Surface-to- Tunnel Natural Tracer Transport Modelling with Various Recharge Variants

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F46747885%3A24620%2F17%3A00004553" target="_blank" >RIV/46747885:24620/17:00004553 - isvavai.cz</a>

Výsledek na webu

<a href="https://portugal.iah.org/gwfr2017" target="_blank" >https://portugal.iah.org/gwfr2017</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Surface-to- Tunnel Natural Tracer Transport Modelling with Various Recharge Variants

Popis výsledku v původním jazyce

In this study we investigate properties of a fractured granite massif around Bedrichov water supply tunnel by means of transport modelling of stable isotopes 18 O and 2 H to determine the water age or velocity. The excavated tunnel is located in the Jizera Mountains, in the north of the Czech Republic and in a portion of the Bohemian massif - the Krkonose-Jizera Composite Massif (Klomínský & Woller 2011). The tunnel length is 2600 m and the first 890 m of the tunnel was excavated with the tunnel boring machine (TBM) method and remaining part utilizing drill-and- blast method. Besides the water supply, the tunnel is used as an underground laboratory for various measurements (geological, seismic, hydrological, hydro-chemical etc.). The stable isotopes are sampled in 14 seepage sites twice or one per month (2/2010-10/2016). The time series of stable isotopes in the precipitation (11/2005-10/2016) come from the nearby experimental catchment Uhlirska operated by the Czech Technical University (Šanda et al 2014). The study follows previous work on 3D numerical modelling of groundwater hydraulics on the site, calibrating the tunnel inflow spatial distribution with inhomogeneous rock model (Hokr et al, 2014). Additionally, the rock transport properties can be determined from the natural tracer transport, with input with the recharging water on the surface and output. Some models of tracer transport related to the site were elaborated within the DECOVALEX-2015 project, in a simplified 3D geometry combining a single fracture and the rock matrix (Hokr et al, 2016). These were done with fictitious tracer pulse injection, allowing to determine residence time from the breakthrough curve. To understand the full context, the isotope transport is also evaluated by a lumped parameter model (LPM). Partial study with shorter data sequence and limited choice of sampling points (Gardner et al, 2016). LPM is based on the assumption that the transit time distribution function of the tracer particles through the groundwater system is known or can be assumed (Maloszewski & Zuber 1996). For this work, we use the software FLOWPC and assume the dispersion distribution function and we evaluate two parameters – mean residence time (MRT) and apparent dispersion parameter (PD= D/vx, reciprocal of the Peclet number). The parameter values must be obtained by “inverse modelling” (calibration), fitting the modelled and measured tracer concentration in the sampling point (output). Several automatic calibrations in UCODE software (Poeter & Hill 1998) are performed on models with various input function of water infiltration/recharge rate and tracer concentration, which generally can differ from the precipitation measurement values. For two basic variants, we suppose uniform infiltration of 20% of the average precipitation, or variable infiltration with the same ratio in monthly steps. For other three variants we consider non-trivial relationships between the precipitation and infiltration, based on assumption that the recharge happens preferably in the events of excess water and it is therefore related to the catchment outflow which is an available measured quantity in Uhlirska. The model input of the isotope concentration is transformed from the precipitation by accounting for the snow cover accumulation infiltrated later in spring months. We performed a computation with raw data as control variant too. For an evaluation of the model quality, the FLOWPC software uses two statistics parameters – the goodness of fit (SIGMA) and the model efficiency (ME). SIGMA expresses how close the model is to measured data (the closer to zero values the better) and ME expresses how well the model fits the shape of the measured data (the closer to 1 value the better). The fitted model data capture the trend of measured data quite reasonably for the shallow seepage sites V7 (-23 m depth) and V6 (-36 m depth) – see the figure. The SIGMA value is 0.22 and 0.18 respectively and ME is 0.42 and 0.48 respectively. The best result comes from the simpler approximation of the infiltration rate. The more sophisticated input approximations do not improve the results. Definitely the approximation of snow cover accumulation works well in the time series of stable isotopes. The results of MRT are 21 months for V7 and 24-26 months for V6 respectively and they can be used in more sophisticated numerical models as initial estimations or calibration data. Contrary to typical interpretation of the water stable isotope temporal evolution capturing the seasonal variations within one year between the input and output, the data in this case study need the longer- period changes to be captured, which makes more requirement on the input evaluation, e.g. to represent the differences between the years. This was in particular difficult for the respective years with quite irregular evolution of weather and hydrologic conditions.

