May 26 - July 31
Evaluation and Uncertainty of MRMS v12 Dual-polarized Radar Quantitative Precipitation Estimation Product
Henry M. Santer and Heather M. Grams
What is already known:
- Quantitative precipitation estimation (QPE) is subject to errors and uncertainties from various sources.
- R(A)rainfall relationships, based on specific attenuation, have higher linearity with rainfall than other variables.
- QPEs tend to underestimate rainfall in areas farther from radar, and overestimate slightly when close to radar.
- The Multi-Radar/Multi-Sensor (MRMS) system creates a high-resolution QPE called Q3EVAP that can use R(A) relationships and applies an evaporation correction.
What this study adds:
- Comparison of Q3EVAP uncertainties to those of its predecessor, Q3RAD.
- Q3EVAP measures lower rain rates than Q3RAD, especially in areas farther from radar.
- Q3EVAP is generally a better and more linear fit to rainfall than Q3RAD.
- When Q3EVAP can use R(A) relationships, it shows significant improvement in estimating rain rates compared to Q3RAD.
- Both products tend toward overestimation under very warm conditions.
- Further analysis is required to separate uncertainties associated with the evaporation correction from those associated with using R(A) relationships.
Abstract:
Two radar-based quantitative precipitation estimation (QPE) products from the Multi-Radar/Multi-Sensor (MRMS) system are evaluated against hourly gauge-based rainfall accumulations. The two products are Q3RAD, the QPE deployed with the initial operating version of MRMS, and Q3EVAP, a new product that uses rain relationships based on specific attenuation A and applies an evaporation correction. The evaluation takes place over the entire conterminous United States (CONUS) over an 18-month period starting in August 2018. Regional analysis of both products suggests that Q3EVAP has a more linear relationship with rainfall than Q3RAD, and in fact is usually closer to the true hourly rainfall than Q3RAD. Q3EVAP mitigates a wet bias that Q3RAD tends to exhibit, but in turn is more prone to underestimation, particularly in areas of worse radar coverage. Both products tend toward overestimation under very warm conditions, but Q3EVAP limits the extent of this with the evaporation correction. For very high rain rates, both products underestimate signifi- cantly, but Q3EVAP consistently provides a similar or better estimate compared to Q3RAD. This is attributed primarily to higher linearity between specific attenuation and rainfall than with other radar variables. The results of this study should be useful to users wanting to understand Q3EVAP’s limitations and to developers in pursuit of probability-based QPE.