The reason for the difference is that this particular USACE graph is stuck at 0900 Nov 7. This graph constantly has problems with dropping offline or missing readings from the gauge.
The NWS graph is generally more reliable though you can see it is currently missing data from Nov 6 to Nov 8.
This USACE Roselle graph below is more reliable. This covers 4 days and gives readings every 15 minutes:
http://www.mvs-wc.usace.army.mil/trans/sf/sfro_s.html
This Roselle graph covers 30 days with hourly data shown but it only updates once a day at 0700:
http://www.mvs-wc.usace.army.mil/trans/sf/sfro_30s.html
The following graphs show precipitation at Farmington and Roselle, and CFS at Roselle: You can toggle between 30 day data and 4 day data, except the 4 day CFS graph isn’t working at the moment.
http://www.mvs-wc.usace.army.mil/trans/sf/sffa_30p.html
http://www.mvs-wc.usace.army.mil/trans/sf/sfro_30p.html
http://www.mvs-wc.usace.army.mil/trans/sf/sfro_30fc.html
Black marks on the graph indicate missing data. The graph extrapolates and fills in missing data to draw a line, but no data will be shown in the tabular listing on the left.
The USACE graph on this webpage has some advantages (when it is working). It shows stage and precipitation at the same time. It can also be customized to show up to a year of data. Delete the “7” at the end of the URL and type in the number of days you want to see, up to 360 days. (Jim Warren was the first to discover this feature, as far as I know.)