Lolli et al. (2025) Evaluating the NASA MPLNET Rain Masking Algorithm at Goddard Space Flight Center and Barcelona sites: Relevance to EarthCARE Cloud Profiling Radar Validation

Identification

• Journal: Atmospheric Research
• Year: 2025
• Date: 2025-10-10
• Authors: Simone Lolli, Jasper R. Lewis, Ali Tokay, Gemine Vivone, James R. Campbell, Erica K. Dolinar, Michaël Sicard, Adolfo Comerón, Alejandro Rodríguez-Gómez, Constantino Muñoz-Porcar, Albert García-Benadí, Mireia Udina, Joan Bech, Ellsworth J. Welton
• DOI: 10.1016/j.atmosres.2025.108535

Research Groups

• CNR-IMAA, Tito Scalo, Italy
• NASA Goddard Space Flight Center (GSFC), Greenbelt, MD, USA
• GESTAR II, Greenbelt, MD, USA
• Naval Research Laboratory, Monterey, CA, USA
• Universitat de Barcelona, Barcelona, Spain
• Universitat Politècnica de Catalunya (UPC), Barcelona, Spain
• SARTI, Universitat Politècnica de Catalunya, Vilanova i la Geltrú, Spain
• LACU, Université de La Réunion, Saint Denis, France

Short Summary

This study evaluates the NASA MPLNET Rain Masking Algorithm (RMA) for detecting rainfall and distinguishing it from non-rain events over multiple years at two distinct sites. The RMA is found to be highly effective, outperforming IMERG in sensitivity and accuracy, and uniquely capable of detecting virga, which is crucial for validating the upcoming EarthCARE Cloud Profiling Radar.

Objective

• To evaluate the performance of the Rain Masking Algorithm (RMA), developed for NASA’s Micropulse Lidar Network (MPLNET), in detecting rainfall events and distinguishing them from non-rain events over multiple years.
• To characterize how the rain occurrence frequency derived from RMA compares to coincident satellite retrievals from the Integrated Multi-Satellite Retrievals for GPM (IMERG) project.
• To present the performance of RMA and a corresponding matchup strategy intended to facilitate future validation efforts for the ESA-JAXA EarthCARE mission's precipitation data.

Study Configuration

• Spatial Scale: Two MPLNET sites: Goddard Space Flight Center (GSFC) in Greenbelt, MD, USA (38.99°N, 76.84°W) and Universitat Politècnica de Catalunya (UPC) in Barcelona, Spain (41.40°N, 2.12°E).
• Temporal Scale: Multi-year data. GSFC: February 2020 to May 2022 (discontinuous). UPC: January 2019 to December 2020 (complete years). Data were aggregated into 60-minute intervals for intercomparison.

Methodology and Data

• Models used:
  • Rain Masking Algorithm (RMA) (V3 MPLNET rain detection algorithm).
  • Integrated Multi-Satellite Retrievals for GPM (IMERG) Version 06, Final run.
• Data sources:
  • Ground-based lidar: NASA Micropulse Lidar Network (MPLNET) lidars (elastic MPL systems) providing 1-minute temporal resolution and 30-meter to 75-meter vertical resolution from the surface up to 30 kilometers altitude.
  • Ground-based disdrometer: Parsivel2 disdrometers (co-located with MPLNET lidars) with a 60-second sampling period, measuring raindrop sizes from approximately 0.2 millimeters to 25 millimeters in diameter.
  • Satellite: Integrated Multi-Satellite Retrievals for GPM (IMERG) providing precipitation estimates at 0.1° × 0.1° spatial resolution and 30-minute temporal resolution.

Main Results

• The RMA is highly effective at detecting rain events, including virga (precipitation that evaporates before reaching the ground), which IMERG often misses.
• GSFC Site Intercomparison (RMA vs. IMERG, disdrometer as reference):
  • Accuracy: RMA 0.72, IMERG 0.66.
  • Sensitivity (Recall): RMA 0.63, IMERG 0.40.
  • Specificity: RMA 0.81, IMERG 0.91.
  • Precision: RMA 0.78, IMERG 0.83.
  • F1-score: RMA 0.69, IMERG 0.54.
  • Matthews Correlation Coefficient (MCC): RMA 0.46, IMERG 0.37.
  • Kappa: RMA 0.41, IMERG 0.29.
  • RMA generally shows superior performance with a more balanced trade-off between precision and recall, and better overall agreement with ground truth. IMERG is more effective at minimizing false positives.
  • IMERG exhibits a significantly higher frequency of undetected rain, especially at lower rain intensities (0 to 1.39 micrometers per second).
• UPC Barcelona Site Intercomparison (RMA vs. IMERG, disdrometer as reference):
  • Accuracy: RMA 0.71, IMERG 0.67.
  • Sensitivity (Recall): RMA 0.46, IMERG 0.49.
  • Specificity: RMA 0.95, IMERG 0.86.
  • Precision: RMA 0.89, IMERG 0.78.
  • F1-score: RMA 0.61, IMERG 0.60.
  • Matthews Correlation Coefficient (MCC): RMA 0.46, IMERG 0.38.
  • Kappa: RMA 0.41, IMERG 0.35.
  • RMA offers better specificity, precision, and overall reliability, making it more suitable for scenarios where avoiding false positives is crucial. IMERG has higher sensitivity but also increased false positives.
  • RMA misses slightly more precipitation events between 0 and 0.833 micrometers per second compared to IMERG.
• RMA Limitations:
  • Struggles to detect low-intensity precipitation (e.g., below 0.694 micrometers per second) where the Volume Depolarization Ratio (VDR) does not exceed the 0.06 threshold.
  • Can occasionally produce false positives due to transient atmospheric artifacts or misclassification of dust aerosols as rain.
  • May misclassify precipitation as clouds, leading to episodic detection.

Contributions

• Provides a robust validation of the NASA MPLNET Rain Masking Algorithm (RMA) against co-located disdrometers and IMERG satellite data, demonstrating its effectiveness across different geographical locations.
• Highlights the unique capability of RMA to detect virga, which is often missed by satellite observations, thereby contributing to a more comprehensive understanding of atmospheric water content and the hydrological cycle.
• Establishes a foundational ground-based validation protocol and performance benchmark for future satellite precipitation missions, specifically for the ESA-JAXA EarthCARE Cloud Profiling Radar.
• Advances global precipitation monitoring and refines meteorological and climate forecasting accuracy by offering a validated ground-based lidar algorithm.

Funding

• NASA Earth Observing System
• NASA Radiation Sciences Program
• NASA Earth Science U.S. Participating Investigator program (Grant No. 80NSSC21K0560)

Citation

@article{Lolli2025Evaluating,
  author = {Lolli, Simone and Lewis, Jasper R. and Tokay, Ali and Vivone, Gemine and Campbell, James R. and Dolinar, Erica K. and Sicard, Michaël and Comerón, Adolfo and Rodríguez-Gómez, Alejandro and Muñoz-Porcar, Constantino and García-Benadí, Albert and Udina, Mireia and Bech, Joan and Welton, Ellsworth J.},
  title = {Evaluating the NASA MPLNET Rain Masking Algorithm at Goddard Space Flight Center and Barcelona sites: Relevance to EarthCARE Cloud Profiling Radar Validation},
  journal = {Atmospheric Research},
  year = {2025},
  doi = {10.1016/j.atmosres.2025.108535},
  url = {https://doi.org/10.1016/j.atmosres.2025.108535}
}