Engineering Reference
In-Depth Information
TABLE 5.3 Median Runoff Sensitivities Relative to the Period 1971-2000 per Degree Global
Average Temperature Change, the Equivalent Standard Error of the Median, and Fraction of
Positive Minus Negative Estimates (FPN) for the U.S. Hydrologic Regions and Alaska
Equivalent Std Error
a
(%)
FPN
b
Hydrologic Region
Median (%)
1
New_England
1.7
0.6
0.47
2
Mid-Atlantic
1.0
0.7
0.29
3
South_Atlantic-Gulf
-2.0
1.1
-0.18
4
Great_Lakes
1.7
1.3
0.15
5
Ohio
1.5
1.0
0.24
6
Tennessee
-2.8
1.1
-0.24
7
Upper_Mississippi
1.2
1.2
0.09
8
Lower_Mississippi
-6.4
1.7
-0.53
9
Souris-Red-Rainy
1.8
1.6
0.21
10
Missouri
-2.0
1.1
-0.12
11
Arkansas-White-Red
-8.4
1.4
-0.56
12
Texas-Gulf
-7.4
2.1
-0.56
13
Rio_Grande
-12.2
2.3
-0.47
14
Upper_Colorado
-6.3
1.7
-0.62
15
Lower_Colorado
-6.1
2.4
-0.44
16
Great_Basin
-5.1
1.4
-0.44
17
Pacific_Northwest
1.2
0.9
0.24
18
California
-3.3
2.2
-0.29
19
Alaska
9.3
0.9
0.91
Colorado
c
-6.2
1.6
-0.59
NOTE: The runoff sensitivity is taken as the average sensitivity (as described below) for IPCC AR4 GCM output
from A2, A1B, and B1 simulations for each basin as derived from 30-year runoff averages centered on the years
for which the global average (for each of 23 models and the 3 emissions scenarios) were 1°C, 1.5°C, and 2°C
minus the 30-year model average runoff for 1971-2000, divided by the global temperature change. The total
number of model and temperature change pairs was 68, consisting of 23 models for 1°C temperature change, 23
models for 1.5°C temperature change, and 22 models for 2°C temperature change. The equivalent standard error
is an estimate of the standard error of the median (Hojo,1931), where “equivalent” pertains to the fact that the
estimate is strictly correct only for normally distributed variates.
a
1.25
σ
/,
n
where
σ
is estimated as one-half the inner 68-percentile range.
b
Fraction of pairs (of 68 total) with inferred positive runoff sensitivities minus fraction with inferred negative
sensitivities. FPN therefore ranges from -1 when 68 inferred sensitivites are negative, to 1 when all are positive.
c
Upper and Lower Colorado combined.
model median runoff sensitivity) across almost all of Eurasia, high-latitude
areas, and most of Australia; (b) decreases across much of the United States,
southern Europe, and Africa; and (c) (not shown) strong agreement as to
the direction of runoff changes at high latitudes and southern Europe, little
agreement across most equatorial areas, and only modest agreement over
much of the United States (for the United States, see also the fraction posi-
tive minus fraction negative in Table 5.3).
Finally, there are some apparent differences in the pattern of runoff
