Generally speaking you may want to include a column for scheduled sampling time in your dataset for later processing. In your case you can simply round your actual sampling time to get the scheduled ones ASSUMING NO LARGE DEVIATIONS FROM YOUR SAMPLING SCHEME. Otherwise there is no way to do it. Check your data carefully before doing that, e.g.


The next step will be calculating median of your concentration data grouped by scheduled time.


Now you have every thing you need for the plot. I still suggest you to do your plot using actual sampling time even though the medians are calculated per scheduled time.


Results:


As your actual time is quite close to the schedule in my approach. There will be little difference if you plot using schedule time.
For comparison: