EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Modeling Nitrate Export at the Catchment Scale using StorAge Selection Functions

Tam Nguyen1, Rohini Kumar2, Stefanie R. Lutz1, Andreas Musolff1, and Jan H. Fleckenstein1
Tam Nguyen et al.
  • 1Helmholtz-Zentrum für Umweltforschung GmbH - UFZ, Department of Hydrogeology, Leipzig, Germany
  • 2Helmholtz-Zentrum für Umweltforschung GmbH - UFZ, Department Computational Hydrosystems, Leipzig, Germany

Catchments store and release water of different ages. The time of a water parcel remaining in contact with the catchment subsurface affects the solute dynamics in the catchment and ultimately in the stream. Catchment storage can be conceptualized as a collection of different water parcels with different ages, the so-called residence time distribution (RTD). Similarly, the distribution of water ages in streamflow at the catchment outlet, which is sampled from the RTD, is called the travel time distribution (TTD). The selection preferences for discharge can be characterized by StorAge selection (SAS) functions. In recent years, numerical experiments have shown that SAS functions are time-variant and can be approximated, for example, by the beta distribution function. SAS functions have been emerging as a promising tool for modeling catchment-scale solute export.

In this study, we aim to integrate the SAS-based description of nitrate transport with the mHM-Nitrate model (Yang et al., 2018) to simulate solute transport and turnover above and below the soil zone including legacy effects. The mHM-Nitrate is a grid based distributed model with the hydrological concept taken from the mesoscale Hydrologic Model (mHM) and the water quality concept taken from the HYdrological Predictions for the Environment (HYPE) model. Here, we replaced the description of nitrate transport in groundwater from the original mHM-Nitrate with time-variant SAS-based modeling, while we kept the detailed description of turnover of organic and inorganic nitrogen in the near-surface (root zone) from mHM-Nitrate. First-order decay was used to represent biogeochemical (denitrification) processes below the root zone and in the stream. The proposed model was tested in a mixed agricultural-forested headwater catchment in the Harz Mountains, Germany. Results show that the proposed SAS augmented nitrate model (with the time-variant beta function) is able to represent streamflow and catchment nitrate export with satisfactory results (NSE for streamflow = 0.83 and for nitrate = 0.5 at the daily time step). Overall, our combined model provides a new approach for a spatially distributed simulation of nitrogen reaction processes in the soil zone and a spatially implicit simulation of transport pathways of nitrate and denitrification in the entire catchment.

Yang, X., Jomaa, S., Zink, M., Fleckenstein, J. H., Borchardt, D., & Rode, M. ( 2018). A new fully distributed model of nitrate transport and removal at catchment scale. Water Resources Research, 54, 58565877.

How to cite: Nguyen, T., Kumar, R., Lutz, S. R., Musolff, A., and Fleckenstein, J. H.: Modeling Nitrate Export at the Catchment Scale using StorAge Selection Functions, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-7967,, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 29 Apr 2020
  • CC1: spatial patterns (getting back to question in live chat), Raphael Schneider, 06 May 2020

    Hi again, and thanks for the interesting contribution. As I asked in the live chat (but then time ran out):
    You show some spatial patterns on slide 14 in your display, and you mentioned that you compared those to previous models. Can you elaborate on how they compare and how plaubsible you think they are? And, talking about input - how did you distribute your N/fertilizer application?

  • AC1: Comment on EGU2020-7967, Tam Nguyen, 06 May 2020

    Hi Raphael,

    thanks for your interest in our study

    Currently, we can only quantitatively validate the spatial pattern of N within the soil zone (compared the spatial pattern in our study with the previous study (Yang et al, 2018) . We are trying to validate (1) nitrate leaching with N surplus data from other data sources. The exact magnitudes of these spatial patterns are highly uncertain but  the  uncertainty of these spatial patterns is expected to be less (you are absolutely right that  evaluation of these spatial patterns should be paid more attention)

    Fertilizer/manure application is distributed base on crop rotation data for each grid cell

    Thanks again for your question


  • AC2: Additional reply to other questions during the oline chat, Tam Nguyen, 06 May 2020

    Q1: The simulations look really good, but seem to show an understimation of peak nitrate concentrations and an over-estimation of of some low flow concentrations - are the groundwater and soil water nitrate concentrations changing - is it the biogeochemical part of the model that is controlling this, rather than the hydrological?

    A1: Thanks for your question, nitrate dynamics within the soil zone is described by the HYPE model, nitrate concentration within this zone depends on input N, biogeochemial processes (e.g., denitrification, mineralization, degradation), plant uptake..

    Nitrate concentration in groundwater is also changing because nitrate in groundwater could be removed by denitrification, which is incorporated into the StorAge selection function in a parsimonious manner

    Q2: Did you consider simulating chloride next to nitrate to better understand the catchment-scale effects of reactivity and shallow storage of N?

    A2: Simulating both chloride and nitrate could significantly reduce the model parameter uncertainty and increase the predictive power of the model. However, we don't have enough data to carry such as simulation