NSFM requires a continuous monthly time series of streamflow and a continuous monthly time series of an ENSO indicator as input data.

Various standard TIME/Toolkit time series formats are supported by NSFM (see eWater Data File Formats 1.2.pdf).

After downloading NSFM to the default directory, example data files can be found in c:\Program Files\Toolkit\NSFM\Data.

NSFM provides outputs of model calibration and cross-verification diagnostics (as tables and plots), forecast skill statistics, and the exceedance probability forecasts (as tables and plots).

The outputs from NSFM can be exported as text files.

NSFM is a robust model that has been tested using data across Australia. It uses a non-parametric approach to empirically fit the data, without making any prior assumption about the model structure. As such, the model results will be as good as the data and the relationships used to derive the forecasts.

In general, at least 30 years (if possible 50 years) of data should be used. The skill in the forecast is directly related to the correlation between the streamflow (or other hydroclimate variable) to be forecast and the predictor variables. A poor model skill will be reflected in NSFM exceedance probability forecasts that are no better than the forecasts derived solely from climatology.