EGU2020-15230
https://doi.org/10.5194/egusphere-egu2020-15230
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Landscape drivers of coastal dune mobility

Thomas Smyth1,2, Ryan Wilson1, Paul Rooney3, and Katherine Yates4
Thomas Smyth et al.
  • 1Biological and Geographical Sciences, University of Huddersfield, United Kingdom (t.ag.smyth@hud.ac.uk)
  • 2College of Science and Engineering, Flinders University, Adelaide, Australia
  • 3Geography and Environmental Science, Liverpool Hope University
  • 4School of Environment and Life Sciences, University of Salford, Manchester, United Kingdom

Coastal dunes are dynamic landforms whose morphology is governed primarily by climate and vegetation dynamics. Over the last 50 years, coastal sand dunes across the globe have dramatically ‘greened’ and wind speeds fallen (Pye et al., 2014; Delgado-Fernandez et al., 2019; Jackson et al., 2019), reducing aeolian transport of sediment and minimising dune reshaping by near-surface winds.  This rapid vegetation has also been attributed to a dramatic decline of several rare species of plants and invertebrates in several coastal dune systems (Howe et al., 2010; Pye et al., 2014). In an effort to increase habitat diversity, large-scale vegetation removal and dune re-profiling are becoming increasingly common interventions. However sustained aeolian activity following intervention appears to be rare (Arens et al., 2013).

In order to better understand the environmental drivers of long-term dune mobility, this work explores the landscape scale physical factors related to self-sustaining ‘natural’ mobile dunes across the United Kingdom. The analysis presented includes the use of geographically weighted regression, a spatial analysis technique that models the local relationships between predictors (e.g. wind speed, slope, elevation, aspect, surface roughness) and an outcome of interest (mobile dunes). It is hoped that the results of this work will help guide decision-making with regards the location, scale and morphology of future interventions in order to maximise their sustainability, minimising the need for maintenance and further intervention.

References

Arens, S.M., Slings, Q.L., Geelen, L.H. and Van der Hagen, H.G., 2013. Restoration of dune mobility in the Netherlands. In Restoration of coastal dunes (pp. 107-124). Springer, Berlin, Heidelberg.

Delgado-Fernandez, I., O'Keeffe, N., & Davidson-Arnott, R. G. (2019). Natural and human controls on dune vegetation cover and disturbance. Science of The Total Environment, 672, 643-656.

Howe, M. A., Knight, G. T., & Clee, C. (2010). The importance of coastal sand dunes for terrestrial invertebrates in Wales and the UK, with particular reference to aculeate Hymenoptera (bees, wasps & ants). Journal of Coastal Conservation, 14(2), 91-102.

Jackson, D. W., Costas, S., González-Villanueva, R., & Cooper, A. (2019). A global ‘greening’of coastal dunes: An integrated consequence of climate change?. Global and Planetary Change, 182, 103026.

Pye, K., Blott, S. J., & Howe, M. A. (2014). Coastal dune stabilization in Wales and requirements for rejuvenation. Journal of coastal conservation, 18(1), 27-54.

How to cite: Smyth, T., Wilson, R., Rooney, P., and Yates, K.: Landscape drivers of coastal dune mobility , EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-15230, https://doi.org/10.5194/egusphere-egu2020-15230, 2020

Displays

Display file

Comments on the display

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

displays version 1 – uploaded on 06 May 2020
  • CC1: Comment on EGU2020-15230, Iain Fairley, 07 May 2020

    Hi Thomas,

    Nice presentation. Can you elaborate on what method you used for the supervised classification and did you test different techniques?

    Thanks,

    Iain

    • AC1: Reply to CC1, Thomas Smyth, 07 May 2020

      Hi Iain, 

      Thanks for the comment. In the poster, we employed the 'Minimum Distance' algorithm using the SCP Plugin in QGIS (). We trained the classification on 20 'patches' of vegetation and grass per image. Unfortunately, a greater number of patches is not possible with the 10 m data.

      I've briefly tested the 'Spectral Angle Mapping' algorithm with our 0.25 m data. Very preliminary results indicate that 'Spectral Angle Mapping' classifies a larger area of bare sand but the classifications in agreement with our ground-truthed locations very marginally declined (78% vs 79%). A full assessment is currently being performed.

      Are there any methods/techniques that you can recommend?

      Thanks again, Thomas

      • CC2: Reply to AC1, Iain Fairley, 07 May 2020

        Hi Thomas,

        Thanks. To remove vegetation from our dune point cloud, I used a basic artificial neural network in Matlab with RGB input and trained with known areas of sand/vegetation ; visually it looks to work well but I've not done any tests to verify that. Previously, we tested ANNs and also random forests on gridded data to classify intertidal sediment type ( https://www.mdpi.com/2072-4292/10/12/1918 ); the ANN worked marginally better and much faster computationally. However, for different studies I've heard people suggest RFs to be much less compuationally expensive than ANNs so computation time probably depends on the problem at hand.

        Cheers,

        Iain

        • AC2: Reply to CC2, Thomas Smyth, 07 May 2020

          Thanks for the information Iain!