I write about the fate of the River Tarrant in Dorset, which now dries every summer as a result of increased groundwater abstraction. I also take issue with some of the ways in which water companies construct conceptual models that artificially limit the orbit of an abstraction’s impact to cones of depression and present this as the whole picture.

The March issue of The Field includes a feature by me on the poor little River Tarrant in Dorset. I first wrote about the unnatural drying of this stream 30 years ago after it vanished through its riverbed in August 1995, leaving 100s of trout dead in the caked mud.

Back then, although abstraction had started to increase, the total loss of the lower river was relatively rare. Now, it dries in its lower reaches every single year. That and the fact that, in spite of this annual, unnatural drying, the river is assessed by the Environment Agency as “supports good ecological status” for flow brings into the spotlight the state of play in our work to mitigate and reverse abstraction’s impacts on chalk stream health. 

The River Tarrant is a metaphor for the stasis in the regulatory system. The way this lovely chalk stream has been exploited and ignored over the decades is shameful.

To condense its story before I get into the techie stuff: the River Tarrant is a small chalk stream in Dorset, a tributary of the Stour. It is a slope-face stream, meaning it rises within the outcrop of the chalk downs which run north from Blandford towards Shaftesbury and flows south and east down the strike of the chalk slope. This makes it a classic, natural winterbourne in its upper reaches. The river length rises and falls as the groundwater in the unconfined, outcropping chalk aquifer rises and falls. It can dry, also naturally, in its lower reaches too, where there is a fault line in the chalk running perpendicular to the valley. As groundwater levels fall through the summer and especially in droughts, the lower river loses water to the ground.

However, the extent – by that I mean the physical and temporal extent – to which the Tarrant is a natural winterbourne is the subject of some debate. Historical evidence strongly suggests that the Tarrant is only a natural winterbourne in its upper reaches downslope as far as Tarrant Launceston, where old maps and LiDAR indicate that a meandering channel, suggestive of perennial flow conditions, begins.

Research by Dr David Solomon for a report recently commissioned by the River Tarrant Protection Society (more on that very comprehensive report, primarily authored by John Lawson, later in this post and a link at the end) has shown that the River Tarrant did occasionally dry further down the valley in the past, before groundwater abstraction was much developed, but so rarely as to make the local press when it did so. 

It dried in Tarrant Monkton – a little further downstream than Launceston – in 1929 and 1933, both years of severe drought. There is no historical evidence of it regularly drying up in the lower reaches, however, albeit in 1970 Dr Stanton, a hydrologist, wrote: “In exceptionally dry years the river is dry just above the confluence with the Stour”. We know it dried in 1976 and it clearly shrank away to almost nothing in 1989 as noted by the National Rivers Authority, by which time the river was being more heavily abstracted.

The River Tarrant once supported several water-mills, at least five of which were Domesday mills. While the Domesday book did record “winter mills” molinum hiemale, none on the Tarrant are so described. All of the Tarrant mill sites are downstream of Tarrant Monkton, lending more weight to the idea that our ancestors preferred to construct mills where flows were reliable. 

The mill site at Tarrant Crawford Abbey seems particularly anomalous to the modern-day ephemeral flow regime of the lower river. This was a substantial mill with a dedicated water-retaining structure and leat, sited very close to the confluence of the main Stour and another watermill – Keystone Mill – that would have been a better option and wholly reliable if the Tarrant had tended to dry.

Meanwhile, Atlantic salmon persist in behaving as if the Tarrant is not a natural winterbourne. They are mentioned in 19th-century issues of The Field as regularly coming up the river to spawn, and there are numerous records of salmon in the river in recent years, including of salmon parr in the now all-too-frequent and necessary fish rescues. As is well known, salmon tend to spawn where they were themselves spawned: they imprint on the chemical signature of the water and stones in which they hatch. That genetic memory and behavioural fidelity – which goes back tens of thousands of years – ought to tell us something: salmon don’t spawn in natural winterbournes. Any salmon that chose natural winterbournes in the past will not long have succeeded in passing genes and progeny to the next generation.

