Effects of sampling strategies and estimation algorithms on nutrient load determination in small agricultural headwater watersheds

Authors: YING, LI - Chinese Academy Of Agricultural Sciences; HAW, YEN - Texas A&M University; Harmel, Daren; QIULIANG, LEI - Chinese Academy Of Agricultural Sciences; JIAOGEN, ZHOU - Huaiyin Normal University; WANLI, HU - Yunnan Academy Of Agriculture Sciences; WENCHAO, LI - Chinese Academy Of Agricultural Sciences; LIMEI, ZHAI - Chinese Academy Of Agricultural Sciences; HONGYUAN, WANG - Chinese Academy Of Agricultural Sciences; WEIWEN, QIU - New Zealand Institute Of Plant & Food Research; HUIZHONG, LI - Liaoning University; JIAFA, LUO - Agresearch; SHUXIA, WU - Chinese Academy Of Agricultural Sciences; HONGBIN, LIU - Chinese Academy Of Agricultural Sciences; XIAOHONG, LI - Chinese Academy Of Agricultural Sciences

Submitted to: Journal of Hydrology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/4/2019
Publication Date: 9/9/2019
Citation: Ying, L., Haw, Y., Harmel, R.D., Qiuliang, L., Jiaogen, Z., Wanli, H., Wenchao, L., Limei, Z., Hongyuan, W., Weiwen, Q., Huizhong, L., Jiafa, L., Shuxia, W., Hongbin, L., Xiaohong, L. 2019. Effects of sampling strategies and estimation algorithms on nutrient load determination in small agricultural headwater watersheds. Journal of Hydrology. https://doi.org/10.1016/j.jhydrol.2019.124114.
DOI: https://doi.org/10.1016/j.jhydrol.2019.124114

Interpretive Summary: In recent decades, numerous studies have been conducted to explore the effects of anthropogenic activities on the natural environment and to inform decision makers on appropriate actions and policies. One such type of assessment is constituent flux estimation. Accurate and precise estimates of constituent flux in streams and rivers are important; however, the error associated with water quality sampling and load estimation has received limited investigation. In this study, the error associated with sampling frequency (3-day, 1-week, 2-week, 4-week, 6-week, 8-week) and load estimation algorithms was examined. Seasonal and annual nutrient loads were estimated in a small headwater stream dominated by groundwater contribution to baseflow. Total nitrogen load (including both dissolved nitrogen and particulate nitrogen) was chosen to reflect water quality parameters, and the accuracy (percent bias), precision (standard deviations), and performance (root-mean-square error) were analyzed for each sampling frequency and load estimation algorithm. As expected, substantial error resulted when using infrequent sampling to estimate total nitrogen loads. Statistical evaluation of differences in error in summer cannot be properly addressed because of the high discharge in storm events. Results indicated that biweekly sampling frequencies load estimation algorithms D (Annual flow multiplied by mean concentration of samples) and F (Linear interpolation of concentrations multiplied by flow) could be sufficient for evaluating the characters of annual constituent load in Fengyu River Watershed. However, high-flow sampling is also required to accurately analyze hydrological variations or seasonal characteristics. Utilizing short-term weather forecasts is suggested to conduct sampling targeted to capture rainfall-runoff events producing high flow. These results can serve as a useful reference for monitoring scenarios in watersheds with similar hydrological characteristics.

Technical Abstract: In recent decades, numerous studies have been conducted to explore the effects of human activities on the natural environment and to inform decision makers on appropriate actions and policies. One such type of assessment is the estimation of nutrient transport in streams and rivers. Accurate and precise estimates of nutrient transport in streams and rivers are important; however, the error associated with water quality sampling and load (mass) estimation has received limited investigation. In this study, the error associated with sampling frequency (3-day, 1-week, 2-week, 4-week, 6-week, 8-week) and load estimation algorithms was examined. Seasonal and annual nutrient loads were estimated in a small headwater stream dominated by groundwater contribution to baseflow. Total nitrogen load (including both dissolved nitrogen and particulate nitrogen) was chosen to reflect water quality parameters, and the accuracy, precision, and statistical performance were analyzed for each sampling frequency and load estimation algorithm. As expected, substantial error resulted when using infrequent sampling to estimate total nitrogen loads. Statistical evaluation of differences in error in summer cannot be properly addressed because of the high discharge in storm events. Results indicated that biweekly sampling frequencies load estimation algorithms D (Annual flow multiplied by mean concentration of samples) and F (Linear interpolation of concentrations multiplied by flow) could be sufficient for evaluating the characters of annual constituent load in Fengyu River Watershed. However, high-flow sampling is also required to accurately analyze hydrological variations or seasonal characteristics. Utilizing short-term weather forecasts is suggested to conduct sampling targeted to capture rainfall-runoff events producing high flow. These results can serve as a useful reference for monitoring scenarios in watersheds with similar hydrological characteristics.