Flood Alleviation by Catchment Process


Flood defence strategy should not only confirm risk but also assign equitable distribution of extreme event water storage;  without doubt, there is no lack of opportunity for beneficial intervention by floodplain development, Run-off Attenuation (RAF’s), and catchment-wide adaptive land-use. This article describes a numerical modelling approach – based on the innovative mapping and analysis of 2km2, 40km2 and 1000km2 catchments in Lancashire and Mid Wales – which is capable of extrapolation to any catchment or stream network. It is submitted as a public interest observation to the current 2017 Competition Coordinators of the 100 EA catchments of England and Wales. The numerical model demands hourly rainfall data. As rainfall data archives are released, calibration to such time-series promises to yield helpful methodology to encourage development of catchments – where appropriate by Ecosystem mechanism. In particular, it is in support of Upper Severn (area 59) & Upper Irwell (area 30) partnership ambition.


The essential role of rainfall data is discussed.  Reference is made that Community authorities have a duty to share resources for FRM.  The background to the problem of impact distribution, highlighted by Storm Desmond in December 2015, is argued alongside fundamental flood defence strategy.  The methodology of diffuse ponding and the communication of such strategy to stakeholders is considered.  The significance of ‘leaky’ ponds below 10,000m3 cubic capacity is explored. Finally, to put technical flesh on the bones of the approach (and marked new below),  method and parameterization of opportunity mapping, flood water detention (by chains of features), and  hydraulic modelling of downstream-stage from rain-on-grid are presented. Taking account of 2016 DEFRA competitive tender for NFM innovation, particular reference is made to the river systems draining to Carlisle and the aspirations of the Carlisle Floods Partnership.


 Figure 1:  The primary river system pouring to Carlisle within cropped relief.  In this catchment, a budget average of 10,000 cubic metres per km2 could accommodate the 18,000,000 m3 ‘Desmond’ overspill now known to cause loss.



The primary parameters of catchment process are elevation, terrain roughness and hourly local rainfall. The progress of mobile technology promises mapping tools to identify on site ‘hot-spots’ not only of rural pollution but also rainfall detention.  2D numerical analysis takes advantage of distributed datasets thereby reducing reliance on imposed 1D equations (of simulated canalization, weirs and storage) which are difficult for stakeholders to understand. Diffuse managed ponding (eg Fig 2), riparian amenity features and corridors of dense vegetation (Fig 4) are useful tools for controlling downstream flow.  Communication and good understanding underpin delivery of risk-gain in areas of diverse stake-holding and ownership. The emergence of open source analysis code alongside automated harvesting of data has created new opportunities for testing solutions and engaging public interest.

bourdon-2016-suds-1311 Fig 2:  Engineered (c. 7000m3) temporary storage pond draining to Lac du Bourdon, (Loire Valley, France, Aug 2016 ) 


Under the Flood and Water Management Act 2010, risk management authorities have a duty to co-operate and to share data.   A key theme of the Pitt Review was for flood risk management authorities to work in partnership to deliver better flood risk management for the benefit of  communities. However Catchment Process can fall between the Environment Agency’s focus on main-rivers, and LA focus on surface water flooding from urban built environment. Partnerships may not always have a lead on flood alleviation.


Fig 3 : Mid Wales amenity with blue-green storage potential upstream of the footbridge (May 2017)


Flood Management is effectively dynamic control of water:  The two primary levers of shallow water hydraulic control are roughness and elevation (form).   In traditional analysis of such hydraulic flow, elevation (and slope) assign force to flood flow, with resistance (roughness) adjusted supportively to achieve solution (of analysis) and calibration. Whilst both parameters are ‘given’ by our natural environment, for strategic/built schemes, it is more usually form which is developed for community and private resilience (levees, dams, structures, etc).  But is it defendable to focus FRM on this expensive (built, smooth and often ugly) ‘form-based’ approach, when the natural sibling ‘roughness’, so often FRM’s Cinderella, is delegated to a mere reactive role?  Roughness grows naturally.  It produces not only useful (albeit porous) capital assets, but also puts down roots to achieve self-maintenance. Roughness can be cropped for food and energy. Whilst it has to be managed for good control, it does not have to be built.


