Elias Massoud Research Scientist, Oak Ridge National Laboratory, USA.

Title: Diagnosing the Representation of Surface and Layered Soil Moisture in Earth System Models

Abstract: Surface soil moisture (mrsos) and vertically integrated soil moisture (mrsol) over the top 10 cm should, by definition, be physically consistent in Earth System Models (ESMs). However, an evaluation of nine CMIP6 models reveals substantial inconsistencies: in some models, mrsos and integrated mrsol agree globally; in others, they align only in specific regions; and in a few, they diverge across all grid cells. These discrepancies arise from a combination of factors, including metadata errors, inconsistent variable definitions, or diagnostic sequencing within the model. We demonstrate how such issues can lead to significant biases, even when both variables are present and seemingly well-defined. As model complexity increases and multi-model comparisons become more common, assumptions about variable equivalence may lead to flawed conclusions. This presentation highlights the need for routine consistency checks, improved metadata standards, and community-wide practices that ensure reliability of derived variables across ESM outputs, particularly in preparation for CMIP7.

Doyeon Kim Postdoctoral Researcher, University of Toronto, Canada.

Title: What Factors Explain the Current Arctic Albedo and Its Future Change?

Abstract: The Arctic has experienced rapid sea ice loss and a substantial surface albedo decline, altering its radiation budget. CMIP6 (Coupled Model Intercomparison Project Phase 6) models capture these trends but show considerable inter‐model spread in the magnitude, distribution, and seasonality of Arctic surface albedo. Over land, spread in AMIP (Atmospheric Model Intercomparison Project) and CMIP6 is associated with snow cover variations, while over the ocean, where sea ice dominates, the sources are less clear. We compare CMIP6 simulations with observations from the Clouds and the Earth’s Radiant Energy System (CERES) and develop a decomposition method to quantify contributions from sea ice albedo, concentration, and extent. Over the Arctic Ocean, all three factors contribute to inter‐model spread in CMIP6. In AMIP, despite prescribed sea ice concentrations, we were surprised to find an inter‐model spread in Arctic Ocean albedo similar to that in CMIP6, driven solely by sea ice albedo. Applying the decomposition to future projections shows that the largest decline occurs in the Central Arctic, driven primarily by reductions in sea ice extent. After 2045, sea ice extent emerges as the dominant driver, highlighting ice edge retreat as key to future albedo decline. These findings pinpoint key sources of inter‐model spread in Arctic surface albedo and offer insights into quantifying shortwave (SW) radiative effect associated with sea ice responses in future projections.

Juliette Lavoie Climate Services Specialist, Ouranos, Canada.

Title: CMIP6 data usage: Lessons learned from more than 200 million downloads

Abstract: As we move into CMIP7, the growing complexity of the CMIP models is putting real pressure on the infrastructure needed to store, distribute, and use this data. In this talk, we ask: what can CMIP6 usage statistics teach us about how to prepare better for future eras? We present an analysis of how CMIP6 data was downloaded across the Earth System Grid Federation (ESGF), the primary repository for CMIP simulations. We break down usage by variables, experiments, institutions and regions. We also look beyond ESGF to data hosted through the community platforms Pangeo on commercial clouds and through the Copernicus Climate Change Service (C3S). Based on these findings, we offer recommendations for data producers and distributors on how to prioritise resources more effectively and how to build better usage tracking going forward.