Estimation of Stream Flow Losses to the Highland Lakes Inflows during the Recent Period of Run-off Non-stationarity

Abstract:

The Highland Lakes are operated by the Lower Colorado River Authority (LCRA) in Texas to provide water supply to municipal, industrial, agricultural users and environmental flows for the river and Matagorda Bay. The Highland Lakes also provide for hydroelectric generation and recreation. Subject to extended droughts interrupted by intense rainfall, the region has the nickname of Flash Flood Alley. Recently the Lower Colorado River has experienced a seven year historic drought. While precipitation have been 80% of average, runoff into the Highland lakes have been at historic lows.

Multiple regression relationships were developed to predict runoff to the LCRA lakes from rainfall and other factors which explain about 2/3 of the variation of observed inflows. This explanation is good considering the inherent error in stream flow measurement and inflow estimation. It is also comparable to the skill of much more complex dynamical models. Review of the residuals from the relationships reveals periods of unfavorable non-stationarity in inflows after accounting for statistically significant climate and seasonality variables. In particular the periods from 1977 to 1987 as well as 2002 to present showed uncharacteristically low runoff as can be seen in the figure below.

Through use of dummy variables for the periods of apparent non-stationarity, the effects of climate and non-stationarity can be quantitatively estimated. At a 90% confidence level, the excess losses in run-off from 2002 to 2015 that can be attributed to lower than median rainfall ranges from 101,000 to 137,000 acre-feet. About another 32,200 to 45,300 acre-feet annually of unrealized inflows can be attributed to typical drought processes. Finally, about another 182,000 to 478,000 acre-feet per year of unrealized inflows can be attributed to unexplained factors in this recent period. These losses are in contrast to runoff during the calendar year of 2011 in which low runoff was better described by extreme climatic conditions.

While the statistical significant variables of the exceptional conditions allows estimation of the impacts of non-stationarity, it does not identify a specific cause. Additional research is needed to understand the drivers of these periods of non-stationarity and the return to typical conditions.