Watershed‐scale mapping of on‐site sanitation‐related environmental issue zones—Case study of Lys River (France and Belgium)

In application of the European Water Framework Directive, the water development and management plan, Schéma d'aménagement et de gestion des eaux (Sage) for the Lys (Lily) watershed area sets out general objectives for the use, development, and quantitative and qualitative protection of water resources and aquatic environments. The Lys Sage issues include the pollution of aquatic media and a provision for reducing the impact of discharges from on‐site sanitation devices. The prioritization of public authority action on on‐site sanitation requires the definition and identification of zones with environmental issues. The objective is to identify priority areas for the rehabilitation of noncompliant on‐site sanitation devices in case of a proven risk of pollution. These zones have been identified and mapped following the development and application of a methodology based on current regulations, feedback, and an innovative approach. It is based on the risk assessment of on‐site sanitation. To do this, a hazard (source hydraulic flow) was crossed with the vulnerability of the environment (target flow impact on stream quality). In addition to vulnerability assessed by the chemical quality of watercourses, the concept of zones of ecological interest was introduced.


| INTRODUCTION
In areas where centralized setting of a wastewater treatment plant connected to a network of sewer pipes would be expensive and not efficient due to low density of houses, the alternative practice both feasible and economical is on-site sanitation (commonly denoted OSS, but in French and in this paper: ANC).This domestic wastewater management mode may be executed by different standard methods all ensuring three functions-collection, purification, and evacuation of wastewater near the house (Crites & Tchobanoglous, 1998).
Due to the high cost involved in sewered sanitation, ANC has emerged as the preferred mode in cities experiencing rapid urbanization, and this practice has put severe stress on groundwater and river quality (Diaw et al., 2020;Fouché et al., 2019;Pujari et al., 2012).OSS systems (e.g., septic tanks) are widely used in low-and middle-income countries (Odagiri et al., 2021;Otaki et al., 2021;Thaher et al., 2022).The widespread prevalence of unimproved sanitation technologies has been a major cause of concern for the environment and public health (Berendes et al., 2020;Cheng et al., 2018;Dasgupta et al., 2021).However, many inhabitants of unsewered rural and periurban areas in developed countries (Daudin et al., 2022) also use on-site wastewater treatment systems (OWTS).In the United States, up to 36% of the population (20% of all residential homes) is served by OWTS (Siegrist, 2017).The challenge is also considerable in Europe, where 150 million people depend on ANC, representing one-third of the total population; for instance, in Ireland.In France, 5.4 million homes resort to ANC, that is, 18% of the population, and in Australia, nearly 20%.This paper assesses the potential for river pollution by ANC in rural or periurban areas using a transferable methodology illustrated here in a case study in France.
The most commonly used OWTS type in regions developed with a water adduction network is based on an anaerobic septic tank with an outlet to spreading drains embedded in infiltration trenches or a percolation bed within the subsoil.The septic tank effluent clarified and digested is also called pretreated wastewater.Infiltration is driven under freeze-vulnerable depth and away from roots into a horizon of in situ soil solicited as the second main part of the system.This basic kind of OWTS relies on the subsoil to achieve an acceptable degree of secondary treatment, mainly through filtering and aerobic biodegradation, by removing suspended solids and some nutrients from pretreated wastewater (Conn et al., 2006;USEPA, 2002).These systems known as on-site soil absorption systems (OSAS) belong to type number 5 in the OSS compendium by Tilley et al. (2008).In case of insufficient soil ability for the treatment, granular or fibrous materials may be substituted for the soil, which combined with different cases for evacuation, induces a wide variety of OWTS.In France, at least 47% of the systems in operation are of the OSAS type; another 40% are OWTS using the soil as the receiving medium for water treated by other filtering materials, different types of sand filters included.The proportion is reversed within recent installations for the last 10 years.Besides, a range of methods was developed over time for ANC; they share the remaining 10%-13% of actual systems, including electrical bioreactors at home and other devices agreed after bench tests (European Technical Approval, ETA) to prove the prominent level of purification.Each OWTS outlet imposes a small punctual source of treated (with variable efficiency) wastewater to the aquatic media, which may appear negligible.At the scale of a region fully deserved by ANC, the great number of OWTS, even efficient, may be the largest source of residual contaminants to the environment.According to Withers et al. (2012), underestimating the number of septic tanks located in a catchment can lead to overestimation of the relative contribution from diffuse sources such as agriculture.Since the purification processed in potable water production plants does not consider the presence in raw water of all contaminants of emerging concern, OWTS may also deteriorate the quality of regional drinking water.From a protection perspective of raw water resources, it is relevant to make an environmental issue zonation, in French Zones à enjeu environnemental (ZEE), with regard to ANC outputs.
More than 10 years after Law no.2010-788 of July 12, 2010, on the national commitment to the environment, which introduced the notion of "proven risk of pollution" caused by OSS, in French Assainissement non-collectif (ANC), very few environmental zonation plans have been drawn up in France.To direct financial aid for work to sectors where the "proven risk of pollution" is greatest, ZEE can be delineated in the Schéma directeur d'aménagement et de gestion des eaux (SDAGE) (orientation A-1 and provision A-1.2 of the SDAGE Artois-Picardie 2016-2021, October 16, 2015;Agence de l'eau, 2015) or in a Schéma d'aménagement et de gestion des eaux (SAGE), which are, respectively, the regional-scale, a watershed-scale, Water Development and Management Plan.In contrast to the Sanitary issue zonation, in French Zones à enjeu sanitaire (ZES), which are clearly defined in the decree of April 27, 2012, relating to the modalities for carrying out the mission of controlling ANC devices (MEDDTL, 2012), the definition of ZEE is not very precise: "zone identified by the SDAGE or SAGE demonstrating contamination of water bodies by ANC on watershed heads and watercourses."Once the ZEE has been defined, a provision must be devoted to it in the SAGE's sustainable development and management plan.ZEE will be considered by the local authority, specifically by Public service for on-site, noncollective wastewater management, in French Service public de l'assainissement noncollectif (SPANC), when monitoring on-site devices and managing rehabilitation work, notably through grouped operations financed by the Artois-Picardie Water Agency.Local authorities will be required to ensure that their urban planning documents (territorial coherence plan, intercommunal local urban planning plans, and communal maps) and wastewater management zoning (two classes: collective or not) are compatible with ZEE.
This article reports on the experience of the SAGE of Lys (Lily) River watershed whose revision began in 2017 and offered the opportunity to carry out such zonation, following in the footsteps of a few other SAGE under the Artois-Picardie Basin committee.The aim is to map environmentally sensitive areas as ZEE.The methodology was developed for the most vulnerable part of the Lys SAGE territory where most of the drinking water catchments are located, which is part of the Pas-de-Calais department of France.The environmental issue is defined by the quality of watercourses.Publicly available data on sanitation and river flows are presented.This is followed by a description of a sampling, measurement, and analysis campaign carried out on low-flow rivers.The determination of ZEE is based on the definition of: ANC impact indicators, pressure zones based on the hydraulic impact assessed at source, and nitrogen pollution vulnerability zones based on the flow impact assessed on the target (river sections).Two zoning scenarios are presented in map form.After considering the indicators used in previous studies, the discussion focuses on the methodology of this zoning study, as well as on the territory's vulnerability criteria.

