Extreme temperature brain teaser

[Arty]

Hello everyone,

I have a small climate-modeling brain teaser that I hope some of you may find interesting.

I am analyzing future temperature projections over Tahiti (French Polynesia) using WRF at 2.3 km resolution. As expected under global warming, mean near-surface temperature increases in the future simulations, with the strongest warming signal generally occurring over the north-west (leeward) coast of the island (Figure 1).

However, when I examine the annual number of days exceeding a fixed temperature threshold (e.g., 32°C), I obtain a rather counterintuitive result: the largest increase in exceedance days occurs along the trade-wind-exposed (windward) coasts, particularly on the eastern and southern sides of the island (Figure 2).

What makes this even more puzzling is that historical in-situ observations suggest that threshold exceedances are currently more frequent on the north-west coast, which is generally warmer, sunnier, and less ventilated.

To provide additional context, I also include precipitation and wind maps (Figures 3 and 4), which illustrate the cloudiness and ventilation patterns. These fields show only limited changes between the historical and future simulations.

I suspect that several interacting processes may be involved, including:

• surface radiative forcing,
• cloud feedbacks,
• temperature and moisture advection,
• land-sea interactions,
• changes in the diurnal temperature cycle,
• and possibly differences in temperature variability rather than mean warming alone.

Has anyone encountered similar behavior in tropical island simulations, or can think of a physical explanation for why threshold exceedances would increase more strongly on the windward side despite larger mean warming on the leeward side?

Any thoughts or suggestions for additional diagnostics would be greatly appreciated.

Thank you!

 

[Arty]

Still interested in some insight, if anyone has one. Thanks.

 

[William.Hatheway]

Did you adjust the base temp parameter in wrf?

 

[William.Hatheway]

Here's my initial thoughts

Changes in regional trade-wind patterns and surrounding sea-surface temperatures could help explain why extreme-temperature days increase more strongly along Tahiti’s windward coasts.

Warmer ocean water would heat and moisten the marine boundary layer before it reaches the island, so the windward side may receive less effective oceanic cooling than it does historically. Even if the leeward coast experiences greater mean warming, many windward locations may currently sit just below the 32°C threshold.

A modest increase in marine-air temperature, reduced ventilation, or longer periods of warm onshore flow could therefore push many more days above the threshold for three consecutive hours.

Small changes in trade-wind speed, direction, stability, or sea-breeze timing could also shift ventilation, cloud formation, orographic effects, and warm-air advection.

 

[Arty]

Here's my initial thoughts

Changes in regional trade-wind patterns and surrounding sea-surface temperatures could help explain why extreme-temperature days increase more strongly along Tahiti’s windward coasts.

Warmer ocean water would heat and moisten the marine boundary layer before it reaches the island, so the windward side may receive less effective oceanic cooling than it does historically. Even if the leeward coast experiences greater mean warming, many windward locations may currently sit just below the 32°C threshold.

A modest increase in marine-air temperature, reduced ventilation, or longer periods of warm onshore flow could therefore push many more days above the threshold for three consecutive hours.

Small changes in trade-wind speed, direction, stability, or sea-breeze timing could also shift ventilation, cloud formation, orographic effects, and warm-air advection.

Hello William,

Thank you for your insights.

As shown below, wind speed only increases slightly in the future climate, with only minor changes in wind direction. I will also extract the SST fields, as you are right that I have not yet checked whether SST patterns evolve differently between the windward and leeward coasts.

What strikes me most is that colleagues of mine working with Météo-France AROME model outputs find a much stronger increase in the number of TX32d events along the northwestern coast, which is what we would expect given the larger increase in mean temperatures there. In contrast, my own results show almost the opposite spatial pattern. This discrepancy has led to some lively discussions, but so far no one has been able to explain what is driving it.

I will need to investigate the underlying physics with additional diagnostics, and my colleagues will do the same on their side. We are not using the same forcing, either. Nevertheless, our simulations generally agree very well in terms of changes in wind, rainfall, temperature, and other climate variables, which makes this difference even more intriguing.

I have also attached Future–Historical maps showing the relatively small projected changes in rainfall and wind.

P.S. I did not adjust the base_temp parameter.