Unrestricted Mesoscale Analysis (URMA)

Description

URMA (Unrestricted Mesoscale Analysis) is NOAA's hourly, high-resolution, near-surface “analysis of record.” It reruns the RTMA system approximately six hours after each valid time, assimilating conventional, mesonet, buoy, ship, shelter, and radar-based observations that arrive too late for RTMA. This also includes advanced precipitation QPE from radar and gauge blends, and snowfall updates from NOHRSC, merged via methods like Whittaker blending.

URMA uses a two-dimensional variational (2D-Var) assimilation with static background error covariances (gridpoint-statistical-interpolation, GSI) to combine observations and a first-guess model field. It operates over multiple NDFD Lambert Conformal grids—2.5 km for CONUS (WEXP), ~3 km for Alaska, and similar resolutions for Hawaii, Puerto Rico, and Guam.

The dataset includes standard near-surface fields such as 2 m temperature, dewpoint, specific humidity, 10 m wind (speed, direction, gust), surface pressure, visibility, cloud ceiling, and precipitation. URMA also integrates wave-height fields offshore and daily minimum/maximum temperature. As the analysis of record, it supports forecast verification, NBM bias correction, ML model training, and situational awareness.

URMA's strength lies in its intelligence in merging multimodal data: mesonet densification, radar-gauge precipitation blends, and snow analysis enhancements ensure better accuracy than RTMA. While RTMA targets low latency, URMA focuses on completeness and quality—making it the most trusted truth source in model evaluation and verification workflows.

Notable upgrades: URMA replaced the legacy Stage II precipitation input with MRMS/Multi-Radar QPE and added NOHRSC snow analysis. In Alaska, URMA inputs fed into the AORC system to improve downscaling of temperature, humidity, pressure, and wind products.

Key use cases include: forecast verification and skill score benchmarks, operational dashboarding, bias-corrected model outputs, and generation of high-quality labels for machine learning and post-processing pipelines.

Detail

• Product page: https://www.nco.ncep.noaa.gov/pmb/products/rtma/
• Key Features:
  • Hourly near-surface analyses; URMA updates after additional observations arrive (time-lagged rerun).
  • Designed as the analysis-of-record for verification and post-processing (e.g., NBM calibration).
  • Delivered on NDFD Lambert grids: 2.5 km (CONUS), 3 km (Alaska), 2.5 km (HI/PR/Guam).
• Resolution: 2.5 km.
• Temporal Coverage: Hourly. Data available from 2019-present.
• Primary Use Cases: Forecast verification, bias correction, backtesting & ML training, operational dashboards.

Parameters: