Empirical Statistical Downscaling Evaluations

---

Our Statistical Downscaling Research Team’s efforts

Overview:

The Empirical Statistical Downscaling (ESD) team at NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL) provides independent, objective evaluations of statistical downscaling and bias correction techniques commonly used to transform dynamical climate model outputs into localized forecasts and projections.  By providing information on the methods’ performance characteristics, we aim to help users select the most appropriate data products and techniques for their specific application, ultimately supporting more robust decision-making and resiliency planning.

Our team’s efforts provide insights of value to three primary communities interested in climate variability and change. They are…

Applied Science Researchers & Decision-Makers

The Challenge: For many climate impacts studies, the raw output from large-scale dynamical climate models is viewed as having inadequate spatial resolution and/or unacceptably large departures from historical, real-world observations that make it unsuitable for direct use in applications of interest.  While bias-corrected and empirical-statistical downscaling (ESD) products are designed to address these shortcomings, choosing the right dataset can be difficult because performance varies by region, season, and methodology.  For this reason, there is no universal answer to the common question asked by users of ESD data products: “Which of the publicly available downscaled climate projection data products is the best one for me to use in my next study?”

Our Impact: Through thoughtfully designed experiments, rigorous analyses, inter-comparisons, and other tests, we generate information that can help researchers determine which ESD products are truly “fit for purpose” for specific resilience planning and decision-making needs on time scales ranging from months to multiple decades. Our findings do more than just point the way toward successful outcomes. They also document specific pitfalls and conditions that users should avoid. For examples, see our Downscaled Climate Projections for Heat & Health Applications page.

Empirical-Statistical Downscaling Community

The Challenge: Lacking observations of the future, it is difficult to evaluate how well bias correction and empirical-statistical (ESD) techniques will hold up when applied to future projections. Additionally, many ESD products were originally developed with specific uses in mind, but are now used as input data in a much wider range of applications with varied data sensitivities and needs. This widespread use can expose unexpected variations in how different methods handle extreme events, temporal sequences, and long-term trend signals passed down from large-scale dynamical models.

Our Impact: We address these uncertainties by using analysis frameworks, such as our “perfect model experimental design“, to conduct rigorous quantitative assessments of downscaling performance in both contemporary and future climates. Our work helps ESD method developers identify and isolate weaknesses and refine their techniques. By investigating the underlying reasons for variations between products, we shine a light on specific performance characteristics and trade-offs of different methods, helping the community understand when and where an ESD dataset is suitable for use.

Global Climate Modeling Community

The Challenge: Global Climate Models (GCMs) are Earth science research tools built to simulate large-scale systems and better understand how our complex climate works. Because GCM developers focus on capturing the key mechanisms and feedbacks that govern global and regional climate variability and trends, re-purposing the GCM simulations for local-scale impacts and adaptation planning are generally not front-and-center when next-generation GCMs are being developed. Using GCM simulation results for local decision-making essentially means taking a foundational research tool and adapting and refining its outputs for applied research that is generally outside its primary scope.

Our Impact: Empirical-statistical downscaling can be viewed as a post-processing step that directly challenges GCM outputs with data from the observational record and subsequently refines the simulations to generate a value-added product. By confronting model simulations with real-world, high spatial resolution historical data, our analyses help isolate specific structural strengths and weaknesses within GCMs. We focus particularly on surface variables, providing dynamical modeling teams with information that can inform future model developments and better support resource managers and decision-makers.

SELECT JOURNAL PUBLICATIONS :

♦ Xu, WR, KW Dixon, N Zenes, and J Lanzante (2026) Examining the Impact of Bias Correction Configurations on a Multivariate Meteorological Index: A Case Study of Heat Index Analysis in the Northeast United States. Journal of Applied Meteorology and Climatology, doi: 10.1175/JAMC-D-25-0128.1

♦ Sun, L., KW Dixon, KE Kunkel, X Sun, and DR Easterling (2026) Comparative Assessment of LOCA2 and STAR-ESDM Downscaled Surface Temperature over the Conterminous United States. Journal of Applied Meteorology and Climatology, doi: 10.1175/JAMC-D-25-0176.1

♦ Xu, WR, KW Dixon, N Zenes, and D Adams-Smith (2025) Sometimes missing the heat: the risk of underestimating extreme heat days with daily maximum heat index approximation. International Journal of Biometeorology, https://doi.org/10.1007/s00484-025-03001-7

♦ Le Roy B, KW Dixon, and D Adams-Smith (2024) High-resolution urban climate simulations for heat and health applications in Philadelphia. Urban Climate, doi: 10.1016/j.uclim.2024.102114.

♦ Wootten AM, KW Dixon, DJ Adams‐Smith, RA McPherson, (2021) Statistically Downscaled Precipitation Sensitivity to Gridded Observation Data and Downscaling Technique, International Journal of Climatology, doi:10.1002/joc.6716

♦ Lanzante JR, KW Dixon, MJ Nath, CE Whitlock, D Adams-Smith, (2018) Some Pitfalls in Statistical Downscaling of Future Climate. Bulletin of the American Meteorological Society, doi:10.1175/BAMS-D-17-0046.1

♦ Dixon KW, JR Lanzante, MJ Nath, K Hayhoe, A Stoner, A Radhakrishnan, V Balaji, CF Gaitán (2016) Evaluating the stationarity assumption in statistically downscaled climate projections: is past performance an indicator of future results? Climatic Change. doi:10.1007/s10584-016-1598-0

---

PEOPLE
Team Members & Collaborators

---

SELECT JOURNAL PUBLICATIONS
Some Papers of Interest

---

PROJECTS & LINKAGES
Past & Present

---

NEWS & PRESENTATIONS

SELECT CONFERENCE PRESENTATIONS: