In situ UV-Vis sensors record spatially and temporally different diurnal patterns in solute concentrations in tropical montane headwater streams: reality or artefact?

There is a need to better understand the driving factors of and interactions between hydro-biogeochemical processes at different temporal and spatial scales to inform water management. Especially for tropical montane ecosystems, which are recognized as important freshwater sources, the required data and knowledge are missing. To address this knowledge gap, a long-term hydrological monitoring network was set up in the Mau Forest Complex, Kenya. The network covers three sub-catchments (27–36 km²) dominated by either tropical montane forest, smallholder agriculture or commercial tea plantations, within a 1021 km² mixed land use catchment. A 5-year dataset of nitrate and dissolved organic carbon concentrations measured at 10-minute interval with in situ UV-Vis sensors was analysed for short-term changes in solute concentrations. The analysis revealed small diurnal patterns (amplitude <0.25 mg N or C L⁻¹, decreasing with increasing discharge) in solute concentrations in all streams. In addition, the timing of the minima and maxima differed between catchments and seasons, suggesting and influence of land use and seasonality on the occurrence of in-stream biogeochemical processes. However, unusual and abrupt changes in the diurnal patterns were observed after a change in sensor position or exchange of sensors. We therefore developed an experiment to test the validity of the observed diurnal patterns. A second, mobile sensor was installed at each site for a period of more than three weeks. After measuring in parallel position to the fixed sensor for two weeks, the position (orientation, depth) was changed or the measurement gap was shaded. In parallel position, the patterns in solute concentrations recorded by the mobile sensor agreed better with those measured by the fixed sensor for dissolved organic carbon (r>0.98) than for nitrate (r=0.43–0.81). However, shading the sensor or a position change resulted in inconsistent changes in the recorded patterns. Larger changes in solute concentrations, e.g. as a result of rainfall events, were reproduced well by the mobile sensor. The results of our study suggest that diurnal changes in solute concentrations with an amplitude close sensor accuracy measured with in situ UV-Vis sensors should be interpreted with caution. The experiment was not conclusive as to what caused the differences in observed patterns. Further experimental work is required to understand the causes and to develop recommendations for the use of UV-Vis sensors in hydro-biogeochemical research.