View source for Naizin or Kervidy-Naizin and Kerbernez, Brittany, France 1975-

== History ==
Hydrological settings
The current observation protocol of AgrHyS consists of long-term records of meteorological, hydrological, and hydro-chemical variables and land cover maps. The longest time series started in 1993. Since that date, the Kerbernez and Kervidy-Naizin sites have been equipped with a weather station that records hourly rainfall, air and soil temperatures, air humidity, global radiation, and wind direction and speed, which allows the calculation of daily Penman evapotranspiration. Between 1995 and 1997, the outlets were equipped with U-notch or V-notch weirs for monitoring the stream water level. Most gauging stations are automated and use float-operated sensors that record stream level sub-hourly (from 1 to 10 min). Since 1999, the level sensors have been connected to a data logger. From 1997 to 2005, a network of piezometers was installed in the two sites to monitor shallow groundwater fluctuations and variations in its chemical composition. Currently, 31 piezometers are monitored on the two sites. The water table level is measured using automatic transducer sensors, sometimes coupled with temperature and electrical conductivity sensors that record water table data every 15 min in most piezometers (and manually in the others every month).

Various elemental concentrations are analyzed in the water depending on the station: major anions (Cl, NO2, NO3, SO4), major cations (Ca, Mg, K, Na), dissolved organic C (DOC) and dissolved inorganic C (sometimes coupled with Fe), total and soluble reactive phosphorus, and suspended solids. These concentrations are measured using grab sampling mostly. Various water quality parameters are also measured using sensors (electrical conductivity, temperature, turbidity, NO3 and DOC concentrations, and total reactive and total P concentrations). Grab samples have been conducted since 1993 in the stream to collect in situ standard physicochemical data. Monitoring has been intensified since 1999. Rain water composition has been investigated since 2010. The highest frequency of sampling in the stream (Kervidy-Naizin) is 1 d since 1999, and the highest frequency of groundwater sampling (Kerbernez piezometers) is 1 wk (between 2005 and 2012). Automatic samplers are used to sample storm flow events every 10 to 30 min over 10 h, which is consistent with the duration of such events in the catchments. Between 4 and 10 events are monitored every year. Since 2007, some complementary, nearly continuous (15–10 min) monitoring of physicochemical variables has been conducted using dedicated sensors: turbidity and concentrations of NO3, organic C, and P in stream water and temperature and electrical conductivity in stream and groundwater.

The evolution of land cover from one year to another is obtained by farm survey or in-field observations or is reconstructed from remote data by analyzing and classifying the information contained in images. The main annual land use in each field is recorded, at least in the case of cover crops. The land cover maps over the period 1993 to 2018 account for the modifications of the field limit over time.
In addition to these data, additional observations are conducted for the duration of the project or even through several successive projects, leading to dedicated medium- to long-term data sets. 

=== Soil === 
The first soil maps were established in the early stages of the research sites (Lamandé, 2003; Walter, 1993) to provide an extensive description of the spatial organization of soils and connect them to their hydrological and geochemical functioning (Walter and Curmi, 1998). High-resolution soil maps (1:10,000 and 1:25,000) have been established on the basis of four criteria: soil parent material, soil depth, soil redoximorphic conditions, and soil types as defined in the French soil classification (Baize et al., 2002). The physicochemical properties of soils (texture, pH, nutrient content, organic matter, trace elements) have been measured over the catchment, and hydrodynamic properties of soil horizons (hydraulic conductivity, porosity) have been quantified for the main soil types. In more recent research projects, additional soil collection campaigns have been conducted in the Kervidy-Naizin site to quantify C, N, and P stocks in the soil compartment and to characterize the biochemical processes controlling C, N, and P cycling in soils. Detailed data on soil C, N, and P content; soil aggregation; and microbial communities (molecular microbial biomass estimated by soil DNA recovery, bacterial and fungal communities analyzed by 16S and 18S rRNA gene pyrosequencing) (Matos-Moreira et al., 2017) have been collected. 
Soil moisture sensors have been in use since 2010 in specific locations of the catchments and according to different strategies. At the Kerbernez site, 21 frequency domain reflectometry sensors are coupled with nested piezometers to monitor the groundwater recharge process. At the Kervidy-Naizin sites, time domain reflectometry sensors (often coupled with temperature sensors) are used either on soil profile (at −5, −20, and −50 cm deep) or in the upper layer of soil (−5 cm) to interpret measurements of gaseous fluxes (CO2, N2O) conducted in successive research projects. The time step of acquisition ranges from 15 to 30 min.

