The stochastic response surfaces of seasonal time varying data can often result in results that are both visually interesting and useful.  In some ways these graphics are reminiscent of wavelet approaches.

Here are representations of the stochastic landscape of the Pacific Decadal Oscillation (PDO) and the Otowi Gage records in North Central New Mexico over the 20th Century plus change.  For each time series, the years are represented by columns and each row represents a separate month.  The rows start from (end of) December at the bottom and go to (end of) November at the top row.

The apparent ergodic properties of the two time series allow an exploration of related ‘stationary waves’.  Consider a metaphor of water flowing over a rocky stream bed.  In this example, the Otowi gage record is the ‘water’ and the PDO record is the ‘rocky stream bed’.  Some caveats include fall and winter Otowi stream flows, which are depressed across the time series, likely due to watershed storage and freezing conditions and not necessarily due to lack of precipitation.

The next image captures the relative proportions of monthly flows for each year of record.

The next image shows the same coverage for absolute flows.  In this image, a viewer can see the actual magnitudes of flow for every month of every year.

In our open approach (with a propietary core) to streamflow forecasting, we employ the best available stochastic practices and data sets.