| STUDY SITE: CONTEXT OF THE LYS WATERSHED
The main natural feature of this region is the absence of major river and significant relief.Rivers and small coastal streams are characterized by low gradients and modest flows.These factors make the rivers highly sensitive to pollution from population density and activities.

| Geomorphological context and hydrographic network
The hydrographic network of the Nord and Pas-de-Calais departments links natural waterways, canalized river sections, and artificial waterways of all sizes.From the southern and western slopes of the Artois Hills, the rivers (Canche, Authie, Somme) flow toward the west, the Boulonnais region and the English Channel; from the eastern and northern slopes, the rivers (Aa, Lys, Scarpe, Escaut, Sambre) flow toward the north, the Netherlands, Belgium, and the North Sea.Historically, a vast river infrastructure has been created (from west to east: Neufossé, Aire, Deûle, and Sensée canals), perpendicular to the natural south-north flow direction, to link Dunkirk to the river ports of Belgium and the Netherlands.The Scheldt (Escaut) International River Basin District's lowest point is 2 m below sea level, situated along Schouwen's south coast (Prunje area).
The source of the Lys (river of lily flowers) is in the Lisbourg municipality, part of the Artois Hills.In its southern half (on chalky soils in Artois), the natural Lys River flows northwards for 44 km from its source to Airesur-la-Lys, with a loose network and deep-lying tributaries.
The geological and pedological nature of the northern half of the watershed (clay subsoil) has given rise to a dense network of natural watercourses, as well as canals and ditches in the plains (Scarpe, Lys, Anterior Flanders): downstream of Aire-sur-la-Lys, the Lys is canalized to a variable gauge.After 85 km in French territory, the Lys marks the frontier with Belgium for 25 km, before flowing 88 km into that country and emptying into the Scheldt (in French, Escaut River) at Ghent.
Although urban areas occupy 15% of the surface area, the Lys watershed is largely agricultural.The main crop is soft wheat, which accounts for 42% of the utilized agricultural area, with the remainder devoted to flowering vegetables, grain and silage maize, and permanent grassland.