Recharge water was sampled using profiles of ceramic cups between −25 cm and −2.5 m on which suction was applied for water collection. At the Kervidy-Naizin site, the chemical composition of soil water has been investigated since 2011 focused on organic matter and P. Bottomland soils and wetland soils have been instrumented with zero-tension lysimeters and mini-piezometers to allow the sampling of free water in soils (Denis et al., 2017a, 2017b; Dupas et al., 2015c; Gu et al., 2017; Lambert et al., 2011). These samples are analyzed to determine their concentrations in anions, dissolved C (and Fe), and dissolved P, in addition to any other variables of interest.
 
=== Surface–Atmosphere Exchange Fluxes ===
A flux tower station was installed in 2015 over grazed grassland, recording surface/atmosphere H2O and CO2 fluxes continuously over a few years to capture their seasonal and interannual variability. During the day, CO2 fluxes are linked to H2O fluxes via photosynthesis and transpiration of the vegetation, which is controlled by bulk canopy stomatal conductance; during the night soil and plant respiration proceed. Fluxes are measured by eddy covariance, and a set of ancillary meteorological and soil variables is recorded on site. The measurement height is 2 m, corresponding to a flux footprint of the order of 1 to 3 ha depending on wind speed, atmospheric turbulence, and stability.

=== Analyzing Water Chemistry at Higher Temporal Resolution ===
Several technologies have been or are being tested regarding their ability to provide water concentrations at resolution >1 h. Sensors that estimate concentrations using indirect methods including in situ ionic specific probes and spectrophotometric probes were installed in 2010 to monitor NO3, Cl, and organic C concentrations. They have been tested in stream and piezometers, leading to satisfactory results for the spectrophotometer, which records NO3 and dissolved organic C every 15 min (Faucheux and Fovet, 2014). Ionic specific probes have been found more reliable for spatial campaigns than for continuous monitoring at a given point. The most recent generation of tools is based on river bank side analyzers, which pump water directly from the stream and determine concentrations by direct physicochemical methods equivalent to those used in the laboratory in real time, avoiding storage and transport issues. A P analyzer has recorded total and reactive P in the Kervidy-Naizin site since 2016 every 30 min (Jordan et al., 2007). The challenge to increase the temporal resolution of analysis for various chemical elements in water is also taken up with the development of a river laboratory prototype (Floury et al., 2017). A second prototype measures major ion concentrations (by ionic chromatography), dissolved silica (colorimetry), and organic C (acid mineralization and infrared CO2 measurement) approximately every 30 min in stream water at the outlet of the Kervidy-Naizin site. 

=== Dedicated campaigns and experiments ===
Campaigns for Dynamic Mapping of Wetlands in Space and Time. According to their role in overland flow generation (Beven and Kirkby, 1979) and their role as hot spots for biogeochemical reactions (Sabater et al., 2003), wetlands have been the subject of several research projects. Wetlands result from the combination of climate, topography, and geomorphology zones in areas where the water table intercepts shallow soil horizons (Crave and Gascuel-Odoux, 1997). Such combinations are dynamic in space and time. Wetlands inventories have been made on the field by visual identification and by helicopter-borne radar (Gineste et al., 1998; Merot et al., 1994). Thanks to the inventory, a method for mapping those wetlands from a digital elevation model and using a climate-topographic index (Merot et al., 2003) has been developed.

