Scientific monitoring proves leaky dams are a success at Hardcastle Crags

By Stuart Bradshaw BSc(Hons) MSc DIC CEng MIStructE

The Hardcastle Crags Natural Flood Management (NFM) interventions have been proven a successful test project by scientific monitoring, showing that NFM does work to help alleviate flooding! 

Therefore, (as we have always known, but it’s nice to have proof) the more NFM and SuDS interventions we can implement, and the more widespread we can make it, the better; and the less devastating future flood events will be.

In this blog post, Stuart Bradshaw discusses the detailed findings from his important work in monitoring our ‘leaky dam’ interventions at Hardcastle Crags NT. The results show that our volunteers have helped to slow the flow of water, as it heads towards the oft-flooded settlement of Hebden Bridge, and the townships beyond, in the bottom of the Calder Valley. 


Following the devastating floods of Boxing Day 2015 in the Calder Valley, in the spring of 2016 Slow The Flow put together a proposal to undertake a Natural Flood Management (NFM) project in Hardcastle Crags, north of Hebden Bridge.  The project proposal comprised the construction of leaky woody debris dams of various types and sizes alongside restoration of 48 ha of woodland at Hardcastle Crags National Trust, which is designated as PAWS (Plantation on Ancient Woodland Sites).  The proposal was successful and work began in April 2017, under the management of the National Trust (NT) and the Environment Agency (EA), with funding from Defra.

Photo credit Victoria Holland – National Trust

The vast majority of the 800+ leaky dams constructed there now are in feeder streams that flow into Hebden Water and Crimsworth Dean Beck. Measuring the efficacy of these measures was an important part of the project proposal, because whilst it is a significant ‘slowing the flow’ intervention in and of itself, the idea behind it was to provide proof and incentive for wider rollout of such interventions, which Slow The Flow believes should be widespread.  Monitoring for the NFM project was funded as a result of an NT/STF proposal in autumn 2017 and subsequent partnership agreements agreed October and December 2018.

Small Leaky dams

The NFM project proposes adopting typically a 30% thin of non-native conifers and species such as Beech and Sycamore, to allow the woodland flora to recover.  The project hypothesis set out in 2016 was that, working in conjunction with the woody debris dams, materials for which are derived from the woodland thinning, a restored and vegetated ground flora would increase surface roughness, intercept precipitation, impede surface run off, reduce erosion and sediment transport, and effectively slow the rate of water flow into the system of ditches and natural water courses throughout the woodland.  

The stream courses flowing through Hardcastle Crags, Hebden Water, Crimsworth Dean Beck, and on the south eastern boundary at Pecket Well Clough, are all tributaries to the River Calder which flows broadly from west to east through the Calder Valley in West Yorkshire.  The background to the origins of the project relates to the long history of fluvial and pluvial flooding in the Calder Valley, which is surrounded by steeply falling hillsides. Consequently, the River Calder and its tributaries respond rapidly to heavy prolonged rainfall – it is described as a ‘flashy catchment’.  In 2012 the towns of Hebden Bridge and Mytholmroyd were flooded twice in the summer floods of that year.  Over Christmas and Boxing Day 2015, Pennine areas had over 60mm of rain fall in 24 hours, and some locations had over 100 mm. The Calder Valley suffered one of the most significant flooding events in recent times.  2,781 homes and 4,416 businesses were flooded all along the Calder Valley, causing unparalleled damage. 


The monitoring proposal involved placing pressure transducers in the feeder streams one at the top and the other at the bottom, to allow the rise and fall of flow in those streams to be measured. 

In summary, initially there were two instrumented streams: an experimental stream with 9 No. leaky dams over an 80 metre reach (Stream 2) and a nearby control stream without leaky dams (Stream 3).  Later a further instrumented stream containing 16 No. leaky dams was added to the project (Stream 1). 


