Flood-frequency comparison from 1995 to 2016 and trends in peak streamflow in Arkansas, water years 1930–2016

In 2016, the U.S. Geological Survey, in cooperation with the U.S. Army Corps of Engineers and the Federal Emergency Management Agency, began a study in Arkansas to investigate possible increasing trends in annual peak streamflow data and the possible resulting increase in the annual exceedance probability flood (AEPF) predictions. Temporal trends of peak streamflow were investigated at 15 selected streamgages on unregulated streams in Arkansas having 30 or more years of peak streamflow data through the 2016 water year. For the period of record at each streamgage, the Mann-Kendall trend test indicated that 14 of the 15 streamgages had no statistically significant peak streamflow trends and 1 streamgage had a statistically significant decreasing peak streamflow trend. Visual examination of the locally estimated scatterplot smoothing technique trend lines of the peak streamflow data indicated a possible increasing peak streamflow trend at 8 of the 15 streamgages since the 1990s.

A sequential series analysis of the 1-percent AEPF at each of the 15 selected streamgages was completed by selecting an initial subset of the oldest peak streamflow data from each site to estimate the initial 1-percent AEPF. This initial peak streamflow data subset was subsequently appended with 10-year increments of additional peak streamflow data until the full period of peak streamflow data was analyzed. The maximum increase in the 1-percent AEPF was 113 percent, and the maximum decrease was 31.9 percent.

Percentage differences between the AEPFs derived from regional regression equations presented in the 1995 and 2016 Arkansas flood-frequency reports were compared. The average percentage differences for the 74 selected locations indicate that the 4-, 2-, 1-, and 0.2-percent AEPFs computed using the 2016 regional regression equations were higher by 3.52, 5.10, 8.59, and 13.31 percent, respectively (25-, 50-, 100-, and 500-year recurrence interval floods), than the same percentage AEPFs computed using the 1995 regional regression equations. The average percentage differences between the 1995 and 2016 AEPFs for the 10-percent AEPF (10-year recurrence interval flood) resulted in 2016 AEPF predictions being 0.41 percent higher. For the 50- and 20-percent AEPFs (2- and 5-year recurrence interval floods), the 2016 AEPFs were less than the 1995 AEPFs by 2.53 and 0.31 percent, respectively.