=== Geophysical Campaigns and Experimental Determination of Hydrodynamic Properties. ===

To characterize the hydrodynamic properties of the weathered rock, slug tests and pumping tests were performed in some of the piezometers to provide estimates of hydraulic conductivity (Martin et al., 2006; Molenat and Gascuel-Odoux, 2002; Molénat et al., 2005; Pauwels, 1994; Vouillamoz, 2003).In the Kerbernez site, these tests were combined with geophysical surveys (electrical imaging, electromagnetic and magnetic resonance sounding) and indicate that the thickness of the weathered granite increases from upslope toward downslope areas of the catchments in the form of a deep graben structure (Legchenko et al., 2004). Subsequently, an experimental determination of weathered material was performed using the Wind method (Rouxel et al., 2012), showing that the retention curve of weathered granite is different from soil retention curves and cannot be easily estimated using pedotransfer function approaches.

=== Tracer Experiments and Use of Geochemical and Isotopic Tools. === 

Hydrological and geochemical deconvolution can use various tracers to identify contributive flowpaths and biogeochemical reaction processes or to estimate water and element residence times in catchment compartments such as soil, vadose zone, or groundwater. Various tracers have been used for different objectives. In the Kervidy-Naizin site, a lot of projects were dedicated to storm flow generation and associated exports of dissolved organic matter (DOM). Storm flow deconvolutions based on water isotopes (18O) showed the small portion of recent water in storm hydrographs (Merot et al., 1995), pointing to the importance of old water from soil and groundwater in the genesis of storm flow. Then, several tracer approaches were used and compared to identify and quantify the contributions of different compartments to DOC export during storm events and thereby also to improve our understanding of subsurface flows. The 13C of C (Lambert et al., 2011, 2013, 2014), the fluorescence signature, and the molecular composition of DOM (Denis et al., 2017a; Jeanneau et al., 2014) confirmed the major contribution of riparian wetlands to storm DOC and emphasized a secondary contribution of downslope areas, which decreased along successive storms. The fluorescence signature also emphasized the potential direct contribution of animal manure to the stream DOM during intense spring storm events. In the Kerbernez sites, tracer experiments have been conducted to gain a better understanding in groundwater recharge. The first attempt to use 2H in tracer campaigns at the scale of shallow piezometers was not conclusive, most likely because of excessive transfer times. Tracer experiments with Br and 2H helped to estimate the water velocity in soils and showed the bimodal properties of the velocities. Enzymatic activity was also used in combination with solute concentrations in laboratory experiments to characterize the biogeochemically reactive transfer in the soil and the weathered zone, showing that heterotrophic denitrification was the dominant process (Legout et al., 2005, 2007). Groundwater age was determined using atmospheric tracer chlorofluorocarbons and SF6, highlighting the distinction between the weathered zone (where apparent ages ranged from 12 to 25 yr) and the weathered-fissured and fractured parts (where apparent age increased with depth and was >25 yr) (Ayraud et al., 2008; Molénat et al., 2013). In both catchments, shallow groundwater feeds the streams; therefore, different experimental tools have been explored to quantify the subsurface fluxes and their temporal or spatial variations. Preliminary tests on radon were started in 2015, and studies on temperature using distributed temperature sensing by optic fiber were started in 2016.
Atmospheric Emissions and Deposition. Several campaigns have been conducted since 2007 to assess various fluxes, such as CO2, N2O, and NH3, at the landscape scale. For example, Buysse et al. (2016) measured soil CO2 efflux with closed dynamic respiration chambers over a 1-yr period (36 weekly to biweekly measurement dates) at 22 sites across the Kervidy-Naizin catchment. They found that water regime, land-use, and crop rotation significantly affect soil CO2 emissions, with lower emissions observed in poorly drained soils either due to lower respiration or to limited CO2 transport in saturated soils. A network of passive and low-cost sensors for atmospheric ammonia (NH3), which integrates concentrations over monthly periods, was also established across the catchment (Tang et al., 2001). These low-resolution atmospheric NH3 data were complemented by mobile NH3 plume measurement campaigns, downwind of the main agricultural NH3 sources (animal housing), using fast-response (1 s) quantum cascade laser technology. These datasets were used to improve landscape-scale estimates of total NH3 emissions as well as local dry deposition, which contribute significantly to the total N load of agroecosystems in areas of intensive animal farming (Bell, 2017).