Nonparametric Stochastic Hydroclimate Simulation for Water Temperature Modeling in Lake Shasta

Abstract:

Reservoir managers on the Sacramento River are required by law to provide artificial cold water habitat downstream for endangered winter-run Chinook salmon. This is enabled at Shasta Lake via a temperature control device installed on Shasta Dam that allows selective withdrawal of reservoir water from different elevations and temperatures. Risk based decision making and planning requires ability to generate ensemble of water temperatures released from the lake - especially when the planning needs to be made under future climate conditions. To this end, we developed a stochastic hydroclimate simulation method that generates ensembles of influent lake streamflow, influent stream temperatures and air temperature over the lake. These, combined with a two-dimensional hydrodynamic model, CE-QUAL-W2 provides ensembles of water temperatures released from the various levels of the lake. A nonparametric K-nearest neighbor based disaggregation method is used to generate streamflow ensembles at five streams entering the lake. Then, conditionally the temperatures of water entering the lake and the air temperature over the lake are also simulated. The W2 model generates lake temperatures. The disaggregation method is also modified to generate streamflows consistent with wet and dry conditions and consequently, the lake temperature scenarios, enabling the water managers to assess various decision options.