IntermtntClimateSummaryJune2006

• Title: Intermountain West Climate Summary
• Date: 6/22/2006
• Description: Water Availability Task Force Meeting Presentation
• Basin: Statewide
• Drought Mitigation - Doc Type: Presentation

FEATURE ARTICLE FROM INTERMOUNTAIN WEST CLIMATE SUMMARY, JUNE 2006

Updated Streamflow Reconstructions
Using an updated and expanded set of tree-ring chronolo-
gies which end in 1997 or later, the authors created high-qual-
ity streamflow reconstructions for four key gages in the Upper
Colorado River basin (hereafter the "WGM" reconstructions) .
The reconstructions \:vere generated by calibrating the tree-ring
chronologies \:vith the most recent naturalized flows from the
USBR, including data that were available as of summer 2004.
The WGM reconstructions span the common years 1569 to 1997,
with the Lees Ferry reconstructions extending to 1490 (avail-

X 1i04
2.4

2.~

~
~ 18
~.
~ 1.6
u:::

1.4

1.2 " " ". . " ":" " .. -" . " ''''1)'" ." . " ." . ":-" .

1
1500 1550 1000 1.65(1 1 '700 1750 1 SOO 1 ~o 1900 1 GOO 2000
EndiJl!l Year

Figure 1 c. Twenty-year running means of four alternative recon-
structions of the annual flow of the Colorado River at Lees Ferry
for common period 1520-1961. Lees-A and D are from WGM
(2006), SJ1976 is the mean of two reconstructions from Stockton
and Jacoby (1976), and HPD2000 is from Hidalgo et al. (2000).
The horizontal lines are the 1906-2004 observed mean (solid)
and the lowest observed 20-year running mean of the 1906-2004
period (dash-dot).

able at: ww\:v.ncdc.noaa. gov /paleo/pubs/\:voodhouse2006/wood-
house2006.html.
Because reconstructions can be sensitive to different model-
ing methods, two different models were tested for the Green,
Colorado near Cisco, and the San Juan gages, and four different
models were tested for Lees Ferry flows. The models vary in
the pool of chronologies used and the statistical approaches used
for handling the data. Ho\:vever, the authors found that differ-
ent modeling methods had little significant impact on important
features of the reconstructions, for example, the long-term annual
mean, or the runs of drought years.
However, one factor that does have some impact on the mag-
nitude of reconstructed high and lo\:v flows is how the persistence
in year-to-year growth is treated in the tree-ring data. In the Lees
Ferry reconstructions, the authors tested the effects of either re-
taining the year-to-year persistence in the tree gro\:vth in the chro-
nologies, or removing this persistence The persistence-retained
chronology models retain a degree of year-to-year persistence
similar to that in the gage record, but overestimate the severity of
multi-decadal droughts (20-year means) in the calibration period.
Comparison with Previous Lees Ferry Reconstructions

Hidalgo et al. (2000) generated reconstructions at Lees Ferry
as well as Stockton and Jacoby (1976). Both of these recon-
structions were calibrated on the years 1914-1961, using similar
sets of tree-ring data but slightly different modeling approaches.
Interannual variations in streamflow are similar across all of the
reconstructions, and a comparison of these Lees Ferry recon-
structions and that of WGM sho\:vs a relatively close match
in variations of streamflow at decadal and longer time scales
(Figure lc). The WGM reconstructions differ from those of
Stockton and Jacoby (1976) and Hidalgo et al. (2000) in suggest-
ing a somewhat higher long-term mean for Upper Colorado River
flows (Table 1, after Table 9 in WGM, 2006). The most conser-
vative and most extreme Lees Ferry reconstructions generated by
this study (called Lees A and Lees D) had long term (1520-1961)
means that ranged from 14.3 to 14.7 maf, compared to Stockton
and Jacoby's preferred Lees Ferry reconstruction, and the Hi-
dalgo et al. reconstruction with means of 13.5 maf and 13.2 maf,
respectively. The set of chronologies and gage records used for
the reconstruction calibration may contribute to these differences,
but the reconstruction methods used are likely also a factor. The
causes of the differences are currently under investigation.

Table 1

Model

Calibration period Long-term mean & 95%
confidence interval

WGM- Lees Ferry-A 1906-1995
WGM- Lees Ferry-B 1906-1995
Stockton & Jacoby (1976) 1914-1961
Hidalgo et al. (2000) 1914-1961

14.7:t 0.2 MAF
14.3 :t 0.2 MAF
13.5 *
13.2:t 0.02 MAF

*no confidence interval calculated, not nzeaningful for the average of
two reconstructions

Highlights of the Reconstructed Streamflows
The recent 2000-2004 drought, as measured by 5-year running
means of water-year total flow at Lees Ferry, is clearly a severe
event \:vhen assessed in the context of the 500-year tree-ring re-
construction. It is highly unlikely (i.e., the probability is less than
10%) that any 5-year flows since 1850 has been as low (Figure
Id). But in considering the uncertainty in the reconstruction
and evaluating the reconstruction probabilistically, flows for the
period 1844-1848 were lower that 2000-2004, and there are eight
periods bet\:veen 1536 and 1850 that had at least a 10% probabil-
ity of being as dryas 2000-2004. In addition, droughts longer
than any in the gage record have occurred in the past. The Lees
Ferry reconstruction includes up to eleven consecutive years with
flows below the 1906-1995 average.
Severe multi-year, decadal, and multidecadal periods of

FEATURE ARTICLE I 3 I~