A proposed cross-site comparison of missing data in long-term hydrologic datasets
All long-term datasets contain gaps in the record. When calculating solute fluxes it is not possible to simply omit missing values of volume or concentration, so they must be estimated. The uncertainty in these estimates is rarely reported or propagated into flux estimates. We are proposing a study to characterize the causes and magnitude of gaps in precipitation and stream flow datasets, with the ultimate goal of identifying the greatest sources of uncertainty to help inform monitoring practices. Our preliminary investigation includes datasets from four LTER sites (HBR, CWT, HJA, SEV) and the Wakayama Experimental Forest in Japan. When calculating deposition, we found that gaps in the chemistry datasets were more common than volume gaps, and the leading cause across sites was contamination from bird droppings. Conversely, the magnitude and causes of gaps in the stream flow record varied across sites, however more gaps were caused by equipment (dataloggers, clocks, and chart recorders) than routine maintenance. We hope to use the LTER All Scientists meeting as a platform to solicit contributions from researchers at other sites to help broaden this cross site comparison.