All of this and more – for example, the memory of ancient locals interviewed by my wife Vicky in 1995 – strongly suggest that the Tarrant is naturally a perennial chalk stream downstream of Tarrant Monkton, and that while it might naturally have diminished in its lower reaches in drought years, it didn’t tend to dry up. There is no documented record of the lower river drying before groundwater abstraction began.

So, how come it now dries in the lower reaches almost every year? 

In 1995, when we rescued all those fish and failed to rescue so many more, the drying came as a shock: I had known the stream since 1987. It had dried in 1976— a drought comparable to 1929— but kept flowing for 99.9% of the next 19 years, with only the lowermost kilometre briefly drying in 1989. I fished it often in the early 1990s when two- to three-pound trout were not unusual. It was my best-kept secret. And yet, in eight of the past nine years, the lower river has been dry for between 46 and 120 days in the year. 

The Tarrant, especially the lower Tarrant, has changed. Why?

“Just one more thing …”

The answer, I believe, lies in the telling fact that the one year— 2017— in the past decade when it didn’t dry in the lower reaches was the one year when Wessex Water shut down its Black Lane pumping station, which lies in the next valley over by the Pimperne Brook.

If Columbo were a natural history detective, he would at this point stop and say, “Just one more thing …” then ask how much it had rained in the preceding winter. Not much, is the answer. The spring of 2017 was a disaster for chalk streams: I was commissioned that year to take photos for WWF of all the chalk streams near London that were bone dry, also largely because of over-abstraction. 

And yet, this one year of all years, the lower Tarrant kept flowing?

Wessex Water claims that the Black Lane pumping station cannot impact flows in the Tarrant, not least because there is a hill in the way. 

And yet, water companies are often the first to point out that subterranean catchment boundaries are not the same as surface watershed boundaries and are dynamic too, capable of naturally migrating as groundwater levels rise and fall. For example, the aquifer under the true Winterbourne in that area, the ephemeral stream that confluences with the Stour on the opposite side of the valley from the Tarrant, and tellingly called “Winterbourne,” is also connected to the Bere Stream in the neighbouring Piddle catchment. In numerous meetings, this natural phenomenon of dynamic groundwater boundaries has been spelled out to me, as water company hydrogeologists caution against the simplistic notion that reducing abstraction in a given valley will lead to restored flows in that valley. 

And yet, conversely, the same water companies have also argued for the existence of glass walls between one chalk valley and the next. The Lea, for example, is very heavily abstracted, but apparently, that impact simply can’t translate across to the neighbouring Ver or Mimram. The abstraction in the Beane simply can’t affect the Ivel, or vice versa, etc. Like those watershed boundaries, the argument is dynamic and tends to suit the point being made.

As John Lawson’s report shows, there are several abstraction pumps in and around the Tarrant, and groundwater abstraction has crept up slowly over the years. The one source within the physical, surface catchment – Stubhamptom – is relatively small and has been running since the late 1950s. But there are much bigger pumps at the aforementioned Black Lane (just over the catchment boundary), and Shapwick (close to the confluence with the Stour), and Sturminster Marshall (a little further down the Stour valley), as well as Corfe Mullen, which is slightly further afield.

The aggregated impact of just the former three climbed from about 5 ml/d in the early 1970s (by when Stanton observed that the lower river dried in exceptional droughts) to about 10 Ml/d in the early 1980s, and peaked at about 15 Ml/d in the 1990s, when I was forced to rescue those trout in the 1995 drought. 

By the early 2000s, the aggregated abstraction of these three pumps had fallen back to between 5 and 10 Ml/d. Then in 2016 and 2017, abstraction was reduced to about 6 Ml/d and in 2017 for a short period, to almost zero. By the end of 2017, it had ramped back up steeply to about 15 Ml/d, as the Wessex Water grid came online and “underutilised” sources in the Stour valley were used to aid reductions in the more protected Avon catchment.

Since then, the Tarrant has dried every year.

Wessex Water maintains that their sophisticated tests and modelling show that the Black Lane and Shapwick pumps don’t affect flows in the Tarrant. 

