EGU2020-5807
https://doi.org/10.5194/egusphere-egu2020-5807
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

A new physically-based catchment modelling tool for reservoir re-engineering and renaturalisation

Daryl Hughes, Geoff Parkin, and Stephen Birkinshaw
Daryl Hughes et al.
  • School of Engineering, Newcastle University, Newcastle Upon Tyne, United Kingdom of Great Britain and Northern Ireland

The hydrological regimes of European catchments have been considerably modified by anthropogenic features such as dams, weirs and water abstractions, with nearly every major river fragmented. The negative impacts of such physical modifications on freshwater ecosystems are being increasingly recognised. Currently, European dam removal initiatives are being driven by factors such as the EU Habitats Directive, and the costs associated with maintaining redundant infrastructure. Climate change and the rewilding agenda may encourage further hydrological renaturalisation initiatives. In the English Lake District, several reservoirs are being actively considered for decommissioning within this decade. To understand how such catchments would respond to lake renaturalisation, robust catchment hydrology models are needed that can represent the effects of changes in physical infrastructure on the hydrological regime. However, many models tend to neglect such human impacts.

We present a new tool that incorporates reservoirs, including impounding structures, river regulations and abstractions. The method involved development of an enhanced version of the freely-available catchment modelling software, SHETRAN. A new ‘reservoir’ module was developed which includes the effects of hydraulic structures and sluice operations on lake stage and river flow. Results for the Crummock Water catchment and reservoir show that the reservoir model generates notably fitter simulations, particularly during dry periods where reservoir operations cause a distinct deviation from the regime expected in natural lake-river systems. Further simulations demonstrate quantitatively how lake renaturalisation might affect future hydrological regimes compared with the baseline scenario. Finally, we discuss the implications of this model for decision-making in the Crummock Water catchment, and the utility of the software for other anthropologically-modified catchments.

How to cite: Hughes, D., Parkin, G., and Birkinshaw, S.: A new physically-based catchment modelling tool for reservoir re-engineering and renaturalisation, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-5807, https://doi.org/10.5194/egusphere-egu2020-5807, 2020

Display materials

Display file

Comments on the display material

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

Display material version 2 – uploaded on 03 May 2020
Version description: I have changed the title slide to make it clear that there are comment boxes throughout the presentation. The viewer[...]
  • AC1: Online Q&A: 1, Daryl Hughes, 04 May 2020

    Q) Sina Khatami: Slide 11) it seems to me that the main difference between the SHETRAN Standard and SHETRAN Reservoir is the event around (2014-07), and all the two models are almost similar across other times. And for that event, Standard is over-estimating and Reservoir is under-estimating. I’d say that is the main contributor to the different NSE values. What do you think about that?

     

    A) Hi Sina. I ought to clarify that the NSE values are for the entire five year simulation periods rather than just the seven months shown on the hydrograph you refer to. Yes, you are correct that the overestimation of post-dry period peaks is a key difference between the Standard and Reservoir models, and what accounts for the difference in model fitness. The other important aspects of the Reservoir model are the dry period flows and reservoir drawdown. Although NSE is quite insensitive to these dry period flows and they account for relatively small volumes of water, they are nonetheless very important for ecological and water resources reasons.

  • AC2: Online Q&A: 2, Daryl Hughes, 04 May 2020

    Q) Luis Samaniego: How do you parameterize the reservoirs? How do you get the management rules? For many dams across the globe we don't get this information

    A) Hi Luis. The reservoirs are submerged grid cells, constrained by surrounding topography on the DEM, and the balance between inputs and outputs. The reservoir outflows (structure and operations) are crucial to ensuring model fitness. The Crummock model uses known geometry from engineering drawings and standard weir equations; this was validated using a downstream gauging station.

    The reservoir abstractions were available as a daily observed time series. This eliminates abstraction uncertainties. However, so long as there is relatively little variation in abstraction and/or the abstracted volumes are small compared with the other outputs, a much coarser and/or less accurate abstraction volume ought to suffice without seriously undermining the water balance.

    As for sluice operations, there were no written rules or records. An operator was interview to understand the general policy. Nonetheless, this made it quite hard to construct a robust model of discharge via the sluices. However, by using the observed discharge, it was possible to determine when and how the sluice was operated e.g. when there was a sudden change in discharge in the absence of rainfall.

    Whilst I’ve had the time to build a detailed model of the Upper Cocker since it is part of my PhD, one can construct good models of reservoir-containing catchments with less time and data. Beyond the requirements for any spatially-distributed model, a few additional data are needed: 1) A good reservoir outflow model is the most important. These can be constructed using elevations and dimensions of dam crests, spillways etc. If engineering drawings are not available, simpler models can be constructed based on estimates. 2) Abstraction. If no abstraction records are available, an estimate may suffice. 3) Reservoir levels. These are useful for calibration/validation; particularly if relying on unknown outflow models.

  • AC3: Online Q&A: 3, Daryl Hughes, 04 May 2020

    Q) Luis Samaniego: is the reservoir delineation automatic?

    A) The reservoir delineation relies on the underlying DTM. Since LiDAR doesn’t penetrate water well, DTMs are often flat. In this case, the bathymetry will need to be included. The modeller can do this using GIS. Alternatively, the grid cell elevations can be changed in the model set up files.

  • AC4: Online Q&A: 4, Daryl Hughes, 04 May 2020

    Q) Kalai: Is the soil property incorporated in the model?

    A) Yes. Multiple layers of soil can be input in SHETRAN, with spatially-distributed varying parameters to pass to the equations describing soil and groundwater behaviour e.g. Darcy, Van Genuchten.

Display material version 1 – uploaded on 01 May 2020, no comments