Real time streamflow forecasting using a snowmelt index model and time series analysis techniques

A snowmelt index model was developed to generate an input time series suitable for forecasting snowmelt and streamflow in the Salmon River Watershed in central New York using a time series model. The snowmelt index model is a modified temperature index model that produces a continuous daily time series. The snowmelt index is generated based on air temperature, a melt factor (which is partially a function of snow density) and a lag coefficient (which is estimated using observed streamflow data). The output of the snowmelt index model is not an estimation of the quantity of melt generated by the snowpack, but rather an index of the magnitude of melt. The snowmelt index is then stochastically related to streamflow using a method of time series modeling known as transfer function noise modeling. -from Authors