Development of a GIS Procedure to Implement a UK Geomorphic Channel Typology
July 2008


Background to research

Both Environment Agency and SEPA have a requirement under the Water Framework Directive to assess the morphological impacts of modifications to rivers. Within Scotland this is based on the MImAS assessment methodology and feeds into the operation of Controlled Activities Regulations and potentially into the type-specific Programmes of Measures. One of the key factors in the sensitivity of rivers is the geomorphic typology of the river, the classification of rivers based on their morphological attributes. Field surveys of individual reaches of homogeneous typology is seen as prohibitively expensive, and despite earlier WFD typologies there is currently no automated approach to assigning typology from secondary derived variables. However, an allocation tree developed in an earlier project WFD49c and a pilot project by Environmental Systems Research Institute (ESRI) suggested that it would be possible to create tools for the allocation of river typologies using an allocation algorithm driven by geomorphic variables extracted from GIS data.
Objectives of research

The aims of the project were to evaluate the input GIS datasets, ESRI tools and to develop a specification for the creation of a suite of open GIS tools within WP1 that could be implemented within EA and SEPA’s IS infrastructure, while also being flexible enough to allow changes to input thresholds on which allocation is based. Different methods of allocating river types were also to be investigated (e.g. fuzzy logic). Within WP2, the GIS datasets were to be collated and prepared and GIS tools were to be developed that extracted various geomorphic parameters and then allocated river reaches within a series of test catchments. The reach scale typologies were based on an assumed unmodified channel and the anticipated natural channel typology, although it is evident that channel modifications will influence the ability to accurately assign types. Allocated typologies were then to be subject to validation against field and other datasets and the algorithms and tools refined where necessary. Finally, WP3 sought to implement this methodology and to generate typologies for all Scottish Rivers.
Key findings and recommendations

The GIS data preparation and allocation tool functions to specification, is intuitive and has been successfully tested and accepted in the 2 agencies (with any minor incompatibility issues resolved during installation). The tool was specified to be flexible and open to amendment, including the potential to modify the decision tree and data extraction rules and the threshold values used to distinguish between types.

However, the results are only as good as the input datasets and the algorithm used to generate the allocation and the conceptual relationship between the variables selected and the typological divisions. The original specification identified a range of other desirable parameters that might have complemented the basic allocation model, but where the availability of nationally consistent data are lacking. Inclusion of additional data would also affect the agreed allocation decision tree which was based on the four core attributes (geology, slope, confinement and sinuosity). Therefore, the user should review the comments on data and also consider closely the logic used by the decision tree.

The project has highlighted the need for additional validation work on the resulting typology and the potential to add variables to improve on the typological allocation process.

The validation results confirm that there are some classes which are very difficult to discriminate in the field (e.g. due to water levels etc.) and difficult when trying to allocate typologies using secondary data within the office when the characteristics of the types and thresholds between types are based on the 4 core variables which display overlapping ranges.

Detailed breakdowns of the types are given nationally and also by catchment within the Appendices. The results for Scotland showed that 65% of rivers are allocated to a ‘firm class’ based on SEPA’s revised decision tree (MImAS types A, B, C, D, F) with the remaining 35% being allocated to low confidence classes or where it is expected that the section is subject to partial modification.

Additional work on the allocation algorithms and the approach to their application and testing requires refinement. The development and testing of fuzzy and genetic algorithm based approaches would allow for the inherent uncertainties related to the classification and threshold values to be tested.

In addition, exploration of the potential to include additional parameters (e.g. bed material, stream power etc.) into the allocation logic is also recommended. There is potential to add predictive level channel attributes (potentially for bed substrate and stream power) to build on the current approaches and seek to reduce low confidence typology allocations.

Keywords: river typology, GIS, WFD, decision tree

Copies of this report are available from the Foundation, in electronic format on CDRom at 20.00 + VAT or hard copy at 15.00, less 20% to FWR members.

N.B. The report is available for download from the SNIFFER Website