The team — Christelle Barthe of the Laboratoire d’Aérologie in Toulouse, Françoise Vimeux of HydroSciences Montpellier and the Laboratoire des Sciences du Climat et de l’Environnement, Camille Risi of the Laboratoire de Météorologie Dynamique, and Sören François of LSCE — has implemented a full water stable isotopologue scheme in Meso-NH, a non-hydrostatic mesoscale atmospheric model developed in France. The new version, called Meso-NH-ISO v1.0, was described and validated in a paper published in Geoscientific Model Development. It allows the model to track two heavy water molecules, HD16O and H218O, alongside ordinary H216O through every stage of a simulated storm, from the first updraft to the last evaporating raindrop.

The significance of this achievement lies in what the isotopes reveal. When water evaporates from the ocean, heavy isotopes are slightly left behind, making the vapor relatively depleted. As air rises and cools in a storm, condensation preferentially removes the heavier molecules, so the remaining vapor and the precipitation it produces become progressively lighter — a process known as fractionation. Because different atmospheric processes leave different isotopic marks, measuring the ratio of heavy to light water in rain and vapor provides a natural tracer of what happened to that water. Rain that has partially evaporated on its way down, for instance, becomes isotopically enriched and loses deuterium excess, a diagnostic that combines deuterium and oxygen-18 measurements. By simulating these processes explicitly, scientists can decode what field observations actually mean.

Previous modeling efforts have taken different approaches. Some cloud-resolving models, such as DHARMA, use bin microphysics schemes that track individual particle sizes with great precision but at enormous computational cost, typically following only one isotopologue. Others, including SAM-ISO, COSMO-ISO, WRF-ISO, and NICAM-WISO, use bulk microphysics schemes that represent entire populations of droplets and ice crystals with averaged properties, allowing both HD16O and H218O to be simulated simultaneously. Meso-NH-ISO follows this bulk approach, coupling the isotope scheme to the model’s ICE3 one-moment bulk mixed-phase microphysics scheme, which predicts the mass of six water species: vapor, cloud droplets, rain, pristine ice, snow aggregates, and graupel.

The implementation required careful treatment of which microphysical processes fractionate isotopes and which do not. Processes that simply convert one form of water to another without a phase change — such as raindrops growing by collecting cloud droplets, or snow forming by aggregating ice crystals — transfer heavy isotopes in proportion to their abundance and leave the isotopic ratio unchanged. Melting is also assumed to be non-fractionating, since each ice crystal melts completely. But processes involving evaporation, condensation, and deposition do fractionate, and the model handles these with different levels of sophistication depending on the particle size involved.

For cloud droplets, which are tiny and equilibrate with surrounding vapor within seconds, the model assumes isotopic equilibrium between liquid and vapor, using well-established equilibrium fractionation factors. For raindrops, which are large enough that equilibrium cannot be assumed, the model explicitly computes the diffusive exchange between the falling drop and the ambient air, accounting for the slower molecular diffusivity of heavy water molecules and the ventilation effects that enhance evaporation for larger drops. For ice particles growing by vapor deposition, the scheme follows the classical formulation of Jouzel and Merlivat, which combines equilibrium and kinetic fractionation effects. Sublimation of ice back to vapor is treated as non-fractionating, consistent with previous modeling studies.

To validate the new scheme, the team chose a legendary test case: the West African squall line observed during the night of June 23–24, 1981, in northern Ivory Coast as part of the COPT 81 field campaign. This storm was extensively documented with radars, soundings, and surface observations, and had already been successfully simulated with a 2D version of Meso-NH in 1994. The researchers ran a two-dimensional simulation with 1.25-kilometer horizontal grid spacing and 46 vertical levels, initializing the atmosphere with the observed sounding and imposing an isotopic profile based on Rayleigh distillation. A small cold pool triggered the convection, and the simulation ran for eight hours — enough time for the squall line to reach maturity.

The results were striking. The simulated storm developed the classic three-part structure of a mature squall line: a convective region with intense updrafts at the leading edge, a transition zone with dry subsiding air, and a broad stratiform region with an anvil cloud trailing behind. Within this structure, the isotopic fields behaved exactly as theory and observations predict. In the convective zone, updrafts carried isotopically enriched vapor upward from the surface while downdrafts brought depleted air down from aloft. Rain falling through dry downdrafts partially evaporated, becoming enriched and losing deuterium excess. In the stratiform region, rain originating from melted snow high in the cloud was more depleted than locally formed rain, while rain evaporating in the dry air beneath the anvil became strongly enriched, with the fraction of evaporated rain exceeding 90 percent in places.

Perhaps most remarkably, the simulation reproduced the famous W-shaped pattern of oxygen-18 in surface precipitation that has been observed along squall lines in the Sahel since the 1990s. Two isotopic minima appear in the convective zone and at the start of the stratiform region, separated by an enriched peak in the transition zone where rain evaporation is strongest. The model also captured the sharp decrease in vapor isotopes at the cold pool front — a feature observed in more than 85 percent of squall lines monitored in Niamey, Niger — and the continuous decline of deuterium excess in both precipitation and vapor along the storm, consistent with field observations from the AMMA campaign.

The team emphasizes that this validation is just the beginning. Their ultimate goal is to apply Meso-NH-ISO to realistic cases of tropical cyclones and squall lines, where recent observations have revealed highly organized isotopic signals in water vapor on hourly timescales that current climate models cannot resolve. By explicitly simulating the convective dynamics at kilometer scale while tracking isotopes through every phase change, the model should allow scientists to quantify exactly which processes — rain evaporation, mesoscale descent, convective detrainment, or ice sublimation — control the isotopic composition of water in these systems. That, in turn, will sharpen the interpretation of isotope records used to reconstruct past storm activity, from tropical cave deposits to the archives being assembled to understand how cyclone frequency has changed over centuries. The code has been released openly, inviting the broader community to push this isotopic lens deeper into the atmosphere’s most dramatic phenomena.

Subject of Research: Implementation and validation of a water stable isotope scheme in the Meso-NH mesoscale atmospheric model using a West African squall line simulation

Article Title: Meso-NH-ISO v1.0: a water stable isotopologue scheme in the non-hydrostatic mesoscale atmospheric model Meso-NH – application to a 2D West African squall line

Article References: Barthe, C., Vimeux, F., Risi, C., & François, S. (2026). Meso-NH-ISO v1.0: a water stable isotopologue scheme in the non-hydrostatic mesoscale atmospheric model Meso-NH – application to a 2D West African squall line. Geoscientific Model Development, 19(20), 9935-9956. https://doi.org/10.5194/gmd-19-9935-2026

Image Credits: AI Generated

DOI: 10.5194/gmd-19-9935-2026

Keywords: water isotopes, Meso-NH, squall line, convection, isotopic fractionation, rain evaporation, cloud microphysics, Sahel, tropical cyclones, deuterium excess, oxygen-18, atmospheric modeling