Geoscience Reference
In-Depth Information
steeply dropping reservoir levels and concerns among water resources professionals
about the resilience of the Colorado River system to severe drought. The question
rephrased in nonstatistical terms then, is how unusual was the low-flow period 2000-
2004 in the context of the 509-year tree-ring record that extends over the years
1490-1998?
The severity of the recent drought is clearly evident in the time series plots of
consecutive sequence, of years below the drought threshold. If the reconstructed
that the 2000-2004 drought is unprecedented in the tree-ring record, as the longest
sequence of years below the threshold is 4 years. Considering the uncertainty in
the tree-ring reconstruction, however, it is impossible to say categorically that there
has not been another 5-year sequence of such low flows in the period 1490-1998.
A more valid interpretation considers the probabilistic nature of the reconstruction.
By this interpretation, the true flow in any year has some nonzero probability of
being higher or lower than the reconstructed value. Depending on the reconstruction
errors, which are unknown, the sequence of true flows will be somewhat different
than the reconstruction itself.
35
O
bserved
30
Mean
25
20
15
10
0.25 Q
u
antile
200
0-
2004 E
v
ent
5
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
35
R
econstructed
30
Mean
25
20
15
10
0.
25 Qua
nt
ile
2000
-
2004 Ev
e
nt
5
1500
1550
1600
1650
1700
1750
1800
1850
1900
1950
2000
Fig. 8.1
Time series plots of observed (
top
) and reconstructed (
bottom
) annual flow of Colorado
River at Lees Ferry. The mean and 0.25 quantile of the 1906-2004 observed flows are marked
on both plots, along with an observed 2000-2004 'event' of 5 consecutive years below the 0.25
quantile. The maximum number of consecutive reconstructed flows below the 0.25 quantile of the
observed flows is 4 years. Sources of data: observed flows are natural flow series from US Bureau