Fig 4 :  Rhizome hedge (Fougeres, France) appropriate for deployment across flood plains to soften hydrograph peaks by enhancing blue-green storage (April 2017)


If downstream discharge (of floodwater) and built defences are at a limit, high value flood-risk-loss can be moderated by distributing ‘blue-green’ storage more benignly upstream, not least to provide amenity (Fig 3). In the case of Carlisle, capacity west of the M6 is beyond breaking-point promoting consideration of rural areas for attenuation augment (ie to the East and South of the M6 crossing ).    Bearing in mind that large-scale barriers (such as the Leigh) are engineered to a higher standard than modest ‘diffuse’ storage ponds (ie those less than 10,000m3), a reasonable budget for distributed detention can be set out, for example, as Table 1 below, to bring the Environment Agency’s target of  18Mm3 for the Eden catchment within reach .

Storage   Vol (m3) Budget Total Comment with reference links
Rural Reservoir 2,600,000 £5/m3 £13  million  Engineered Storage   with control; (Leigh 2Mm3 budget being £10M)
Rural Pond 12,000,000 £0.5/m3 £ 6  million Detention chain: (60m x 100m x 1m)  diffusion/storage per km2
Rural Floodplain 4,000,000 £0.25/m3 £ 1  million High-friction 40 km2  floodplain 0.1m deeper
Rural Total 18,000,000 £20 million 18Mm3 = approx. 2015 urban overspill
Urban Repair £5 million Tidy up gaps in 0.5% AEP provision
TOTAL  18,000,000 £25 million Carlisle’s share of Cumbria budget

Table 1: Budget for system attenuation distributed by Catchment Process

[Note: 18Mm3 relates approximately to a downstream reduction of some 10% by discharge and 7% by stage –  additional catchment storage always being assessed in terms of both medium and extreme event concentration times. Notionally, a barrier of gabions under the M6 could disallow flows greater than 1300 m³/s.]



In broad terms, intense rainfall is applied to synthetically ‘hyper-rough’ terrain to create an ‘extreme’ inundation perimeter. This perimeter effectively maps potentially useful green storage ‘hotspots’ capable of development on the ground. There is evidence from pilot study that chains of such attenuation features (developed primarily in roughness but also in form) diffuse downstream stage peaks. 2D analysis under historic hourly rainfall of terrain with alternative distribution of detention ponds can then compare the benign effect downstream.  Figs 5 & 6  below (grid SO2397) show relief overlayed with the (purple) inundation perimeter of extreme hydraulic analysis marked up with (yellow) green-storage opportunity  ‘hotspots’,  (ie locations for ‘Leaky’ ponds below the 10,000m3 ceiling which fall outside reservoir legislation).

        cl-opmapstr-chirb2     cl-opmapstr-chirb3

Figs 5 & 6 (above) show green storage opportunity mapped under extreme inundation


Fig 7: (SO2397 AP40 marked-up) shows green storage development of the location highlighted in red

The location ringed red in Figs 5 & 6 is ground-truthed in Fig 7  as a potentially helpful pond of approximately 3000m3.  In this 20 km2 area alone there are interesting locations for some 30 ponds.  Once the software has mapped locations of green-storage potential, historic rainfall (hourly data being available) can be applied to alternative land-cover until the desired stage reduction is achieved. Fig 8 below (approx. 1 sq km tile SJ2212) shows a typical variable grid and a snapshot (Fig 9) of the stream network as the flow increases under catchment rain-on-grid.


Figure 8 (Grid) & Figure 9 (Flow) illustrate 40km2 Mid Wales 2D pilot model  


Fig 9: Tile SJ2212 snapshot as flow increases.

An animation screen-grab (Figure 10 below) illustrates the flow mid reach of a 1000 km2 Upper Severn 2D model similar in many respects to the Eden catchment. With a variable 10/100m computation grid (ie generally as Fig 8 above), the shallow water equations of this larger catchment solve in approximately 90 minutes.


Fig 10:  Mid reach section of Upper Severn 1000km2 model showing a screen-grab rain-on-grid flow animation.

This larger 1000km2 catchment has been tested (Fig 10) under open source code with open lidar form, synthetic roughness and synthetic rainfall. In the longer term, given data and reasonable coordination, opportunity exists for every catchment in the country to be subjected to indicative analysis of this nature at low cost. It would be helpful also to run such simulation with alternative 2D code to establish a standard nationwide template of analysis.


