Estimating Uncertainties in the Multidecadal Variability of Colorado River Flows from Treering Data
The Colorado river supplies water to forty million people. The river system in crisis due to; an ongoing twenty year drought, a historical overallocation of river flows and an increasing demand for water due to an increasing population. One historical cause of the crisis was the assumption one hundred years ago by the signatories of the Colorado River Compact that the natural flows of the Colorado River do not vary significantly on timescales longer than a decade. Since then, one hundred and fifteen years of gauged Colorado river data indicate that the natural flows fluctuate significantly on all the timescales of the gauged record (up to one hundred years). Rhee et al (2019) have used a gaussian inference model to predict future river flows on the timescale of a few decades. A key parameter of this model is the slope (spectral index) of the fourier power spectrum of the yearly river flow fluctuations. This parameter can be best estimated from the thousand year record of Colorado river flows inferred from tree ring measurements. The trees are located in the upper Colorado River Basin at locations where tree growth is most sensitive to annual precipitation. In a recent paper, Butgen et al., 2021 have demonstrated that the estimation of long term fluctuations in tree ring proxies is quite sensitive to the detrending method as well as the proxy calibration method used. We use their approach to estimate the uncertainties in the Colorado river flow multidecadal variatiability (power spectrum spectral index). These uncertainties are critical in estimating the probabilities of reduced river flows for the coming decades which are critical for risk assessment and determining river management policies.