The graphic is illustrating how the current year compares to the cluster of previous years. It would be more informative if it were interactive and you could hover over or label other years, but that requires Javascript abilities that this news outlet doesn't have. I think I've seen other presentations of this data and there is a clear later-is-hotter trend, but it's subject to the normal El Nino / La Nina oscillation plus random fluctuations.
The bar graph is measuring a different quantity: it's ocean temperatures in the equatorial Pacific (vs. all over the world), and specifically illustrates the El Nino cycle. The reason this is important is because overall ocean temperatures tend to be cooler in La Nina years anyway. One point of this research is that the naturally cooler temperatures from La Nina may have masked ocean warming that would've happened over 2020-2023, and now that we're switching to El Nino conditions, the temperature increase will come back with a vengeance.
All the lines are averages. Each line represents a year.
The red line is the current year, which is why it stops at August.
>And if each gray line represents a year, we can't tell if there is a steady increase year/year, all we see are clusters for different years.
We can however see that 2023, starting from April, is consistently higher than any and all previous years.
It sounds like you're interpreting "average" as being an aggregate over time, but it is actually an aggregate over space at a given point in time.
The graph you're getting the 2015 number from is not comparable. It is only about the temperatures in the limited area in the Pacific where the El Nino effect happens.