In a previous post I explored a demonstration of another ARMA based forecast for a stream in the Platte River Watershed of Central Colorado. It is only a demo but may capture a significant persistence of cycles which often overlap with others which go by different names, including the often – noted Pacific Decadal Oscillation, or the patterns featured in a paper I authored earlier this year and featured in the Hydrological Sciences Journal . In that article I compared solar based forecasts to “conventional” auto regression based forecasts. It turns out that in some areas, the ARMA forecasts can have great skill with long lead times and small moving averages. This forecast of the Cache La Poudre tributary of the Platte River may be a good candidate.
So far the forecast which represents a 15 year projection of the annual average flow in cfs calculated from USGS stream-gage data at the ends of the first and second quarters of each calendar year, appears to be working out fairly accurately. For this year, one could say that the forecast nailed it. Moreover, perhaps this accuracy is unprecedented for such a precise unit of time (annual) for such a long span.
Yet this demonstration is simply a very easily reproducible example of how effective the conventional methods can be for forecasting, if one knows where to look. As with any forecast featured here, these are simply for demonstration purposes and there is no guarantee of future performance. Moreover this is draft material from a work in progress and is only informally shared.
 Wallace, M.G., 2019, Application of lagged correlations between solar cycles and hydrosphere components towards sub-decadal forecasts of streamflows in the Western US. Hydrological Sciences Journal, Oxford UK Volume 64 Issue 2. doi: 10.1080/02626667.2019.
1220total visits,1visits today