Changing fieldwork practice to enable effective archive-model comparison -
a suggested framework
1Briant, R.M., 2Cohen, K.M., 3Cordier, S. 4Demoulin, A., 5Macklin, M.G.,
6Maddy, D., 7Rixhon, G., 8Veldkamp, A., 9Wainwright, 10Whittaker, A.
Lead author email address:
[email protected]
1Department of Geography, Environment and Development Studies, Birkbeck,
University of London, Malet Street, London, WC1E 7HX, U.K.
2Dept. Fysische Geografie, Fac. Geowetenschappen, Univ. Utrecht, Postbus
80.115, 3508 TC UTRECHT, The Netherlands
3 Département de Géographie et UMR 8591 CNRS- Université Paris 1-Université
Paris Est Créteil, Créteil Cedex, France
4Dept of Physical Geography and Quaternary, University of Liège, Sart
Tilman, B11 - 4000 Liège, Belgium
5Institute of Geography and Earth Sciences, University of Wales,
Aberystwyth, Ceredigion, SY23 3DB, U.K.
6School of Geography, Politics and Sociology, Newcastle University,
Newcastle upon Tyne, NE1 7RU, UK
7University of Cologne, Institute for Geography, Albertus-Magnus-Platz,
50923 Köln, Germany
8ITC, Faculty of Geo-Information Science and Earth Observation of the
University of Twente, PO Box 217, 7500 AE Enschede, The Netherlands
9Durham University, Department of Geography, Lower Mountjoy, South Road,
Durham, DH1 3LE, UK
10Department of Earth Science and Engineering, Imperial College London,
South Kensington Campus, London SW7 2AZ, U.K.
Comparison of field-based fluvial data with landscape modelling output of
various types and at different time-scales is not simple due to mismatches
in the types of data generated. For example, field work generates detailed
site-specific field data with significant and uncertain time gaps even with
the most robust chronologies compared with continuous three-dimensional
model data which often lacks detail. Also, time-series outputs at the
catchment outflow are the simplest output to produce from models, but not
available from field data except in rare situations when a system drains
into a closed basin. There may also be a different focus required when
assessing tectonic or climatic drivers. Fine-tuning of both field data
collection and model outputs can do much to ensure greater comparability,
and thus improve our understanding of how systems operate.
These issues were addressed in December 2013 by a group of European
specialists in both numerical modelling and field investigation of fluvial
systems at a workshop in Twente in the Netherlands. Following this meeting,
two workgroups were convened to deliver the objectives listed below:
Workgroup 1:
We will prepare a research proposal which aims to benchmark and
document a wide-range of Landscape Evolution Models.
We will prepare an outline framework for rigorous model testing.
Workgroup 2:
We will prepare an outline framework for the comprehensive recording
and reporting of field based observations targeted for model-data
comparison.
We will establish a robust set of 'real-world' model testing datasets.
This paper reports on the outline framework from Workgroup 2 as it has been
developed so far, and provides an opportunity for us to seek feedback on it
from the wider fluvial archives community. The framework suggests changes
to both site-based and catchment-scale observations, to provide data for
both initial model conditions and evaluation of model outputs. Key points
are the importance of chronologies – both at-a-site (as is standard
practice) but also catchment-wide (addressing the importance of sampling
strategies); recording more three-dimensional field data and recording
field data on hillslopes as well as fluvial landforms / sediments – i.e.
the whole catchment, as is modelled. We stress also the need to generate
field data that better facilitates the evaluation of model output using
pattern-matching – i.e. looking for similar emergent properties from model
output to the field data rather than an exact match. The pattern-matching
approach enables the stochastic properties of both natural and modelled
systems to be accommodated.
