Quality control development for near real-time rain gauge networks for operational rainfall monitoring

Abstract

Water, especially fresh water, plays an unmissable role in nature, agricultural activities or as drinking water. How much precipitation actually reaches the ground is therefore important to monitor. The number of official rain gauges is however limited. If present, weather radars are able to indicate rainfall with a high spatial and temporal resolution but do not grasp actual rainfall amounts. Therefore, in recent years methods are proposed to automatically quality control (QC) measurements of third party near real- time rain gauge networks by comparison of a measurement to measurements obtained nearby. Here, a method is proposed that includes weather radar observations to quality control an unprecedented large amount of measurements of rain gauges employed by Royal Netherlands Meteorological Institute (KNMI) and third parties as water boards and Personal Weather Stations (PWSs). Results show that the method is able to improve data quality of various real-time rain gauge networks. For PWS data of Netatmo collected in Amsterdam in 2018, the data set quality measured for 5-min rainfall accumulations is improved to a relative bias of 0.06 (-0.11 originally), a coefficient of variation of 6.8 (53.2 originally) and a Pearson correlation coefficient of 0.59 (0.07 originally) by removing 27.3% of all data. Application of the existing method however filters less data while yielding the same results. For water boards and KNMI data, this existing method cannot be applied due to the lack of nearby rain gauges. The newly proposed method however is able to improve all data sets, despite all differences in setup, maintenance and intrinsic quality. It is furthermore shown that with parameter optimization, it is able to filter 20%-point less measurements without quality loss, hence showing the wide applicability of the QC method if real-time weather radar data is present. Application of the proposed method on a national scale for 1 year of Netatmo data, shows that quality controlled third party data can be used as viable source for operational rainfall monitoring.