Peer-Reviewed Article
Assessment of Uncertainty in Doppler- Radar Estimates of Precipitation for Use in Geoscience Studies
Authors
T. Walsh, Instructor, Department of Mathematical Sciences, USMA , M. Scioletti Asst. Professor, Department of Mathematical Sciences, USMA, P. Rao Professor, Department of Civil and Environmental Engineering, California State University, Fullerton, T.V. Hromadka II, AS-0020, Professor, Department of Mathematical Sciences, USMA; H. McInvale, Professor, Department of Mathematical Sciences, USMA
Abstract
Doppler radar data forms the underpinnings of various applications in hydrometeorology, engineering, floodplain man- agement, and weather forecasting, (among other uses) necessitating the importance of scrutinizing its accuracy, which depends on the accuracy of measured precipitation estimates obtained from gaged monitoring sites. This article explores the collective use of the WRS-88D Doppler radar system, given its long history, from the assemblage of several thousands of published data pairs of Doppler radar precipitation estimates with actual rain gauge precipitation gauge readings. De- tailed statistical analysis of these data pairs shows that the evaluation of the uncertainty in the Doppler radar estimated precipitation can be accomplished using standard techniques, and the display of the computational results can be com- municated using scatter plot visualization techniques readily available. The resulting distributions depict the degree of uncertainty associated with Doppler radar estimates of precipitation.
Introduction
Weather radars are playing an important role in predict- ing precipitation characteristics. The Weather Surveillance Radar (WSR-88D) is a Doppler Radar first introduced in 1988. This is the usual name for the 159 high resolution S-band Doppler weather radars which are part of the NEXRAD (Next Generation Radar) network, and are operated by the National Weather Service. The WSR-88D radar operates by sending and receiving microwave pulses, in the 2-4 GHz range, known as S band. During 1988-2013, many researchers quantified the performance of Doppler Radars by comparing the Doppler radar derived rainfall with the associated relevant gauge observations (considered the “bench mark” data). These com- parison studies highlight factors that can affect the reliability of Doppler predictions, including the often used ZR power law relationship, radar miscalibration, signal attenuation and range effect, among others.
Focusing specifically on the data accumulated by the WSR- 88D Doppler Radar system, (prior to the completion of the system upgrades to Dual Polarization by 2013), of particular interest is the comparison between the reported precipitation gauge readings and the related Doppler radar estimate of precipitation. In this analysis, published literature in cited references 1-10 contains the data in the form of scatter plots and tables. The data compares the Doppler-radar-derived rainfall estimates with the observed local gauge values, spread across multiple storms and geographical domains with the overwhelming majority categorized via total storm accumu- lation. We used digitizing software to read the graphs and
tabulate the data in each reference for later concatenating. 22 TPG •
Jan.Feb.Mar 2019
Method
The raw data file consists of two columns of rainfall data; namely, Doppler Radar Estimated Precipitation (“DREP”) and Gauge Estimated Precipitation (“GEP”). The DREP column includes radar estimated values (in mm) from the Doppler WSR-88D equipment whereas the GEP column includes precipitation values (in mm) as measured by recording pre- cipitation gauges. Combining the two columns creates a set of ordered pairs resulting in 8846 ordered pairs for the subject Doppler data file.
Below, Table 1 summarizes the data characteristics for the Doppler Radar column. Based on the published graphs and/ or tables from the cited references, the compiled radar and gauge precipitation values in the current paper specifically focus on total storm accumulation Doppler Radar data for further analysis, as opposed to the other types of radar data available, e.g. Dual-Polarization data.
Table 1 - Summary of Doppler (WSR-88D) Data Characteristics
Figure 1 depicts the data in Table 1 in its raw form. Using the standard normalization technique to provide scalability, yi = (xix DREP and the GEP variables to produce a set of normalized data pairs. Figures 2 and 3 show the DREP and GEP as con-
www.aipg.org
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64
