Given the highly discontinuous temporal and spatial nature of precipitation, accurate observation of this quantity remains one of the more challenging tasks required of ground based in-situ networks. This is particularly true for near-real-time automated networks, where automated quality assurance (QA) methods are relied upon to detect and flag faulty data, within the most recently received observation time window. One step that can be taken to increase the reliability of the measurements is through redundancy. By increasing the number of independent measurements of the weight of the bucket, the continuity of the record can be improved with pair-wise inter-comparisons used to confirm observations. However, this also leads to a substantial increase in the complexity of calculating one precipitation amount from these independent observations. Hence, even well documented precipitation QA procedures need to be reexamined at times to understand both their limitations and effectiveness.

This presentation will describe recent studies of the US Climate Reference Network (USCRN) precipitation algorithm, including potential improvements that have been identified. Several precipitation QA algorithms were developed for testing with the USCRN precipitation gauge, the Geonor T200-B series with three vibrating wires. Algorithms were tested under a variety of atmospheric (light to heavy precipitation rates, light to moderate winds, seasonal changes) and instrumental (noisy wires, evaporation from gauge) conditions in both real-world environments and artificially generated precipitation events. Some of the most challenging factors to overcome were noisy wires, gauge evaporation, and extreme events/environments. While the current algorithm used by USCRN is quite effective, several recommendations have been generated for improving the methodology for calculating precipitation from raw gauge depth and wetness sensor inputs.