Investigation of watershed nutrient export affected by extreme events and the corresponding sampling frequency

Although the real-time monitoring technique has been widely applied due to the improvement of sensors, development of traditional sampling methods is still worth of being discussed due to the economically feasibility. Currently, extreme events (e.g. extreme rainfall caused by climate change) play a relatively important role in nutrient export. However, impacts of extreme events on the optimization of sampling strategy is still not well addressed despite the uncertainty of different frequency sampling programs has been sufficiently discussed in previous studies. Therefore, the corresponding impact of extreme events impact on the optimization of sampling strategy was investigated by examining temporal (i.e., inter-annual and seasonal) variations of available data. Uncertainty of nutrient flux estimates under different sampling frequencies was explored by subsampling daily monitoring data. Results showed that uncertainty in flux estimates differed between nitrogen and phosphorus. The relative error (RE) in annual TN flux estimates ranged from -4.2% to 2.4% (once per three-day) to -21.4-31.1% (monthly sampling), while the RE in annual TP flux estimates varied from -14.1% to 8.2% (once per three-day) to -65.9%-163.4% (monthly sampling). Biweekly and weekly sampling routines are considered the optimal sampling program for total nitrogen (TN) and for total phosphorus (TP) when the extreme events impact were not been considered. The uncertainty of flux estimates with different sampling frequencies increased with the increasing extreme events. High level of uncertainty occurred in years with the most extreme events in 2012 (RE: 21.4-69.0% for TN, 33.3-96.6% for TP), while the lowest can be found in 2011 (RE: 0-20.7% for TN, 0-48.3% for TP) (with the fewest extreme events). In addition, uncertainty in TN and TP flux estimates was generally greater during summer season than during other seasons. These results highlighted the important role of extreme events in nutrient export. Approximately half of the annual TN and TP flux occurred in some extreme days that only accounted for less than 20% in the same year. The onset of these extremes of nutrient export was likely due to the stormflow with addition of external fertilizer and the direct discharge of surface ponding water from paddy fields during special periods of time. These results would be helpful for the optimization of sampling strategy.