TL;DR
While global warming is expected to increase extreme rainfall intensity by about 7% per degree Celsius due to increased atmospheric moisture (Clausius-Clapeyron relationship), this study reveals significant spatial variations, particularly strong negative scaling over tropical oceans. This suggests that atmospheric circulation dynamics, especially those related to tropical cyclones, play a crucial role and must be considered alongside thermodynamics when predicting future rainfall extremes.
What Problem Does This Paper Address?
A central challenge in climate science is understanding how extreme rainfall events will change in a warming world. The prevailing thermodynamic theory, based on the Clausius-Clapeyron (CC) relationship, predicts a roughly 7% increase in rainfall intensity for every 1°C rise in temperature due to increased water vapor in the atmosphere. However, observational data and climate models show that this scaling isn't uniform globally. This paper investigates the reasons for these spatial heterogeneities, specifically focusing on the role of atmospheric and oceanic dynamics in tropical regions, where extreme rainfall can have devastating impacts. The authors aim to disentangle the thermodynamic influence from dynamic processes to provide a more nuanced understanding of future extreme rainfall.
Methods
The researchers employed a global analysis of temperature and rainfall data, focusing on tropical oceans. They defined extreme rainfall events as those exceeding the 90th percentile of wet-time rainfall rates. To investigate the temperature-rainfall relationship, they analyzed the scaling between near-surface air temperature and extreme rainfall intensity. A key methodological innovation was the focus on pre-rainfall cooling – the decrease in temperature in the hours leading up to an extreme rainfall event. They developed a method to quantify this pre-rainfall temperature gradient and its correlation with rainfall intensity. To isolate the effects of rainfall on temperature, they specifically selected rainfall episodes that occurred after at least 48 hours of dry conditions. They also conducted a separate analysis focusing only on rainfall events associated with tropical cyclones (TCs) to assess their specific contribution.
Datasets
- Tropical Rainfall Measurement Mission (TRMM) 3B42 V7 dataset: Provided 3-hourly rainfall estimates at a 0.25° spatial resolution for the period 1998-2018.
- ERA5 reanalysis dataset: Supplied 1-hourly near-surface air temperature data at a 0.25° spatial resolution for the same period.
- NOAA OI SST V2 High Resolution Dataset: Used for average sea surface temperatures (SSTs) from 1998-2018 at 0.25° resolution.
- International Best Track Archive for Climate Stewardship (IBTrACS): Utilized to identify and tag rainfall episodes associated with tropical cyclones.
Key Insights and Ideas
- Thermodynamics isn't the whole story: While the global median scaling aligns with the CC relationship, the spatial patterns are highly variable, especially over tropical oceans. This highlights that dynamic processes are crucial for understanding regional rainfall changes.
- Pre-rainfall cooling is key over oceans: A significant finding is the strong negative correlation between pre-rainfall cooling of near-surface air temperature and extreme rainfall intensity over tropical oceans. This means that as the temperature drops before a storm, the storm itself tends to be more intense.
- Tropical cyclones drive the negative scaling: The authors attribute this pre-rainfall cooling and its link to intense rainfall to the atmospheric and oceanic dynamics associated with tropical cyclones. TC passage can mix cooler subsurface waters and draw in cooler air, leading to this observed pre-storm cooling.
- Cyclonic activity dominates rainfall generation: The scaling behavior observed for all rainfall episodes over tropical oceans is very similar to that observed for TC-tagged episodes, suggesting that cyclonic activity is the primary driver of extreme rainfall in these regions.
Key Results
- Global average vs. regional reality: The global median temperature-rainfall scaling factor is approximately 6.0-6.3%°C⁻¹, close to the CC expectation of ~7%. However, spatial patterns show significant deviations.
- Negative scaling over tropical oceans: Over tropical oceans, a substantial portion (around 33%) of locations exhibit negative temperature-rainfall scaling, meaning rainfall intensity decreases with warming temperatures. In some areas, this negative scaling is as strong as -40%°C⁻¹.
- Robust correlation: A robust linear correlation was found between pre-rainfall temperature cooling and extreme rainfall intensity, with a Pearson correlation coefficient (PCC) of -0.987 in tropical ocean regions.
- TC influence: Even when considering only TC-tagged rainfall episodes, the observed scaling patterns and correlations remain remarkably similar to those for all rainfall episodes, reinforcing the dominant role of cyclonic activity. For instance, the scaling between temporal temperature gradients and rainfall intensity over northern tropical oceans was -4.02 ± 0.0088 mmh⁻¹/0.1°Ch⁻¹ for all episodes and -3.98 ± 0.0369 mmh⁻¹/0.1°Ch⁻¹ for TC-tagged episodes.
Limitations and Future Work
- Data limitations: The study acknowledges that the TRMM rainfall product may underestimate extreme rainfall compared to finer-scale radar data, particularly over oceans. However, it remains the best available dataset for this region.
- TC definition: The IBTrACS archive only includes TCs with sustained wind speeds above a certain threshold, meaning some cyclonic activity might be missed.
- Other dynamic influences: While TCs are identified as a primary driver, the authors suggest that other (thermo)dynamic contributions to scaling, especially positive ones, warrant further investigation.
- Future projections: The authors emphasize that future projections of extreme rainfall must incorporate these dynamical effects, as they are likely to lead to pronounced spatial heterogeneity and substantial deviations from purely thermodynamic expectations. Further research into the specific mechanisms driving these dynamical influences is needed.
Summary updated