SR SSD 99-3


Technical Attachment

Verifying River Forecasts with and without Quantitative Precipitation Forecasts

Bill Lawrence, DOH

Arkansas Basin River Forecast Center

Tulsa, OK

Since February 1997, the Arkansas Basin RFC in Tulsa has been issuing daily river forecasts that contain both QPF and non-QPF time series. The intent was to allow our WFOs to see the effect of their QPF on river forecasts, as well as to allow the ABRFC to easily do comparative verification for the two forecasts. The chart below displays the results for all of the daily forecasts (that is, all the daily river forecast points) for each month of 1998.

The graph shows interesting results. January, February and October 1998 showed the best improvements of the river forecasts with QPF, over those without the QPFs. Those three months, all during the cooler, non-convective season, were also the wettest months of the year. During the cool season, when stratiform situations dominate, QPF has proved very valuable in improving river forecasts. On the other hand, during May, June and July, the QPFs actually hurt our forecasts overall, but only by a small amount. This is consistent with what we found during 1997, when during the summer convective season, QPF had little if any positive value in terms of improving the river forecasts.

The last two years notwithstanding, we do not want to make any hard and fast decisions about the operational value of QPF in forecasts until we have a larger data set, perhaps five years worth. Also, flood forecasts were not included in this study, only daily forecasts. More data can be viewed on the ABRFC web site at The web site includes data from 1997, and individual daily forecast verification for each month of 1998.