The published  Carlisle Flood Investigation Report (compiled by CH2M and precised here) states “Carlisle is at the confluence of three major rivers, the Rivers Eden, Caldew and Petteril, and is therefore highly prone to flooding . . .  events in 2005, 2009 and 2015 were driven by all three rivers . . . there was also an event in June 2012 . . .  current flood defences are designed to reduce the flood risk for an event with a 0.5% probability of flooding occurring in any one year  . . .  £25 million of £3bn funding has been earmarked to improve flood risk management in Carlisle” .

The 2016 competition launch presentations stated (here again in precis) that:-  “In 2015, 18,000,000m3  ended up behind defences in Carlisle . . . The Eden is a diverse catchment in terms of geography, with communities affected from headwaters to lowlands . . . The EA usually model just river and flood plain, (ie once water is in the channel) but want now to build on this with catchment modelling to manage water both overland and in smaller tributaries. . . . . . .  The Agency seeks wider integrated (CaBA), outcomes, including natural flood management with all catchment management groups represented.  . . . Cumbria RFCC supports the Agency, Local Lead Authorities and CaBA Partnerships in tactical roles.  . . .  Natural and engineered solutions are needed with outputs assessed against different storm conditions”.

(Risk awareness, uncertainty, and the legacy of 1D numerical channel simulations were rolled  out also.  Advances in Open Data were hailed and new listings of catchment analysis tools introduced. There was little mention however of solutions derived from modelling. Whilst reference was made to ‘single,  scalable assessment  by  open, and trusted software,  download links to such code and data were not disclosed.)

                                      “Why was a competition not launched in 2006 ?”.

Four Carlisle flood events in 10 years were described in terms of downstream historic channel AEP which, to public ears, is arguably misleading – even discredited. Forecast Catchment-Rain-on-Grid (ie the impact of 200+ mm rainfall on Black Moss, Greystoke and Skiddaw in the same 48 hour period and its modelled effect downstream) could be a better way of connecting cause and effect, not least for negotiation with upland landowners. From the CH2M Report, it would seem that, in respect of flooding, the built environment to the west of the M6 is unfit for purpose,  even if it is considered possible to increase channel outflow. From Q&A at the competition launch event,  it would appear also that little or none of the 2005-2015 spend was assigned to catchment (ie rather than channel) process. Reading the 2009 Eden CFMP,  CaBA orientated Policy 6 is considered only for some 20% of the catchment with no evidence of ‘WwNP’ delivery. Furthermore there is no assessment of catchment reaches capable of Barrier-Storage, or indeed reference and quantification of  Catchment-Process.   Furthermore, access to national land-cover and hourly rainfall data is still restricted and  undermines catchment analysis. Indeed this censorship caused a pilot (40 km2 Fig 11) catchment study to stall prior to up-scaling to Carlisle’s c. 2000 km2. The pilot study, supported by national flood-risk teams and by the Rivers Trust and deriving from a 2011 study of a reach of the Upper Severn, had been in progress for several months prior to the competition.Flood-risk-gain is a goal for ALL vulnerable communities (ie it is not just a goal for Cumbria).

Figure 11:  The 40km2 Upper Severn pilot sub-catchment showing stream-network and floodplains


Whilst open-source modelling utilities fit well alongside open-data for hydraulic analysis of catchments under diverse ownership and stake-holding,  such code can be less versatile than commercial alternatives. In particular, for the exercises described above, it was not possible to input a moving storm to a 2D computational area. Before a national template is set out for productive 2D analysis, input and calibration data should be to hand so that grid resolution can be adjusted to give pragmatic run-times. Once a template is tested, a model can be set up for any catchment within a few days and run to test alternative outcomes within a few hours , even when large stream networks are loaded .

9.        SUMMARY and CONCLUSION  

The article is therefore

1.  A technical precis of methodology to promote adaptive land-use for flood alleviation (ALFA). A default budget of 10, 000m3 ‘blue-green’ modelled storage per km2 of catchment is commended – ie, functional short-term flood alleviation for 1 day in every 10,000 days.

2. Evidence that restrictions on data distribution  compromise the search for solution –  which in turn risks increasing flood risk to vulnerable communities.   Without calibration data,  analysis is incomplete and, for stakeholders, unconvincing.

Thereafter, the problems to be overcome include

1. Poor data sharing

2. Unbalanced investment in form over roughness

3. Unbalanced focus on risk awareness over engineered solution.

Significant proposals are

1. Open hourly rainfall historic archive to allow calibration and completion of catchment pilot projects nationwide

2. Genuine follow through of competition aspirations

3. Proactive focus on mechanisms to connect  smart floodwater storage with productivity by improved land drainage.