The pressure transducers record the stream flow height in real time and allow a hydrograph to be drawn at each monitoring location, which are at the top and bottom of each stream.  A typical hydrograph is shown at Figure 1 which shows the hydrograph of the top location versus the hydrograph at the bottom of the stream after passing through 9 No. leaky woody debris dams.  Figure 2 shows similar curves for the adjacent control stream.

Figure 12 shows two delays of 1 hour and 1.5 hours observed at two peaks for Stream 2, compared with no delay for either peak at the control Stream 3 (Figure 13); this was found to be typical in the data recorded throughout the project.  A delay is defined by measuring the stream flows at the top and bottom of each reach against time and where the hydrograph peaks, determining the time difference between those peaks. The data set records delays ranging between 15 minutes and 105 minutes for the experimental stream (leaky dams) against no delays for the control stream (no leaky dams).

9 leaky dams, over an 80m stretch, results in 1.5h and 1h delays of the peak flows, compared to the parallel control stream nearby

Table 52, which is colour coded, provides a summary of the stream flow data for the whole project.  This shows the data in green which indicates a delay at the peak is prevalent on the Stream 1 and 2 data, in comparison with data in yellow (no delay) which dominates the Stream 3 data.  

Control stream 3 often shows no delay in peak flow from top to bottom, the streams with leaky dams generally show a positive time delay

An attempt to develop a correlation with the apparent rainfall intensity, represented by the rate of change of the rise on the rising limb of the hydrograph with delay time is shown at Figure 53, but no correlation is apparent when looking at the complete data set.  The only observation that can be made is a cluster of data points from Stream 1 and 2 that correlate with a delay at the peak in comparison with those from Stream 3 which do not, this is effectively colour coded Figure 52 data presented graphically.

Times of the peaks differ markedly between the streams, only in four cases do they broadly coincide on 7/11/19, 19/01/21, 4/7/21 and 29/7/21.  It is apparent that there are implications for the larger catchment size draining into Stream 3 against Stream 2 which may be the reason for the differences in the timing of the peaks along with the influence of effluent flow from groundwater.

Figure 54 offers an association between delay time at peak, with time to peak, with Streams 1 and 2 tending to be above 10 hours’ time to peak and Stream 3 less than 12 hours.  This suggests that longer developed periods of rainfall, possibly on a dry catchment, are associated very approximately with a delay time at peak in excess of 30 minutes for the streams containing the leaky dams. Higher intensity rainfall, possibly on an already saturated or partially saturated catchment, produces a shorter time to peak, more typically associated with no delay for Stream 3 or a delay of 15 minutes or less for Stream 2. 


These observations lead to the conclusion that on this project a succession of leaky woody debris dams placed at regular intervals in stream courses have successfully produced delays to the flood peak as measured downstream of the interventions, compared with the measurements upstream of the same interventions.  The delay time has ranged on this project to between 30 and 105 minutes and is more commonly observed on rainfall events that produce a time to peak in excess of 10 hours.  

Leaky dam


The effects of rainfall intensity and catchment wetness have not been investigated to any extent other than simplistically looking at the rate of change on the rising limb of the hydrographs.  Antecedent conditions will play their part along with the degree of soil saturation.  With the appropriate budget and instrumentation these effects could be studied further.

The effects of leaky woody debris structures on a catchment wide scale have not been measured during this project but the time delays at the peak observed here could be incorporated into hydrological computational models for the overall catchment and the results observed.  

Extending the project outside of the National Trust boundaries onto the farmland and moorlands above would test the theory that additional interventions higher up the catchment increase the delay times at peak observed here.

Sufficient data has now been collected for Stream 3, the control stream, for this to be considered representative for a “before and after control impact (BACI)” research study.  The way the project was implemented and funded did not allow for any monitoring to be implemented for the streams prior to installation of interventions.  By installing similar leaky woody debris structures in Stream 3 at this stage, would allow the effects of these structures on stream flows to be observed and compared with past data prior to installation of those interventions.

Stuart Bradshaw BSc(Hons) MSc DIC CEng MIStructE