The team quantified precipitation seasonality with a deliberately simple but powerful metric: for every year and every grid cell, they computed running six-month precipitation totals and identified the consecutive wettest and driest six-month windows. The difference between those two accumulations defines a seasonality index, or SI. Unlike fixed calendar-season approaches, this moving-window framework lets the timing of the wet and dry phases shift from year to year and from place to place, which matters in a region where monsoon onset and withdrawal dates vary considerably. A larger SI means a sharper contrast between the rainy and rainless halves of the annual cycle, and it is this contrast, rather than annual totals alone, that governs when water is actually available to farms, forests, and reservoirs.

Three observation-based products told a consistent story. The gauge-based Climate Prediction Center dataset recorded a seasonality increase of 8.92 percent per decade between 1981 and 2020, the satellite-gauge blended CHIRPS product showed 3.30 percent per decade, and the merged satellite-gauge GPCP dataset showed 4.24 percent per decade, all statistically significant. The ERA5 atmospheric reanalysis also produced a positive trend, a striking 20.40 percent per decade, but with such wide uncertainty that the authors treat it as a supplementary reference rather than confirmatory evidence. For context, a recent global analysis found that precipitation seasonality over global land areas has been increasing at roughly 3.1 to 3.85 percent per decade, which means the Indochina Peninsula is amplifying at least as fast as, and in some datasets considerably faster than, the planet as a whole.

Crucially, the study dissects where in the annual cycle the amplification comes from. Because the seasonality index is simply the difference between wet-season and dry-season accumulation, either component could in principle drive the trend. The observations point firmly to the wet side. In the CPC dataset, wet-season precipitation rose by 25.37 millimeters per decade while dry-season precipitation was essentially flat at 0.26 millimeters per decade. CHIRPS showed a 17.00 millimeter per decade wet-season increase alongside a 4.93 millimeter per decade dry-season decline, which compounds the contrast in the same direction. GPCP and ERA5 showed weaker and partly offsetting component trends, but across all datasets the dominant signal was the same: the monsoon half of the year is getting rainier, while the dry half is changing comparatively little.

That asymmetry has a plausible physical explanation rooted in thermodynamics. According to the Clausius-Clapeyron relationship, the atmosphere’s moisture-holding capacity rises by roughly seven percent for every degree Celsius of warming. In a monsoon climate, that extra moisture is most consequential when convection is already active, so the wet season absorbs most of the added atmospheric water vapor while the dry season, limited by circulation rather than by moisture supply, sees little benefit. The authors are careful to note that their study does not explicitly diagnose circulation changes or moisture transport pathways, and they flag monsoon dynamics, land-sea thermal contrast, and moisture transport as targets for future mechanistic work. But the thermodynamic argument aligns neatly with what both observations and models show.

To move from correlation to attribution, the team turned to the Detection and Attribution Model Intercomparison Project, part of the Coupled Model Intercomparison Project Phase 6. Five models, ACCESS-CM2, ACCESS-ESM1-5, CanESM5, MIROC6, and MRI-ESM2-0, provided parallel simulations in which the historical period is driven either by all external forcings combined, by greenhouse gases alone, by anthropogenic aerosols alone, or by natural forcings such as solar variability and volcanic eruptions alone. Trends in each experiment were estimated with the Theil-Sen slope estimator, a robust regression technique that limits the influence of anomalous years, and then averaged with equal weights across the models.

The results were strikingly directional. Under greenhouse-gas-only forcing, the multi-model mean seasonality index climbed by 15.32 millimeters per decade, driven almost entirely by a 15.96 millimeter per decade increase in wet-season precipitation, and four of the five individual models agreed on the positive sign. Under aerosol-only forcing, the picture inverted completely: all five models produced negative seasonality trends, with a multi-model mean of minus 16.07 millimeters per decade, as wet-season precipitation fell by 17.22 millimeters per decade. Natural forcing produced negative trends in four of five models, averaging minus 8.89 millimeters per decade. The all-forcing experiment landed in between, at plus 12.13 millimeters per decade, consistent with aerosols partially offsetting the greenhouse-gas-driven amplification. Spatially, greenhouse-gas forcing produced coherent positive seasonality trends across nearly the entire peninsula, while aerosol forcing produced widespread negative trends, and the all-forcing response was comparatively weak and mixed, exactly what one expects when opposing forcings compete.

The authors stress that the magnitude of the simulated responses varies substantially among models, and that only one realization per model and forcing experiment was used, so some of the spread reflects internal climate variability rather than structural differences alone. They also acknowledge that the five models were selected because they all offer the required single-forcing experiments over a common period, not because they were pre-screened for regional precipitation skill, although a prior evaluation of 35 CMIP6 models against the APHRODITE observational dataset found that most capture the major features of mainland Southeast Asian rainfall climatology. Sensitivity tests shifting the analysis start year to 1982 or 1983 left the sign and relative ordering of the forcing responses unchanged, suggesting the conclusions are not artifacts of a particular window.

The stakes for the region are considerable. The Indochina Peninsula hosts the Mekong, Irrawaddy, Salween, and Chao Phraya river systems, all of which are paced by monsoon rainfall, and its densely populated agricultural economies depend on the seasonal timing of water availability. A wet season that concentrates more rain raises flood risk in basins like the Mekong, while a dry season that shows little change in precipitation, combined with rising atmospheric evaporative demand under warming, points toward intensified seasonal water stress. Because the observed amplification is driven by wetter wet seasons rather than drier dry seasons, adaptation strategies may need to focus less on drought preparedness alone and more on managing heavier monsoon-season runoff, reservoir operation, and flood defenses.

The study also highlights a broader lesson for climate attribution science. Annual precipitation totals and even precipitation extremes, the metrics that dominate most climate assessments, can mask profound reorganizations of when rain falls within the year. Here, modest changes in annual totals coexist with a strengthening seasonal contrast that reshapes water availability, agriculture, and hazard profiles. As global analyses continue to identify anthropogenic fingerprints on the annual precipitation cycle, this regional study demonstrates that monsoon domains can amplify the global signal, and that separating the competing influences of greenhouse gases, aerosols, and natural variability is essential for projecting how the world’s most seasonally dependent societies will experience a warming climate.

Subject of Research: Attribution of observed changes in precipitation seasonality over the Indochina Peninsula to anthropogenic greenhouse gas forcing using observations and CMIP6 single-forcing simulations

Article Title: Changes in precipitation seasonality over the Indochina Peninsula during 1981–2020: observational evidence and attribution of anthropogenic forcing

Article References: Zhang, Y., Yang, K., & Du, S. (2026). Changes in precipitation seasonality over the Indochina Peninsula during 1981–2020: observational evidence and attribution of anthropogenic forcing. Frontiers in Earth Science, 14, Article 1937024. https://doi.org/10.3389/feart.2026.1937024

Image Credits: AI Generated

DOI: 10.3389/feart.2026.1937024

Keywords: precipitation seasonality, Indochina Peninsula, monsoon climate, CMIP6, DAMIP, greenhouse gas forcing, aerosol forcing, wet-season precipitation, Mekong River basin, climate attribution, Clausius-Clapeyron, water resources