| Hydrogeological context
The superimposition of surface water and groundwater subcatchments demonstrates a close communication between the Artois Hills rivers and the chalk aquifer.The water table contributes 80% to the flow of the Authie and Canche Rivers, and 70% to that of the Lys and Aa Rivers.During periods of low water in the Lys, its flow is sustained by draining the chalk aquifer.During rainy seasons, the trend is reversed, and the river's high water locally feeds this aquifer.
Around half of the watershed's drinking water supply comes from three groundwater bodies (with a European identifier, MESO): some are abstracted from the deep Carboniferous limestone aquifer (MESO No. FRAG015) and most resources come from the chalk aquifer, hidden by silts but outcropping in riverbeds (Deûle valley chalk MESO No. FRAG003; Artois and Lys valley chalk FRAG004).Located in the southern part of the watershed area, its recharge surface is 763 km 2 , with a much larger impluvium outside the Lys watershed.Effective recharge of the chalk aquifer represents an average of 7% of annual rainfall, and a study indicated an average annual recharge of 47 million m 3 /year over the entire impluvium (SO-GREAH, 1998).Its quantitative status is therefore satisfactory.The quantitative status of the Carboniferous limestone aquifer, on the other hand, is under surveillance.Groundwater quality depends on land use and geology.Most of the chalk water table is unprotected from infiltration, making it highly vulnerable to pollution from agricultural inputs.

| National and international regulatory context of the Lys SAGE
The Lys watershed SAGE territory is located in the new Hauts-de-France region (territorial reform of April 22, 2015).It covers an area of 1834 km 2 (32% in the Nord department and 68% in the Pas-de-Calais department) and includes 222 communes (municipalities), of which 172 are in the Pas-de-Calais department.
The Joint Association for the Lys SAGE, in French Syndicat mixte pour le SAGE de la Lys (SYMSAGEL), is the operational arm of the Lys SAGE.It was created by prefectorial decree on December 22, 2000.By prefectorial decree of December 28, 2009, SYMSAGEL was recognized as a watershed public territorial establishment, in French Etablissement public territorial du bassin-Lys, thus becoming EPTB-Lys.This label gives the legitimacy to organize in its own name and on the scale of the Lys watershed: flood prevention, balanced management of water resources, and preservation of wetlands and management.Revision of the SAGE began in 2017 (SYMSAGEL, 2019) and was under review until 2019 by the (national) Environmental Authority.EPTB-Lys's scope of action is that of SAGE's territory as set out in the prefectorial decree of May 29, 1995.Given the cross-border nature of Lys River and its watershed, the EPTB-Lys carries out its missions within an international framework (Figure 1).
Multilateral coordination for implementation of the European Water Framework Directive (WFD) (Council of the European Communities [EEC], 2000) in the Scheldt International River Basin District (IRBD) is governed by the Scheldt Treaty, concluded between the governments of three countries: France, Netherlands, and Belgium (federal government), and the assemblies of the three regions making up Belgium: Wallonia, Flanders, and the Brussels-Capital Region.The Scheldt IRBD comprises 79 surface water bodies (with a European identifier, MESU) in five subwatersheds, including the Scheldt-Lys Basin.This international framework is important, as it is an incentive to set ambitious targets such as ZEE, which are still very rare elsewhere in France.AND METHODS The main environmental issue in the Lys watershed is the quality of the raw groundwater and surface water used to supply drinking water.Another important issue is biodiversity and associated green tourism, an issue represented by Ecological interest zonation, in French Zones d'intérêt écologique (ZIE), and classified sites (landscapes), all not linked to the definition of ZEE, and on which we will open a perspective in Section 4.

