Smarter Radar, Better Forecasts: A More Accurate Way to Measure Rainfall

At a Glance

Researchers at Colorado State University have developed an improved dual-polarization rainfall estimation method based on region-based hydrometeor classification. This approach uses spatial patterns in radar data to produce more stable and accurate rainfall measurements, reducing noise, and improving reliability even when data quality is low. Testing shows outperforming both single-polarization and dual-polarization techniques. The system is designed for real-world, operational weather monitoring.

Background

Dual-polarization radar improves rainfall estimation by capturing more detailed information about raindrop size and type. However, current methods often rely on pixel-by-pixel classification, which can become unstable in noisy conditions and struggle with mixed-phase precipitation, such as melting snow. These limitations reduce accuracy and reliability in operational forecasting. A more robust approach that accounts for spatial relationships in radar data is needed to improve performance.

Overview

This technology introduces a region-based hydrometeor classification framework that incorporates spatial coherence and self-aggregation of dual-polarization radar signals. Instead of analyzing each radar point independently, the method evaluates neighboring data collectively, improving classification stability and reducing noise sensitivity. This allows for more consistent identification of precipitation types, which directly enhances rainfall estimation accuracy.

The system follows the standard rainfall estimation workflow of quality control, hydrometeor classification, and precipitation quantification, but improves the classification stage by leveraging spatial context. It also mitigates common issues such as brightband contamination in the melting layer, a known source of error in radar-based rainfall measurement. Validation using NPOL radar data demonstrates that this method provides significantly better performance than traditional fuzzy-logic-based dual-polarization approaches and single-polarization Z-R relationships, making it well-suited for operational deployment.

Figure 1: This figure compares four rainfall estimation methods and shows that DROPS2.0 has the lowest Normalized Mean Absolute Error (NMAE) across all time periods, the most stable and smallest increase in Root-Mean-Square Error (RMSE) as rainfall accumulates, and the highest Pearson Correlation Coefficient (CORR) with real-world measurements. This means the new DROPS2.0 method gives more accurate rainfall estimates, stays reliable as more data is collected over time, and matches real-world measurements better than the other methods, especially over longer periods of time.

Benefits

More robust performance in noisy or low-quality radar data conditions
Improved rainfall estimation accuracy compared to existing methods
Utilizes spatial relationships in radar data for more stable classification
Reduces errors from mixed-phase precipitation and brightband effects
Demonstrated effectiveness using real-world radar datasets (NPOL)
Compatible with existing operational dual-polarization radar systems

Applications

Weather forecasting and real-time precipitation monitoring
Aviation weather systems and flight safety planning
Flood prediction and water resource management
Climate monitoring and hydrological modeling
Integration into commercial weather analytics platforms

Publications

H. Chen, et al (2017) “An Improved Dual-Polarization Radar Rainfall Algorithm (DROPS2.0): Application in NASA IFloodS Field Campaign” J. Hydrometeor., 18, 917–937, https://doi.org/10.1175/JHM-D-16-0124.1.

Last Updated: June 2026

Opportunity

Available for Non-Exclusive Licensing

IP Status

Inventors

V. Chandrasekar
Haonan Chen
Renzo Bechini

Reference Number

17-053

Licensing Manager

Aly Hoeher
Aly.Hoeher@colostate.edu
970-491-7100