Název v anglickém jazyce

Surface-to- Tunnel Natural Tracer Transport Modelling with Various Recharge Variants

Popis výsledku anglicky

In this study we investigate properties of a fractured granite massif around Bedrichov water supply tunnel by means of transport modelling of stable isotopes 18 O and 2 H to determine the water age or velocity. The excavated tunnel is located in the Jizera Mountains, in the north of the Czech Republic and in a portion of the Bohemian massif - the Krkonose-Jizera Composite Massif (Klomínský & Woller 2011). The tunnel length is 2600 m and the first 890 m of the tunnel was excavated with the tunnel boring machine (TBM) method and remaining part utilizing drill-and- blast method. Besides the water supply, the tunnel is used as an underground laboratory for various measurements (geological, seismic, hydrological, hydro-chemical etc.). The stable isotopes are sampled in 14 seepage sites twice or one per month (2/2010-10/2016). The time series of stable isotopes in the precipitation (11/2005-10/2016) come from the nearby experimental catchment Uhlirska operated by the Czech Technical University (Šanda et al 2014). The study follows previous work on 3D numerical modelling of groundwater hydraulics on the site, calibrating the tunnel inflow spatial distribution with inhomogeneous rock model (Hokr et al, 2014). Additionally, the rock transport properties can be determined from the natural tracer transport, with input with the recharging water on the surface and output. Some models of tracer transport related to the site were elaborated within the DECOVALEX-2015 project, in a simplified 3D geometry combining a single fracture and the rock matrix (Hokr et al, 2016). These were done with fictitious tracer pulse injection, allowing to determine residence time from the breakthrough curve. To understand the full context, the isotope transport is also evaluated by a lumped parameter model (LPM). Partial study with shorter data sequence and limited choice of sampling points (Gardner et al, 2016). LPM is based on the assumption that the transit time distribution function of the tracer particles through the groundwater system is known or can be assumed (Maloszewski & Zuber 1996). For this work, we use the software FLOWPC and assume the dispersion distribution function and we evaluate two parameters – mean residence time (MRT) and apparent dispersion parameter (PD= D/vx, reciprocal of the Peclet number). The parameter values must be obtained by “inverse modelling” (calibration), fitting the modelled and measured tracer concentration in the sampling point (output). Several automatic calibrations in UCODE software (Poeter & Hill 1998) are performed on models with various input function of water infiltration/recharge rate and tracer concentration, which generally can differ from the precipitation measurement values. For two basic variants, we suppose uniform infiltration of 20% of the average precipitation, or variable infiltration with the same ratio in monthly steps. For other three variants we consider non-trivial relationships between the precipitation and infiltration, based on assumption that the recharge happens preferably in the events of excess water and it is therefore related to the catchment outflow which is an available measured quantity in Uhlirska. The model input of the isotope concentration is transformed from the precipitation by accounting for the snow cover accumulation infiltrated later in spring months. We performed a computation with raw data as control variant too. For an evaluation of the model quality, the FLOWPC software uses two statistics parameters – the goodness of fit (SIGMA) and the model efficiency (ME). SIGMA expresses how close the model is to measured data (the closer to zero values the better) and ME expresses how well the model fits the shape of the measured data (the closer to 1 value the better). The fitted model data capture the trend of measured data quite reasonably for the shallow seepage sites V7 (-23 m depth) and V6 (-36 m depth) – see the figure. The SIGMA value is 0.22 and 0.18 respectively and ME is 0.42 and 0.48 respectively. The best result comes from the simpler approximation of the infiltration rate. The more sophisticated input approximations do not improve the results. Definitely the approximation of snow cover accumulation works well in the time series of stable isotopes. The results of MRT are 21 months for V7 and 24-26 months for V6 respectively and they can be used in more sophisticated numerical models as initial estimations or calibration data. Contrary to typical interpretation of the water stable isotope temporal evolution capturing the seasonal variations within one year between the input and output, the data in this case study need the longer- period changes to be captured, which makes more requirement on the input evaluation, e.g. to represent the differences between the years. This was in particular difficult for the respective years with quite irregular evolution of weather and hydrologic conditions.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

10501 - Hydrology

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2017

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů