EMDNA: an Ensemble Meteorological Dataset for North America

Summary

This paper presents EMDNA, a novel ensemble meteorological dataset for North America combining station observations and reanalysis outputs to quantify uncertainty in spatial precipitation and temperature fields. The methodology integrates multiple reanalysis products via Bayesian model averaging and optimal interpolation, with validation showing substantial improvements over raw reanalysis estimates, particularly in data-sparse and topographically complex regions. The dataset provides a probabilistic framework useful for evaluating the impact of meteorological uncertainties across agricultural, hydrological, and climate-related applications.

UK applicability

EMDNA is specific to North America and not directly applicable to UK conditions. However, the methodological framework for fusing station observations with reanalysis data using Bayesian model averaging could inform development of analogous ensemble meteorological datasets for the United Kingdom, particularly for improved uncertainty quantification in regions with sparse gauge networks.

Key measures

Spatial resolution (0.1° grids, ~10 km), temporal coverage (1979–2018), ensemble member count (100), meteorological variables (daily precipitation amount, mean daily temperature, daily temperature range), accuracy metrics for probabilistic evaluation

Outcomes reported

The study developed EMDNA, an ensemble meteorological dataset comprising 100 members with daily precipitation, mean temperature, and temperature range at 0.1° spatial resolution (approximately 10 km grids) from 1979–2018 across North America. The dataset was created by fusing station observations with reanalysis model outputs using Bayesian model averaging and optimal interpolation, with evaluation demonstrating improved performance in high-latitude and mountainous regions.

Topic tags