Difficulties in measuring remote ephemeral systems: A long term comparison of weir derived and manually measured discharge in the Canadian Rocky Mountains.

Ephemeral stream networks represent the majority of stream reach length, yet are the focus of relatively fewer field studies. The intermittent and sometimes flashy nature of many ephemeral streams make them difficult to gauge. There is often only a brief window to measure an adequate range of flows to develop a stage discharge relationship. The highest flows are infrequent and difficult to measure which leads to uncertainty in accurate peak flow predictions using rating curve methods. Weir installation in ephemeral streams can be advantageous because it reduces field work in remote sites, and captures the full range of discharge in real time. Most weir focused literature explores weir design in laboratory settings and implementation in urban/industrial environments; less literature has demonstrated the difficulties and tradeoffs of applying weirs to remote natural watersheds. Moreover, it is uncommon to find direct comparisons of weir derived and manually measured discharge over a long time period. 
Here we compare 13 years of paired weir-derived and measured discharge data in an ephemeral, snow-melt dominated catchment in the southern Canadian Rocky Mountains. Errors associated with the weir routinely occurred during the start of the melt freshet and post-cessation dates. Freezing in the weir reservoir created uncertainty in determining early season discharge, and small post-cessation events were not measured if the weir reservoir did not fill enough to reach the bottom of the weir outflow. Furthermore, the weir consistently underestimated discharge relative to paired manually measured discharge, which suggests significant rates of leakage from the weir. On average, 37% of the measured discharge was not accounted for by the weir (rate of leakage). Additionally, the rate of leakage was not constant in time. Rates of leakage varied seasonally and interannually (annual averages ranged from 18%-55% of measured discharge) and were positively, but only weakly correlated with discharge. This suggests several environmental factors could impact leakage rates. Using the measured discharge points to scale the weir data reduced the overall average discrepancy between weir and measured discharge to 4%. Once applied across the discharge record, the leakage scaling correction resulted in an increase in average annual water yield of 80mm across 13 years. These data suggest that weir leakage in natural systems may be a significant source of error and highlights the importance of regular field calibration.