Because the graphs are averaging X samples over Y time. The larger Y the more samples are smaller since IP traffic is generally bursty. If you were to run a constant-bit-rate application at R for 45 minutes then you would see a 30 minute average at the rate since all of the samples would be at R.
The title of the graphs say it all. Daily is a 5 minute average, weekly is a 30 minute average. It’s not recording the peak, the daily graph is taking all the traffic that happened over 5 minutes and is giving you an average for those 5 minutes in a single time space. Then the weekly is taking all the traffic from the 6 of the “5 minute” time slices and averaging them together to give you a time slice on the weekly graph.