The spatial and temporal sampling errors inherent in low resolution radar estimates of rainfall

Abstract

ABSTRACT:The errors introduced into radar estimates of rainfall by making observations at spatial and temporal resolutions that are coarse compared with precipitation systems' characteristic length and time scales are explored in this study. High resolution (200 m, 50 s) X-band radar data from 48 mid-latitude precipitation events are downgraded progressively in spatial and temporal resolution so that estimates of this sampling error can be made by comparing 10 min rainfall accumulations of this data to accumulations calculated from the original high resolution data. The analysis shows these errors to be of significant magnitude. For 2 km and 5 min sampling, this error varies from 17 to 64% of the mean rainfall accumulation. A relationship is shown between the error introduced from the reduction in spatial resolution and the characteristic length scale of the precipitation system along with a metric of precipitation intensity.

X-band dual-polarized radar quantitative precipitation estimate analyses in the Midwestern United States

“…An average of 5-minute scans were used 152 for each of the variables, which were aggregated to hourly totals to be compared to the hourly tipping-153 bucket accumulations. In spite of previous reports suggesting 5 minute to hourly aggregates can have 154 significant effects on QPE (Fabry et al, 1994), evidence has been presented that overestimation in 155 accumulations may not exceed 26% for a pixel size of 1 km (Shucksmith et al, 2011). 156…”

A terrain‐based weighted random forests method for radar quantitative precipitation estimation

“…In China, the China Next Generation Weather Radar (CINRAD, similar to WSR‐88D) is widely used in operation and mostly relies on the radar reflectivity factor ( Z ) to measure rainfall rate ( R ) through empirical Z–R relationships. Through operational use, the sources of errors consist of: (1) Z bias including ground clutter, anomalous propagation, signal attenuation, beam blockage, bright band (BB) and variation of the vertical profile of reflectivity (VPR), (2) R bias including changes in the precipitation before reaching the ground and (3) inappropriate Z–R relationships (Villarini and Krajewski, ; Charba and Samplatsky, ; Shucksmith et al ., ; Erdin et al ., ; Lee et al ., ).…”

“…Through operational use, the sources of errors consist of: (1) Z bias including ground clutter, anomalous propagation, signal attenuation, beam blockage, bright band (BB) and variation of the vertical profile of reflectivity (VPR), (2) R bias including changes in the precipitation before reaching the ground and (3) inappropriate Z-R relationships (Villarini and Krajewski, 2010; Charba and * Correspondence: W. Zhang, Institute of Automation, Chinese Academy of Sciences, Beijing, China. E-mail: zhangwenshengia@hotmail.com Samplatsky, 2011;Shucksmith et al, 2011;Erdin et al, 2012;Lee et al, 2015).…”

Applicability of open rainfall data to event-scale urban rainfall-runoff modelling

“…The resolution requirements depend also on the storm spatial structure and therefore storm type (Emmanuel et al, 2012;Peleg et al, 2013;Shucksmith et al 2011). In a relatively homogeneous rain field, e.g.…”

“…Too coarse a gauge network may either miss the convective cells, or especially with small catchments, measure additional rain if the gauges are located outside the catchment. Radar measurements, on the other hand, suffer from sampling errors especially if the spatial resolution is coarser than the length scale of the rain feature (Shucksmith et al, 2011). This can lead to over-or underestimation of rainfall amounts.…”