Taking into account the legislative scheme and the context in which it came into being, we consider that our task is to decide whether the Secretary of State was wrong to identify the water into which the land drains as polluted. ‘Wrong’ in this sense has the meaning given in Cook v General Medical Council [2023] EWHC 1906 (Admin) at [20]-[21], citing Waltham Forest LBC v Hussain & Ors [2023] EWCA (Civ) 733 at [64]. While we are not confined to public law errors, to disturb the decision we must disagree with it despite having accorded the special weight appropriate to a judgment of the body given decision-making power by the statutory regime, as well as having the relevant institutional and technical expertise.

Notwithstanding the simplicity of Mr Nash’s case as summarised at paragraph 8, this is a particularly complex appeal. In formulating the reasons for our decision we have applied the well-established principles summarised in the Senior President’s Practice Direction ‘Reasons for decisions’ of 4 June 2024. We have not identified every piece of evidence relied upon, nor every step in our reasoning; what follows is our conclusions on the main issues in dispute and a sufficient explanation of how we reached them to enable the parties to understand why they won or lost. Nor will a lay reader find an explanation of the relevant science, methodology and terminology involved in designating an NVZ, and instead will need the Methodology and the NVZ S522 datasheet to hand.

The regulations implement Council Directive 91/676/EEC on the protection of waters against pollution caused by nitrates from agricultural sources. There are two relevant judgments of the Court of Justice concerning the extent to which pollution must come from agricultural sources to justify designation: R. (Standley) & Ors (Environment and consumers) [1999] EUECJ C-293/97, and Commission v Belgium (Environment & consumers) [2005] EUECJ C-221/03. We adopt the summary of those cases given by the Upper Tribunal in PJ v Secretary of State for Environment, Food and Rural Affairs [2015] UKUT 207 (AAC):

The Environment Agency has also referred to authorities where around 10% was ‘significant’, albeit in the context of the Directive on urban waste-water treatment rather than nitrate pollution.

This catchment NVZ was expanded upon this designation to incorporate an additional river catchment at the outlet of the River Adur. The pre-existing NVZ was 71% of the size of the new NVZ. Figure 4 to the datasheet shows the location of the monitoring points and the Burgess Hill wastewater treatment works (“WWTW”), which discharges at Goddards Green.

The lowest designating monitoring point is F0002492. It recorded a 95%ile concentration of 22.39 mg/l TIN, significantly above the threshold of 11.3mg/l. Under the Methodology this gives a monitoring confidence level of 6. A time series for F0002492 is provided showing that the figure has increased since 1990.

No land use modelling was performed to assess the contribution made by agriculture, according to the datasheet because “the proposed NVZ is smaller than the minimum size [where] we can apply the land use model with confidence”. Yet, as observed by Mr Nash, the Methodology sets that minimum size at 20 km². The 2017 designation is 131.55km² and historically was 102.34 km², with the catchment being approximately the same size.

The Environment Agency has not continued to argue that the NVZ is too small, nor provided any explanation for how this was ever its approach. Instead, it says that omission of the SEPARATE model (explained in the datasheet) outputs from the datasheet “was an oversight”. We find it difficult to understand how this can be reconciled with its statement that the Methodology was nonetheless properly applied, as designation was explicitly done on the basis that no modelling data was available. The result was that when monitoring and modelling results were combined according to section 6 of the Methodology, a zero was given for land use model results on the evidence matrix. This raised the stated need to exercise discretion in all cases, “but particularly where the decision is informed by only one strand of evidence”.

We nonetheless consider what the Environment Agency now says justifies designation. It refers to land cover data showing about 60% of the catchment to be agricultural as corroborating the SEPARATE source apportionment model that estimates the agricultural contribution to the annual nitrogen mass load at 51%. Applying the panel’s expertise, and based on the cogency of each party’s explanation, we entirely agree with Mr Nash’s criticisms of reliance on land use data and SEPARATE alone to draw meaningful conclusions in this specific case. We make the following assessment:

We further agree with Mr Nash’s argument that SIMCAT modelling would have made a major contribution to the evidential picture. SIMCAT is described in the datasheet, as well as in more detail by the Upper Tribunal in PJ at [31] (although this refers to an earlier version). While NEAP-N does appear in the datasheet, for the reasons described in the Methodology at section 5.1 it cannot compensate for the lack of SIMCAT modelling.

A workshop was apparently convened to consider local factors arising from designation, but we have no specific information to show that any of the criticisms made in these proceedings were taken into account. Given the confusing picture as to whether land use modelling has been conducted and applied, we are reluctant to accept that the workshop was compliant with the Methodology at section 8. Absence of records or minutes is surprising given that the extension of an existing NVZ ought to be “the highest priority for discussion”, and it is likewise surprising that an effective workshop would have missed the lack of clarity on land use modelling.

Also tending against confident identification of the water as polluted is the analysis conducted by Hafren Water that target exceedance generally occurred at low flows, and not under average conditions. The Environment Agency has first taken this as arguing that a monitoring point is in a mixing zone, which is to say that it may not be representative of the overall water quality downstream from the point source. We agree with Mr Nash that this misses the point being made, which was restricted to the contribution made by the point source to the overall water quality. Second, the Environment Agency argues that Mr Nash’s suggested contribution of 80% by the point source is itself likely to leave room for a ‘substantial contribution’ by agriculture. We will address that point later.

At page 245 of the bundle the Environment Agency claims to be satisfied that “the monitoring points pass the series of checks outlined within section 4.2”. We have already set out our agreement with concerns raised by Mr Nash as to these, but we finally draw attention to the question “Does ammonia form a significant portion of total TIN as N concentration at the monitoring location?” If this data is available, we cannot see that any account has ever been taken of it.

We are acutely aware that the Environment Agency has the institutional and actual expertise to identify water as polluted, and that we must place weight on its assessment. This includes recognising that the assessment need not be perfect, and that the Environment Agency is entitled to use appropriate statistical methodologies to draw a reasonable conclusion: see our decision in Towse v Secretary of State for Environment, Food & Rural Affairs [2024] UKFTT 174 (GRC).

In this case the Environment Agency has based its decision on monitoring data alone, without undertaking the checks suggested by the Methodology or, we conclude, required by the circumstances. There is force in Mr Nash’s submissions that seasonal and low-flow sampling calls into serious question whether agriculture makes a meaningful contribution to pollution in this case, by indicating a much higher point source contribution than estimated by the Environment Agency. The Environment Agency’s estimate suffers from the multiple evidential frailties identified above, together with showing signs of not having been cogently formulated in the first place. We reluctantly conclude that there was no proper evidential basis upon which to conclude that agriculture makes a meaningful contribution to pollution in this NVZ. The making of even a 10% contribution, argued by the Environment Agency as still being significant, would still simply be too speculative based on the evidence available.

We are fortified in the above conclusion by the distinct possibility that better evidence might yield the same result. Everyone agrees that there is local agriculture that leaches nitrates, the question is simply to what extent it is dwarfed by the WWTW. If we had been provided with either SIMCAT modelling or ammonia results pointing towards significant agricultural contribution, this might have been enough to tip the scales towards designation. Nothing in this decision excludes the possibility that designation might not be the sustainable outcome of a future decision. In this case however, we find that the Secretary of State was wrong to identify the water as polluted on the evidential basis relied upon. The appeal is accordingly upheld under regulation 6(2)(b).