On this stream gage in Beaumont, the storm surge is strong enough to reverse the flow in the river. The gage height graph at the top shows the typical water-level rise of a flood, but the bottom graph shows streamflow (discharge) that went negative when the storm hit. Positive values are flow downstream, negative means the river is flowing upstream. (The negative values will show up on the default graph for 7 days, then you can adjust the plotting period to see them for 60 days before they age off.)
Cool. I didn't know our gages could even record that. (OK, I'm a ground-water hydrologist, not a surface-water one, so I never thought about the problem before). Must be the relatively new equipment that measures velocity as well as gage height. The old-fashioned method of stream gaging was/is that you install a continuous monitor for the gage height, and then you go measure the streamflow over the range of gage heights and construct a stage-discharge curve so you can translate any gage height to a streamflow. Well, any height that's in the range of your curve...thus the desire to go measure the stream anytime the river gets higher or lower than you've measured before, and the reason that field technicians go out to try to measure streams at the peak of a flood. But a gage height monitor wouldn't let you know that the river was running backwards.
I note that this is a gage right at the coast, and it shows flows that are just barely negative regularly (tides). Under Ike's influence, it's flowing at quantities well above the daily maximum recorded there, and upstream. However, it looks like the flow has already reversed and that water will all now go the other way--and the daily average may well come out close to zero. I'll be interested to see...