No abstract

Although water is involved in many ecosystem services, the absence of monitoring data restricts the development of effective water management strategies especially in remote regions. Traditional monitoring networks can be expensive, with unaffordable costs in many low-income countries. Involving citizens in monitoring through crowdsourcing has the potential to reduce these costs but remains uncommon in hydrology. This study evaluates the quality and quantity of data generated by citizens in a remote Kenyan basin and assesses whether crowdsourcing is a suitable method to overcome data scarcity. We installed thirteen water level gauges equipped with signboards explaining the monitoring process to passers-by. Results were sent via a text-message-based data collection framework that included an immediate feedback to citizens. A public web interface was used to visualize the data. Within the first year, 124 citizens reported 1175 valid measurements. We identified thirteen citizens as active observers providing more than ten measurements, whereas 57% only sent one record. A comparison between the crowdsourced water level data and an automatic gauging station revealed high data quality. The results of this study indicate that citizens can provide water level data of sufficient quality and with high temporal resolution.

Abstract. Conversion of natural forest (NF) to other land uses could lead to significant changes in catchment hydrology, but the nature of these changes has been insufficiently investigated in tropical montane catchments, especially in Africa. To address this knowledge gap, we aimed to identify stream water (RV) sources and flow paths in three tropical montane sub-catchments (27–36 km2) with different land use (natural forest, NF; smallholder agriculture, SHA; and commercial tea and tree plantations, TTP) within a 1021 km2 catchment in the Mau Forest complex, Kenya. Weekly samples were collected from stream water, precipitation (PC) and mobile soil water for 75 weeks and analysed for stable isotopes of water (δ2H and δ18O) for mean transit time (MTT) estimation with two lumped parameter models (gamma model, GM; and exponential piston flow model, EPM) and for the calculation of the young water fraction. Weekly samples from stream water and potential endmembers were collected over a period of 55 weeks and analysed for Li, Na, Mg, K, Rb, Sr and Ba for endmember mixing analysis (EMMA). Solute concentrations in precipitation were lower than in stream water in all catchments (p < 0.05), whereas concentrations in springs, shallow wells and wetlands were generally more similar to stream water. The stream water isotope signal was considerably damped compared to the isotope signal in precipitation. Mean transit time analysis suggested long transit times for stream water (up to 4 years) in the three sub-catchments, but model efficiencies were very low. The young water fraction ranged from 13 % in the smallholder agriculture sub-catchment to 15 % in the tea plantation sub-catchment. Mean transit times of mobile soil water ranged from 3.2–3.3 weeks in forest soils and 4.5–7.9 weeks in pasture soils at 15 cm depth to 10.4–10.8 weeks in pasture soils at 50 cm depth. The contribution of springs and wetlands to stream discharge increased from a median of 16.5 (95 % confidence interval: 11.3–22.9), 2.1 (−3.0–24.2) and 50.2 (30.5–65.5) % during low flow to 20.7 (15.2–34.7), 53.0 (23.0–91.3) and 69.4 (43.0–123.9) % during high flow in the natural forest, smallholder agriculture and tea plantation sub-catchments, respectively. Our results indicate that groundwater is an important component of stream water, irrespective of land use. The results further suggest that the selected transit time models and tracers might not be appropriate in tropical catchments with highly damped stream water isotope signatures. A more in-depth investigation of the discharge dependence of the young water fraction and transit time estimation using other tracers, such as tritium, could therefore shed more light on potential land use effects on the hydrological behaviour of tropical montane catchments.

Land use change alters nitrate (NO3‐N) dynamics in stream water by changing nitrogen cycling, nutrient inputs, uptake and hydrological flow paths. There is little empirical evidence of these processes for East Africa. We collected a unique 2 year high‐resolution data set to assess the effects of land use (i.e., natural forest, smallholder agriculture and commercial tea plantations) on NO3‐N dynamics in three subcatchments within a headwater catchment in the Mau Forest Complex, Kenya's largest tropical montane forest. The natural forest subcatchment had the lowest NO3‐N concentrations (0.44 ± 0.043 mg N L−1) with no seasonal variation. NO3‐N concentrations in the smallholder agriculture (1.09 ± 0.11 mg N L−1) and tea plantation (2.13 ± 0.19 mg N L−1) subcatchments closely followed discharge patterns, indicating mobilization of NO3‐N during the rainy seasons. Hysteresis patterns of rainfall events indicate a shift from subsurface flow in the natural forest to surface runoff in agricultural subcatchments. Distinct peaks in NO3‐N concentrations were observed during rainfall events after a longer dry period in the forest and tea subcatchments. The high‐resolution data set enabled us to identify differences in NO3‐N transport of catchments under different land use, such as enhanced NO3‐N inputs to the stream during the rainy season and higher annual export in agricultural subcatchments (4.9 ± 0.3 to 12.0 ± 0.8 kg N ha−1 yr−1) than in natural forest (2.6 ± 0.2 kg N ha−1 yr−1). This emphasizes the usefulness of our monitoring approach to improve the understanding of land use effects on riverine N exports in tropical landscapes, but also the need to apply such methods in other regions.