J2.1
Quality assurance of PACRAIN tipping bucket gauge data
![](images/flag.png)
![](images/blue_ribbon.png)
Quality assurance (QA) is an important part of maintaining any observation network. QA methods are essential to identify both isolated bad data and systematic errors that indicate problems with a specific observation site. The layout of the PACRAIN/PI-GCOS network and the nature of rainfall in general create a challenging QA problem. QA methods that rely on correlation in time and space to identify outliers are of limited use. Rainfall is a highly discrete process that can vary by orders of magnitude over small time and distance scales. This is compounded by the geography of the Pacific, where rainfall gauges can be hundreds of kilometers apart and orographic effects can eliminate any correlation between even nearby gauges.
Several QA strategies need to be utilized. A limited number of sites have coincident gauges--a manual-read gauge and TBG in the same location with overlapping periods of record. Many techniques can be used to identify outliers in these cases. Some sites have collocated gauges--a manual gauge and TBG in the same location but not operational at the same time. For these sites the TBG data should be statistically similar to the manual data, although such comparisons must account for any long-term trends in rainfall distribution at that location. Most of the sites in the network are isolated gauges for which there are no data to compare against. High-resolution satellite rainfall estimates are of some value, but their utility is limited by the highly localized nature of rainfall. Techniques that can be used with data from a single TBG are being examined for their suitability to the PACRAIN TBG dataset.