Thunderstorms in the Climate Maze ✛
Context
Thunderstorms are not just a spectacular sight; they are also a source of many hazards: lightning, hail, squalls, flash floods, and even wildfires from 'dry' strikes. Understanding how often and where they occur is critical for planning infrastructure, agriculture, and warning systems. However, direct observations of thunderstorms worldwide are uneven, and climate models are too coarse to reproduce small-scale processes. Therefore, scientists use indirect indicators: combinations of temperature, humidity, and wind that point to favorable conditions for thunderstorms. This work joins a line of research that tries to extract a signal from the noise of weather, relying on multi-year data and statistical methods.
Methods
The authors used ERA5 reanalysis data from 1979 to 2023, a kind of 'reconstruction' of the atmosphere based on observations. For each grid point with a step of 0.25 degrees, they computed two key parameters: convective available potential energy (CAPE), which shows how much energy is available for air to rise, and wind shear between the surface and a height of 6 km, which helps organize the storm. Multiplying these quantities yielded the thunderstorm environment index (BTE). The threshold for a 'thunderstorm day' was tuned so that the frequency matched lightning sensor data from 2012 to 2023. The scientists then compared the variability of this index with indices of El Niño, the Indian Ocean Dipole, and polar oscillations, and assessed long-term trends using a bootstrap test and an ensemble of 12 climate models.
Results
The variability map showed that thunderstorm activity 'shivers' most in the tropics and over the western ocean basins, while at high latitudes it is more stable. El Niño turned out to be the main conductor: about 25% of the entire studied area shows a statistically significant link to its phases, and in the tropics this share reaches 52% over the ocean. The Indian Ocean Dipole is noticeable in September–November, but its influence is closely intertwined with El Niño. Polar oscillations act mainly at high latitudes, adding information where tropical modes are silent. Over 45 years, 15% of grid points saw an increase in thunderstorm environments, and 8% a decrease, but these changes are small compared to natural fluctuations. Models predict an increase in thunderstorm activity with warming: on average, 30 more thunderstorm environments per year per degree of global warming, mainly due to increased humidity rather than wind shear.
Implications
The study shows that thunderstorm variability cannot be reduced to a single factor: El Niño dominates in the tropics, but the poles have their own logic. What sets this apart from neighboring studies is that it is the first to apply a unified global thunderstorm environment index calibrated against lightning and consistently compared with four climate modes. This allows a view of the big picture without losing regional details. For science, this means that seasonal predictions of thunderstorm activity could rely on El Niño phases, but with caution: the signal is not uniform everywhere.
Future development
The topic will evolve toward finer tuning: using high-resolution regional models, accounting for specific types of hazardous phenomena (hail, squalls, lightning), and expanding the ensemble of next-generation climate models. It may be possible to uncover nonlinear interactions between modes, for example, how El Niño and the Indian Ocean Dipole amplify each other through convection over the Maritime Continent. This will require new data and machine learning methods.
Impact
The results could be useful for seasonal risk forecasting in agriculture, energy, and water resource management. Insurance companies and emergency services would get more robust estimates of the likelihood of thunderstorm disasters.
Next steps
The immediate next step is to test how well the BTE index predicts thunderstorms in regions with sparse observations and to compare it with direct lightning network data beyond 2012–2023. It is also worth studying how the relationship with El Niño changes under different warming scenarios.
Key open problems
In meteorology, an open question remains: how to distinguish an anthropogenic trend from natural variability in short data records. This work shows that for thunderstorms, the warming signal is still comparable to noise, and longer records or more sensitive methods are needed. Also unresolved is the problem of parameterizing convection in models; the BTE index sidesteps it but does not eliminate it.
The knowledge machine's view
This section is addressed to the author of the paper and their colleagues. Neighbouring works were found by meaning-based search across our archive, not by word overlap. We do not review: we show the field around the work.
What the work does
The authors built a global map of thunderstorm activity over 45 years using ERA5 data, computing a thunderstorm environment index (BTE) as the product of CAPE and wind shear, with the threshold calibrated against lightning sensors. They compared BTE variability with El Niño, the Indian Ocean Dipole, and polar oscillations, and assessed long-term trends using a bootstrap test and an ensemble of 12 climate models.
Where its strength is
The strength of the work lies in a unified global thunderstorm environment index calibrated against lightning observations, which allows systematic comparison of the influence of different climate modes and identification of regional features.
Where it could go next
- It might be interesting to see how the proposed thunderstorm environment index relates to changes in precipitation regimes: a nearby paper shows that rains are becoming less frequent but more intense, and one could check whether this shift is accompanied by increased thunderstorm activity in the same regions.based on: Rain Is Learning to Fall Less Often but Harder arXiv:2601.06408
- Another possible step is to account for the influence of solar activity on atmospheric circulation: the archive contains an analysis of seasonal variations in space weather, and it would be interesting to test whether it modulates the thunderstorm environment on decadal timescales.based on: Space Weather by the Seasons arXiv:2605.27057
- One could also consider a connection with oceanic processes: a paper on internal waves in the ocean interior may suggest how deep dynamics affect thunderstorm conditions through surface temperature.based on: Invisible Waves in the Ocean's Depths arXiv:2601.01476
Worth trying
- Test whether the thunderstorm environment trend intensifies in regions where precipitation becomes more intense, using data from paper 2601.06408.
- Add a solar activity index to the regression analysis and see whether it explains part of the residual variability in BTE.
- Compare BTE anomalies with the phases of internal ocean waves to identify a possible teleconnection.
Why this may matter
If thunderstorm activity can be linked to other climate indicators, this will improve seasonal forecasts of hazardous phenomena and our understanding of how thunderstorms respond to global warming.
Nearby works in our archive
- The Fourth Spring of Climatesimilarity 0.51
- Invisible Waves in the Ocean's Depthssimilarity 0.50
- The Sun's Tidal Rhythmssimilarity 0.50
- Gravitational Visitors: How Passing Planets Rewrote Earth's Historysimilarity 0.50
- How a Desert Remembers Its Historysimilarity 0.50
- Rain Is Learning to Fall Less Often but Hardersimilarity 0.49
- Space Weather by the Seasonssimilarity 0.49
The archive is sparse around this work: the closest review is only 0.51, against a usual 0.64. It looks like few have drilled nearby.
The threshold for a thunderstorm environment is unique at each point on the planet: it is tuned so that the number of 'thunderstorm' hours matches the number of recorded lightning strikes. In dry deserts, this threshold can be tens of times lower than in humid tropics.
Key numbers
- Data period: 1979–2023
- Share of points with significant El Niño link: 25% of the entire area
- Average increase in thunderstorm environments per degree of warming: about 30 per year
- Standard deviation of annual number of thunderstorm environments: about 28
- Share of points with increased thunderstorm environments over 45 years: 15%