Their case rests on the analysis of what are called switch-off tests, when they shut down pumps and look for the rebounding impact on groundwater levels in the surrounding area. This helps to define the ‘zone of influence’ (a misnomer – see my argument below) or more accurately the ‘cone of depression’, a funnel-shaped depression in the water table: a bit like the vortex that descends to a plug hole. 

The cones of depression of either Shapwick or Black Lane do not reach the Tarrant Valley. At least not enough to influence drying, Wessex Water says, adding that an impact is exceptionally unlikely because the chalk between the valleys is less ‘transmissive’ (i.e. the movement of water through the chalk is more restricted) than within the valleys. 

Their findings have been accepted by the Environment Agency.

Bringing sticks to a gunfight.

As a community of campaigners, we have struggled to fight the over-abstraction of our chalk streams in no small part because we don’t easily understand the way groundwater drives river flows or the ways groundwater abstraction impacts river flows. As a perfect example, the idea of lessening abstraction in summer – which more or less every lay person still thinks is the right time to reduce abstraction (or the worst time to increase it) – made it all the way from being a collective NGO recommendation to becoming an Ofwat incentive, when it is generally the least effective time to limit groundwater abstraction, because the impact of the reduction is not generally felt at the time it is made.

We are easily baffled, firstly by the sometimes counter-intuitive ways in which groundwater and flows respond to abstraction and secondly by the sophisticated arguments of the water companies or their consultants. If we were universally better informed, we might do a better job of combating the radar chaff thrown in front of our heat-seeking ire.

I knew nothing about all this groundwater stuff this time in 1995. I just had a strong hunch that abstraction must be to blame because it is logical to conclude that if you take water out of a natural river system, it must impact flows from that system. That logic still holds, by the way. But there are complexities to this mephistophelean science, and it is best we try to understand them.

Ob-fudge-scation

Three decades ago, water companies, the river authorities, and their consultants generally tried to fudge the impact of abstraction on river flows. In systems as complex as spring-fed rivers, it was easy and too tempting to dissemble and deflect.

As just one example, in the 1990s, Atkins wrote a report on abstraction impacts in the River Kennet. They stated:

• “There is no clear evidence that groundwater levels over the catchment have been affected by groundwater abstraction.”
• “There is no strong evidence for any change in the distribution of minimum groundwater levels over the catchment…”
• “Ongoing abstraction does not have a cumulative effect… Both groundwater levels and surface flows stabilise at a lower level…”

That last statement is particularly disingenuous, deliberately designed to confuse the lay reader. It allows for the stabilisation of groundwater at a lower level, having previously stated that there was no evidence this was occurring and then conflates the idea of ecological damage only with cumulative groundwater mining. 

Notice the phrasing, the refrain of “no strong evidence” as a caveat? We could characterise this as systemic minimisation and an insistence on evidential uncertainty. It is designed to disable the protest. People had noticed the drying up of chalk streams, but countering this kind of specious guff was very challenging unless you had a PhD in hydrogeology and didn’t work for a water company, the agencies or one of their consultants. Which is no one with a PhD in hydrogeology.

In fact, the NGO movement has been turning up at gunfights armed with sticks for several decades. Only John Lawson – who started out battling that exact same Atkins report and the unsustainable abstraction of the River Kennet where he lives – has made a really effective stand here on behalf of the NGOs. He deserves a sainthood, let alone a knighthood. One of the cleverest people I know and a very good, highly qualified engineer, he and his work are routinely dismissed and patronised by the hydrogeological cabal. 

Computer says “we don’t know”

While the socio-economic benefits are still just as real today as in the 1990s, water companies no longer fudge the theoretical case quite so brazenly. In practice, however, and on a case-by-case basis, they argue the toss every single time. They have very strong economic motivations to do so. And they manage quite successfully too, because as our knowledge of these systems has become more sophisticated, the ability to hide action behind uncertainty has increased.

Our understanding of the chalk aquifer has developed over recent decades, so that what was once seen as three basic strata is now understood as many different strata comprising different chalks of varying age, hardness and structure, intersected by layers of less transmissive horizons, riven through with karstic flow-ways. About this infinitely complex underworld, the more we know, the less we know for sure.

