Let's see if we can outline the main points we want to cover for the temporal bias section, as well as some evidence that can be included for each:
The database has changed what it records over time
Initially, only tornadoes
Then things like hail added
Then added lots of others in 1996
Evidence for this point can include the information in the database documentation about what events they started recording then, other resources on the history of this database.
From the "Details" vignette, I think we have a figure that shows how dramatically the database size has changed over time. This is something that we could include to illustrate this temporal evolution.
Population growth over time means there are more people to spot and report events.
I think there are some papers on tornadoes that talk about how more are reported in areas with more people. There may be for other types of hazards in the database, too.
If we wanted, we could take a place with high population growth over the last 50-ish years and look at how population growth and reporting of tornado events have evolved (and compare it with someplace that was already pretty densely populated in the 1950s). However, there might be enough evidence already through papers that have focused just on this topic.
While we tend to think of temporal trends in terms of long-term trends (over years to decades), there are also patterns across the year---by season---that might result in shifting probabilities in whether an event that happens in reported. This can be related to population movement---a tourist beach might have a much higher chance of reporting a rip current in the summer, when it's full and people are swimming, than in the winter. We have the rip current plot we could use to illustrate this, plus your research on the paper on that.
Technology has changed, and this has enabled us to detect more stuff.
We could try to make a plot that shows how number of reports from different sources have changed over time (I had this as a suggestion for certain event types earlier---this could be the same figure here). I'm guessing there are some reporting sources that have only come "online" more recently. CocoRaHS, for example, I think is a pretty new network, so they wouldn't have reported any events before they existed. Social media is another source that probably has only reported things starting pretty recently, since that's a new-ish technology. This article might be promising: https://journals.ametsoc.org/view/journals/mwre/138/8/2010mwr3270.1.xml
Let's see if we can outline the main points we want to cover for the temporal bias section, as well as some evidence that can be included for each: