Using in situ UV-visible spectrophotometer sensors to predict phosphorus species concentrations in Lake Champlain tributaries

Accurate riverine phosphorus concentration measurements and load estimates are critical to meeting water quality targets in the Lake Champlain Basin. Currently, total phosphorus load estimates rely on proxy variables such as turbidity and discharge, and little is known about important phosphorus fraction dynamics for dissolved and soluble reactive phosphorus that are more readily available forms for cyanobacteria uptake. Optical water quality sensors can make rapid sub-hourly measurements and have been shown to reduce uncertainty in load estimates and reveal high-frequency storm dynamics for solutes such as nitrate and dissolved organic carbon. We tested the utility of in situ UV-Visible spectrophotometers to predict total, dissolved, and soluble reactive phosphorus concentrations in streams draining agricultural, urban, and forested land use / land covers in the Lake Champlain Basin. We present the first statistically robust validation technique applied to optical water quality sensors in order to emulate how the sensors may perform in predicting unknown phosphorus species concentrations. Models to predict dissolved and soluble reactive phosphorus explained a greater portion of the variance than any other known proxy variable technique, and results varied by land use / land cover. Total phosphorus predictions from UV-Visible spectra were most accurate when data from all sites were combined, and the amount of variance explained was similar to results of others that rely on discharge and turbidity. This monitoring approach is unique in its ability to predict not only total phosphorus concentrations with reasonable accuracy, but also concurrent dissolved phosphorus fraction concentrations in addition to concurrent nitrate and dissolved organic carbon concentrations, and turbidity. This technology may be particularly useful for researchers or managers that aim to quantify and monitor the chemical partitioning of phosphorus species at a high-frequency. While these next-generation sensors are currently expensive, this technology is evolving rapidly. In the foreseeable future this type of instrumentation may extend our ability to monitor critical nutrients at times and places that would be difficult to sample in any other way. However, these instruments present new challenges for maintenance, calibration, and data management. We will discuss these benefits and challenges.