Groundwater modelling has also become much more high-tech over recent years. We now have conceptualised, computerised aquifers, with gazillions of cells imitating the aquifer properties, which can run numerous scenarios and impacts. And yet – so it seems – we can never know enough to state anything for certain. Thus we have replaced the fudging with an insistence on preceding any action with full knowledge, attained only via immensely sophisticated modelling that is privy to water companies and the Environment Agency, the two players who are motivated to maintain the status quo. 

There’s too much money at stake for water companies. Too much work at stake for consultants. The socio-economic value of groundwater renders the precautionary principle a pipe dream. eNGOs and protesters are still impotent and hobbled by their relative lack of expertise and resources.

And thus every chalk stream in which the abstraction pumps run is a new arena for endless investigation.

So, it is with the Tarrant. The complexity allows the can to be kicked. Again and again and again.

Newsflash!

The River Tarrant Protection Association commissioned John Lawson to write his report well over a year ago. It contains a very cogent case for casting doubt on the joint Wessex Water and Environment Agency conclusion that the river Tarrant is unaffected by local groundwater abstraction. It includes a pragmatic proposal for how to relieve pressure on the River Tarrant and asks, at the very least, for the Tarrant to be properly included in the next round of AMP investigations.

There is a great deal of historical and anecdotal evidence in the report, but the bulk of the report covers John’s analysis of the formal modelling and his own, simpler, but empirically accurate modelling methodology (which hydrogeologists tend to dismiss). This shows that the recent actual flows only fit the modelled outputs when the pressure on the regional groundwater table, including that exerted by the pumps at Shapwick and Black Lane, is taken into account. 

John’s conceptual model – I will try to unpack it below – underpinning his numerical model is based on the idea of the aquifer as a regional entity, that allows for the inclusion of neighbouring catchments. In theory, this inclusion could radiate as far as the boundaries of the aquifer. In practice, John does limit it to keep the data input manageable. John’s conceptual aquifer is – I will argue – theoretically accurate, but hotly denied by any water company that encounters John’s work.

John shared his report with Wessex Water and with the Environment Agency last September. Christopher Greenwell, the EA’s water resources strategic lead, replied a few weeks later, promising a detailed review of the report and a meeting to discuss, with an update by the 12^th December 2025. 

The update never came and had to be chased. The EA replied in mid-January to say that the groundwater team was reviewing the report. In March this year, Christopher Greenwell replied, saying that his groundwater colleagues would commission a review (so had they been reviewing the reports in January or not?) and a comparison of John’s groundwater approach, as well as the methodology used in the Wessex Water / EA chalk basin model, “to be carried out by an external party to avoid bias and give greater confidence to all parties”. Asked for a bit of clarification, Christopher confirmed that the review would not concern itself with anything in the report other than a comparison of the conceptual models.

So, John’s David to the Wessex Water / Environment Agency Goliath then? With the review conducted by a third party who is fully objective? We shall see.

In the meantime, back to the complexity … and those conceptual models.

Groundwater for Dummies

It really isn’t that complex. Or rather, any complexity that exists overlies a basic, unarguable simplicity. Groundwater abstraction from a spring-fed system will lower stream flows from that system. There’s no real, honest debate over whether it does, or even by how much it does (over time, by the same amount as the abstraction). The debate, such as it exists, is over where the impact is felt and more challengingly, when it is felt. 

This provides all the wriggle room needed, however.

Time

The role of time can be difficult to get your head around, but – at first – let’s imagine the underground aquifer as a large surface lake, with many streams and rills running into it from the surrounding hills, but only one river draining it. Loch Tay, for example. If we eliminate evaporation, and any loss of water through the lake bed, the amount of water flowing out of the lake must equal the amount coming in.

But the correlation between inflow and outflow is not continuously simultaneous. When it rains it will take time for the water in the lake to rise and force more flow to leave the lake. And when it stops raining and the lake level is high, it will take time for the lake level to fall and the outflow to diminish, until it once again balances the inflow.

The system is in a state of dynamic stability and equilibrium. Inflows, lake level and outflows all vary all the time, as rain comes and goes, but ultimately inflows equal outflows over time.

Once that is grasped, it only takes a bit of imagination to increase the size of this mind’s-eye lake and appreciate that time will lag to a greater and greater degree between the impact of inflow on lake level and outflow as the lake gets bigger. A headwater stream in flood rushing into Lake Windermere (6 square miles) will force up the level of the lake and the outflow more quickly than it would rushing into Loch Neagh (150 square miles).

If we now add another form of outflow (let’s say we drill a great big pipeline through the hill and divert a large proportion of the lake’s recharge into a neighbouring valley), the former outflow (the river draining the lake) will have to go down. 

However – this is key – it will take time for the new outflow to drain the lake down to the point where it captures that former outflow. 

But capture it, it will.

How long it takes depends on the relative size of the new outflow and the size of the lake. But, no matter the size of the new outflow, there must eventually be an impact on the former natural outflows that exactly equals the new outflows, and the mechanism for that impact is the lowering of the lake level. Even if the abstraction is small, maybe only 5% of the former discharge, and the new outflow is a long way away from the natural outflow, it will be felt, in time, once the lake has lowered by the commensurate amount. 

That surface system is fairly easy to get your head around.

Going underground

The complexity comes when you turn the surface system into a groundwater system. The fundamental concept of the water balance remains exactly the same, however, meaning that you can’t add an extra form of discharge (abstraction) without lowering the former natural discharge.

Historically, water companies have argued that you can and have used obtuse arguments to say as much. Setting the precedent for this, in the mid-19th century, the London Water Company started to abstract water from the River Gade, a chalk stream still bedevilled by abstraction. When the millers on the Gade complained, the London Water Company recruited scientists to argue that their abstraction didn’t and couldn’t lower the river flows. They were taking water, so the Victorian boffins argued, from a limitless well so deep underground that it wasn’t connected to the surface system, something the millers were able to show was nonsense. The courts found in the millers’ favour but kicked any reparation into the long grass of the future, a debate and conclusion which is eerily familiar in the Chilterns to this day. 

Knowing therefore that when the lake becomes an aquifer, the impact of the abstraction remains non-negotiable, the components of when and where do become harder to pin down.

When and where

We have already seen in the lake analogy that the timing of the arrival of impact at the outflow varies according to the size of the lake because the new outflow must reduce the water level in the lake in order to have an impact on the former outflow. 

This is the same with an aquifer. A groundwater abstraction must first lower the water level in the aquifer in order to then capture former natural flow from the stream. Thus, in the first instance, an abstraction – depending on where it is sited – may have an almost undetectable influence on former stream flows while it takes water from the volume stored in the aquifer. Eventually, however, the abstraction will reduce the storage and lower the groundwater levels enough to establish a new dynamic balance in the level of the aquifer, after which ALL of the abstraction is captured from the former natural flow. This split over time between water taken from storage and then water taken from flow is illustrated in the diagram below.

The key idea is this:

The means by which groundwater abstraction captures stream flows is by a reduction in the storage of the aquifer and a lowering of groundwater levels to establish a new dynamic balance.

This lowering of the water table changes the hydraulic head across the entire aquifer, which is what lowers the natural flow.

This basic and inalienable truth was defined by Theis in 1940, but is befuddled all the time by water companies as they smudge the issue of whether any particular abstraction is having an impact on any particular stream.

Wessex Water, for example, argues that relatively short-duration switch-off tests (mentioned earlier) define the boundaries of the cones of depression in their analysis of abstraction impacts around the Tarrant and then misleadingly conflate these boundaries (partly by now renaming the ‘cones of depression’ as ‘zones of influence’) with the limit of the influence of the particular, individual abstractions.

They argue this, in spite of the fact that the impact of any given abstraction extends – in theory – infinitely from the location of the abstraction, its only real boundary being time. The cone of depression is simply the measurable drawdown around a pumping well, forming a gradient that drives water to the wellhead. The edge of the cone is not a physical boundary at all, it is a detection threshold where the drawdown is less than any possible measurement error. 

In short, the use of the cone of depression is a water resources construct to help simplify modelling and analysis.  In reality, the system actually responds as a continuous, radiating field of change in hydraulic head.

How groundwater abstraction actually affects natural systems.

The so-called zone of influence is actually only the measurable depression, one that is imposed on a much more widespread reduction in the dynamic level of the water table, that reduction exactly equating, over time, to the reduction in head needed to capture former natural discharge and establish a new dynamic balance. Groundwater abstraction alters the natural distribution of hydraulic head everywhere in the system.

Although the system’s boundaries are – in theory – only limited by time, in practice they may effectively be defined by other factors (and Wessex water contends that these factors are relevant). These could include:

• impermeable layers – for example thick layers of clays and flints between layers of chalk.

• aquifer boundaries – the chalk aquifer obviously ends where the chalk ends. Although it may well connect with other aquifers, the boundary of the chalk may mark an effective boundary for the purpose of analysis.

• recharge boundaries – large rivers or lakes / wetlands can supply the water being demanded by the abstraction and therefore limit the effective growth of the water-table reduction. It’s possible the River Stour floodplain is a recharge boundary in the case of the Tarrant.

• changes in transmissivity – this is what Wessex Water rests much of its case on, especially the transmissivity under the catchment boundary.

Those effective, possible limits notwithstanding, groundwater abstraction is actually a time-dependent redistribution. As I have shown, it can take a long time for abstraction to capture former natural flows.

Stage 1. Duration – days, weeks or months.

The water demanded by the pump comes almost entirely from storage. The impacts are localised and at this point they are indeed mostly felt within the cone of depression.

Stage 2. Duration – months to years.

The abstraction begins to impact nearby streams and wetlands. This is when the abstraction starts to impact hydraulic gradients in an ever-widening orbit, taking less and less water from storage and increasingly more from changes in the boundary flows.

Stage 3. Duration – long-term new equilibrium, years to decades.

This is the stage beyond which the groundwater abstraction must, according to immutable laws of physics, take its water from the capture of former natural discharge. Most of this will be a reduction in the former groundwater discharge to the chalk streams. Some will also come from “induced recharge”, driving leakage from the stream bed (this is absolutely what is happening in the lower Tarrant), and from lowered evapotranspiration (because the water table is lowered). 

Over time, the depletion from former natural discharge approaches the pumping rate, until almost all is taken from the stream system.

Just as it can take a long time for abstraction to lower groundwater levels and capture former natural outflows, it must therefore take a long time for a reduction in abstraction to lead to fully recovered groundwater levels and a full restoration of former natural flows.

Switch-off tests lasting only a few days, weeks, or even months will do nothing more than allow the cone of depression to refill and so help define its boundary. This is not the same thing as the limit of the boundary of impact.

The important point to underline in the case of the Tarrant is that groundwater divides are based on hydraulic head and that divide can move with abstraction pressure. Abstraction can also steepen the gradient of the hydraulic head on one side of the divide relative to the other, causing water to flow towards the pump, relative to the natural system

This means that theoretically, it doesn’t necessarily matter if the Black Lane abstraction is in the neighbouring valley, or that there is a hill in the way, or even that the chalk is marginally less transmissive under the hill. These pumps can very much impact flows in the Tarrant by altering the hydraulic boundaries a long distance from the pump in ways that might not easily be detectable except as diminished stream flow.

Short duration switch-off trials

The switch-off trials that were used to define the boundaries of the cones of depression around the Tarrant lasted only several weeks. 

The Black Lane pump was shut down between 9^th June and the 15^th August 2016. Continuous pumping at a high rate then followed for the next month until September 15th. Both the recovery of groundwater levels and the ensuing drawdown and second recovery were analysed by Wessex Water and used to infer that the so-called zone of influence didn’t extend to the Tarrant. Adding to the weight of evidence was an analysis of groundwater gradient undertaken in 1985 which showed steep groundwater contours under the Pimperne – Tarrant interfluve, suggesting that the chalk here is of lower transmissivity.

The Shapwick pumps were shut off for only eight days in July 2017. They studied groundwater responses around Shapwick and up to and in the Tarrant valley. There was a visible rebound near the pump, but no discernible rebound on the Tarrant interfluve, or the Tarrant valley. Wessex concluded that the river gravels in the Stour valley supply most of the water to the Shapwick pump. 

In both cases, a midsummer switch-off of nine weeks, let alone of only eight days, cannot reasonably be deemed long enough to allow regional groundwater levels to recover, when true recovery would need one or two years or longer.

It seems to me that Wessex Water and EA’s logical reasoning is flawed.

The degree to which other limitations – such as lower transmissivity under the interfluve, or a possible recharge boundary in the Stour gravels – have a controlling impact is debatable on a case-by-case basis, but one needs to get the evidence and reasonable conclusions in the correct order.

In other words, it is not sustainable to:

• subject a natural system to abstraction 
• observe a subsequent change in flows in the River Tarrant
• confidently attribute that change to some other as yet unknown cause (Wessex Water posits a more leaky river bed than in the past)
• confidently state that the impact cannot be caused by abstraction because 
• the boundaries of a measurable zone of drawdown do not reach the Tarrant
• and short-term switch-off tests do not show a recovery in borehole levels that are some distance from the pump

This is confirmation-bias reasoning.

What do we KNOW in theory?

We know for certain that abstraction must – eventually – cause a commensurate reduction in former natural discharge, mostly stream flows.

We know that it can take a long time (months to years) for abstraction to remove storage and lower water-table levels to the extent that it captures those stream flows.

We know – vice versa – that it can take a long time (months to years) for the cessation of abstraction to lead to a full recovery of water-table levels and former stream flows.

We know that there is no theoretical limit to the distance over which an abstraction’s impact may be felt.

We know that factors such as transmissivity may create an effective limit to that distance. 

What do we KNOW in practice?

We know that aggregate groundwater abstraction in and around the valley has climbed steadily since the 1970s.

We know that before 2017 the lower River Tarrant dried in 1976, and in 1989/90 and 1995 (all severe droughts).

We know that since 2015 the lower River Tarrant has dried every summer except one (none of which were severe droughts).

We know that the recent drying outside droughts has coincided with the highest aggregate abstraction rates from the three pumps in and close to the valley, over 10Ml/d from Stubhampton, Black Lane and Shapwick, and over 20Ml/d when including Sturmister Marshall. 

We know that the one year in the last decade when the lower Tarrant did not dry – 2017 – was both a very dry year AND the one year when the Black Lane pump was switched off for several months and the aggregate abstraction rate fell right back only a few Ml/d .

We know that there is no documentary evidence that the lower Tarrant regularly dried before the era of groundwater abstraction, although there is such evidence for drying in the upper river, which one would expect to dry naturally from time to time. 

We know that there is evidence to suggest the lower river flows were reliable before the era of groundwater abstraction. 

Occam’s Razor …

states that when faced with competing hypotheses, the one with the fewest assumptions is probably the best.

The River Tarrant Protection Society’s hypothesis is that the drying of the lower Tarrant is caused by the local groundwater abstraction, especially its increasing rate since 2017. 

Wessex Water’s hypothesis is that abstraction cannot impact flows in the Tarrant because the chalk under Tarrant Pimperne interfluve appears to be less transmissive and because short-term shut-off tests show that the measurable cones of depression do not reach the Tarrant valley, even though we know that cones of depression do not mark the limit of potential influence. They posit that the drying of the lower Tarrant may be caused by an unexplained and un-evidenced increase in the leakiness of the riverbed and that the lack of drying in 2017 is explained by some heavy rainfall in the late summer. 

I truly hope that the review commissioned by the Environment Agency proves to be genuinely independent. 

Wessex Water is not objective. Their regional water plans and the use of Stour sources to relieve abstraction pressure in the highly protected Avon SAC are upset by the idea that abstraction should also be reduced in and around the River Tarrant.

But the Tarrant is a chalk stream and a nursery stream for critically endangered Atlantic salmon. The Environment Agency would not be doing its job properly if it ignored clear evidence that abstraction was damaging salmon spawning and nursery habitat.

The River Tarrant Protection Association is asking that the abstraction impact on the River Tarrant be properly investigated, not dismissed, or filibustered and that their pragmatic proposal for a solution to the problem be considered.

JOHN LAWSON’S REPORT CAN BE READ HERE