WSPC04476

• File Number: 8283.200
• Description: Colorado River Basin-Colorado River Computer Models-Colorado River Decision Support System-RAY
• State: CO
• Water Division: 5
• Date: 12/29/1993
• Title: Hydrosphere-Colorado River Model-Technical Overview
• Water Supply Pro - Doc Type: Report/Study

'I
I
I
I
.
I
I
.
'.
.
I
I
I
I
.
I
I
I
I

002101

Hydrosphere's Colorado River Model

December 29, 1993
Page 8

However, there are limitations associated with the indexed sequential method. These
include the short period of recorded data (which may not include any critical period) and the
fact that only historical traces, which may never be repeated even approximately, are used in
the streamflow synthesis. Because of these limitations, we have developed alternative
streamflow traces, using stochastic modeling techniques and reconstruction from tree-ring
records.

Synthetic Hydrology

Synthetic flows were generated using SPIGOT (Grygier and Stedinger, 1988 and 1990),
a stochastic streamflow generation software package developed at Cornell University. SPIGOT
can be used to generate synthetic streamflows which have similar statistics to those of the
historical record. The generated streamflows preserve cross-correlations between stations and
serial correlations between months as well as the standard deviations and lag-I serial
correlations observed in the historic traces.

We have used the SPIGOT package to generate 1,400 years of streamflow for each of
the 29 inflow points in Hydrosphere's Colorado River Model. Two data sets have been
generated: a trace with 15 maf as its mean and a trace with 13.8 maf as its mean. The 15 maf
trace was developed using the 1906-1983 hydrology and the 13.8 maf trace was developed
using the 1930-1983 hydrology. The annual means of the flows generated by SPIGOT are
within 0.5 % of the historical means at Lee Ferry for their respective parent periods.

Reconstructed Pre-Historic Hydrology

Reconstructed pre-historic streamflow at the 29 inflow points in Hydrosphere's Colorado
River Model have been also been developed. They were derived from the annual streamflows
reconstructed from tree rings at Lee Ferry for the period 1520 to 1961 (Stockton and Jacoby,
1976). The annual reconstructed flows were disaggregated over time into monthly flows at
Lee Ferry, and again disaggregated over space to the 29 inflow points in the basin used in
Hydrosphere's Colorado River Model (Tarboton, 1993).

Climate Change Hydrology

Hydrosphere applied the Colorado River Model to a study of the impacts of climate
change on water supply in the basin. Several hydrologic traces were developed for this study.
Two of them, which correspond to I.C and 2.C changes are provided with the model.

A number of approaches were evaluated for the development of streamflow data sets for
the study. These methods included generation of streamflow data directly from global change
models (GCM's) and using GCM results to generate synthetic temperature and precipitation
traces, from which streamflow traces could be derived. However, the inability of GCM's to
replicate observed conditions in the Colorado River basin and the limited length of GCM time
series data (10 to 25 years) eliminated these methods. Ultimately, the relatively simple
approach of adjusting the stochastic traces was used. The Upper Basin streamflows were
adjusted by factors determined from a relationship developed by Revelle and Waggoner (1983)
which related mean annual flows at Lee Ferry to temperature and precipitation. The Lower
Basin inflows were adjusted according to Stockton and Boggess (1979) to reflect reductions
which were twice the reduction applied in the Upper Basin.

Hydrosphere Resource Consultants 1002 Walnut Suite 200 Boulder, Colorado 80302