Berkeley Earth is presenting new research at the European Geosciences Union’s EGU2025 Conference, taking place from April 27th – May 2nd in Vienna, Austria. This year, our team will highlight significant differences between historical temperature reconstructions and present improvements to future climate projections.
Chief Scientist Dr. Robert Rohde will present new findings that highlight important differences between observational and reanalysis temperature datasets—particularly in the last decade. His talk will shed light on the implications these divergences have for understanding past climate trends.
Additionally, Dr. Devin Rand will debut our latest bias-corrected, downscaled CMIP6 dataset—a high-resolution tool designed to help planners and policymakers apply global climate model data at the local level. This work builds on our mission to improve climate data transparency, usability, and real-world impact.
You can find Berkeley Earth at the following EGU25 sessions:
Monday April 28, 2025
15:05 CEST
Session CL5.3 – Room 0.49/50 (Austria Center Vienna)
EGU25-13840: Contrasting Berkeley Earth’s New High Resolution Temperature Dataset with ERA5
Robert Rohde presents a focused comparison between Berkeley Earth’s new high-resolution temperature dataset and the ERA5 reanalysis product. While both offer global surface air temperatures at a quarter degree spatial resolution, their methods of construction differ: Berkeley Earth directly interpolates weather station and ocean measurements, incorporating quality control and machine learning to correct for biases and improve spatial accuracy. ERA5 assimilates a wide range of remote sensing data and observations with a physically consistent weather model—but only uses station temperatures indirectly, leading to potential discrepancies.
This session will highlight systematic differences between the datasets in seasonality, warming trends, and regional behavior.
Friday May 2, 2025
09:55 CEST
Session CL4.9 – Room 0.49/50 (Virtual Presentation)
EGU25-13866: A New Synthesis: Bias-Correcting and Downscaling CMIP6 for Local Decisions
Devin Rand presents a new high-resolution climate projection product from Berkeley Earth, designed to bridge the gap between global climate models and local decision-making needs. This dataset combines 45 CMIP6 models (374 runs total) with ERA5 reanalysis and Berkeley Earth observational data to produce bias-corrected, statistically downscaled surface temperature projections at a quarter degree resolution.
The result is a globally consistent, locally relevant synthesis that provides accurate probability estimates of future weather events. Dr. Rand will explain how this synthesis improves accuracy, captures uncertainty, and delivers actionable insights for climate risk analysis, infrastructure planning, and policy development. A forthcoming extension will apply the same framework to precipitation.
