A Statistical Analysis of Radar and MRMS QPE in the Northern Plains and Mid-Atlantic
Loading...
Authors
Steen, Matthew
Date
2023-05
Type
Thesis
Language
en
Keywords
Alternative Title
Abstract
ABSTRACT
A Statistical Analysis of Radar and MRMS QPE in the Northern Plains and Mid-Atlantic
by
Matthew C. Steen
Plymouth State University, May, 2023
Over the last 40 years weather radar has provided an immense amount of data
across the United States. Radar precipitation estimates provide a much higher spatial and
temporal density of observations than ground-based measurements. In recent years, dual-
polarization capabilities introduced to NEXRAD WSR-88Ds provided improvements to
precipitation estimations allowing for more accurate forecasts and warnings. Also, the
Multi-Radar/Multi-Sensor System (MRMS) has developed a set of hydrometeorological
based algorithms which take in data from radars, satellites, surface and upper air
observations, models, and lightning detection systems to create a wide array of products
that assist in decision-making and provide improved weather forecasting tools. The
MRMS and Dual-pol radar Quantitative Precipitation Estimations (QPEs) are being used
more by analysts in place of the previous radar precipitation estimation algorithm, the
Precipitation Processing System (PPS). The goal of this research is to compare the Sioux
Falls, SD (KFSD) and Dover, DE (KDOX) WSR-88Ds dual-polarized radar rainfall
estimates and MRMS radar-only rainfall product estimates to rain gauge measurements
for precipitation events in these regions. KFSD was selected because there have been few
radar QPE studies in the Northern Plains. KDOX was selected for similar reasons (few
studies in the mid-Atlantic) along with providing a perspective from a different climatic
xi
regime within the United States. Hourly rain-gauge precipitation estimates within 100 km
of KFSD for events with at least one hour of observed rainfall greater than or equal to
five millimeters were analyzed. These observed values were compared with high and low
resolution dual-polarization QPE and MRMS radar-only estimates.
Results demonstrate that the MRMS radar-only product produced better
precipitation estimates than both high and low resolution dual-polarized estimates at both
KFSD and KDOX. The mean absolute error (MAE) for MRMS was lower than both
radar products at KFSD while the high-resolution radar product produced a similar MAE
at KDOX. The differences in MAE between the low resolution dual-pol QPE and the
MRMS and the high resolution dual-pol QPEs were determined to be statistically
significant at both stations. The average bias of the MRMS was lower than both radar
products for both radars, with a larger difference between the MRMS and low resolution
dual-pol QPE than high resolution estimates. Like MAE, the difference between the low
resolution dual-pol QPE and the MRMS and the high resolution dual-pol QPEs were
determined to be statistically significant at both stations. Based on the results of this
study, forecasters may be more inclined to favor the estimations of MRMS-based
products for the forecasting of rainfall and issuing of rainfall related watches and
warnings.