Update: Spurious trace precipitation in Primary Precipitation data product
January 2, 2024
A long-standing issue with NEON's weighing gauge precipitation sensors is frequent spurious measurements of trace precipitation. Data from the weighing gauges are published in DP1.00006.001 along with precipitation measurements collected by secondary and throughfall tipping buckets, which are unaffected. Data from the weighing gauges are published with "PRIPRE" in the file name.
Daily sums of spurious precipitation range from \< 0.05 mm/day to > 0.5 mm/day, with higher sums generally corresponding to sites and time periods of greater evaporative demand. The underlying cause of the spurious trace precipitation is sensor noise and susceptibility to environmental variation. NEON scientists are currently working on a new algorithm designed to be far less sensitive to these issues. The new algorithm is expected to be released in mid-late 2025 and all previously collected data will be reprocessed. Additionally, NEON is in the process of acquiring a new weighing gauge precipitation sensor to reduce or eliminate this issue entirely.
In the meantime, modifications to the existing algorithm have been made to make the noise level of the sensor more obvious (to better distinguish actual precipitation events), reduce bias in long-term precipitation totals, and improve the quality flagging to better indicate spurious measurements. Specifically, negative values are no longer removed when computing 5-minute and 30-minute precipitation sums and data are quality flagged when more than 10% of the raw precipitation measurements over the sum interval were negative. All provisional data have been updated with this change and the full data record will be updated in RELEASE-2024 (planned for late January 2024).
Users are advised to filter data using the final quality flag and to apply additional filtering and review as needed to ascertain the suitability of the data for the intended use. The updated quality flagging is largely successful at identifying spurious trace precipitation. However, actual precipitation events that are short in duration or fall within the noise range of the sensor are likely to be marked as spurious, and some spurious trace precipitation remains unflagged.
An update on this issue will be provided when the new algorithm is released, the new sensors are installed, or by the end of 2025, whichever is earlier. For more detailed information on the problem please Contact Us. Select DP1.00006.001 in the Data Product dropdown list.