| Issues in the Lys watershed
The Lys catchment area is unique in that groundwater (54%) and surface water (46%) are used almost equally for drinking water production.The water intake of the Moulinle-Comte drinking water depuration plant in the Lys stream is the only surface water withdrawal for drinking water.

| MESO
The first objective of the SDAGE (2016SDAGE ( -2021) ) for the MESO inside the Lys watershed was to maintain good chemical status for the Landenian sands (MESO No. FRAG014), which cover roughly the northern half of the Lys watershed, and for the Carboniferous limestone aquifer (FRAG015) to the east.In addition, the objective of achieving good chemical status has been set for the two MESOs of the chalk aquifer (FRAG003 and FRAG004), which correspond to the southern half of the watershed and are currently classified as poor status (since 2011) according to the WFD criteria; over the 2006-2011 period, nitrate (NO 3 -) (with an upward trend) and selenium were the elements responsible for the downgrading of FRAG003; aminotriazole and desethylatrazine were the compounds responsible for the downgrading of FRAG004; moreover, glyphosate and its by-product amino-methylphosphonic acid (AMPA) were present in excess in these two MESOs.This is why, for the chalk aquifer, the objective of good status has been postponed to 2027.Actions concerning the quality of MESO in the Lys SAGE territory include: reducing diffuse pollution by phytosanitary products and NO 3 -; protecting the feeding areas of priority drinking water catchments, with the aim of preserving the quality of the water abstracted, to limit catchment closures and the need for more curative treatments; and provisions aiming to preserve the water quality in drinking water-sensitive zones and restore the quality of degraded catchments (Figure 2).

| MESU
The Order of July 27, 2015, amended the Ministerial Order of January 25, 2010, on methods and criteria for assessing the ecological status, chemical status, and ecological potential of surface waters taken in application of Articles R.212-10, R.212-11, and R.212-18 of the Environment Code.
The good chemical status of a surface water body is determined by the achievement of environmental quality standards: contaminant levels must remain below pollution threshold values.Of the eight MESU in the Lys watershed, only the canalized Lys (MESU No. FRAR31) has good chemical status.Polycyclic aromatic hydrocarbons are the main substances responsible for the chemical downgrading of rivers in the Lys watershed.Moreover, the "nutrient concentration" quality item, made up of the following parameters, total phosphorus (P tot ), NO 3 -, and ammonium (NH 4 + ), found in high concentration in all the MESU, classifies them on average to poor chemical status (Supporting Information: Table A1).
The ecological status of MESU results from an assessment of the structure and functioning of aquatic ecosystems.It is determined using biological, hydromorphological, and physicochemical quality criteria.The Lys River maintains good ecological status, while the upstream Lawe River is in poor condition.The lowland rivers have been intensively developed.Of the eight MESU in the Lys watershed, there are six heavily modified water bodies and one artificial MESU (Table 1).For these seven water bodies, the objective of good ecological potential replaces that of good ecological status (with a system of four classes instead of five).In the Lys watershed, measuring stations are used to assess the following indices that characterize the ecological potential of a watercourse: the standardized global biological index (IBGN, with five stations); the standardized diatom biological index (IBD, with 24 stations); the standardized river fish index (IPR, with six stations).
After studying the aquatic invertebrate population, the IBGN assigns a score from 0 to 20.The value of this index depends on both the quality of the physical environment (bed bottom structure, condition of banks) and the quality of the water.The five measuring stations have an average (two stations), good (two stations), or very good (one station) IBGN.
For IBD, a positive trend was observed at some stations, depending on the period.The distribution of stations according to status class over the 2012-2013 period was: good (33%), average (50%), poor (13%), and bad (4%).The IPR, which compares with a reference population (i.e., little disturbed), shows an improvement over the 2007-2013 period: canalized Lys at Erquinghem (mediocre to good); Guarbecque (poor to mediocre); upstream Lawe (good to very good); Lys River (good to very good).
Despite some progress during the last decade, the Lys watershed still has disturbed surface aquatic environments.Once we had identified the communes with areas of ANC, the most accurate method of determining the number of ANC dwellings would be to refer to the inventory drawn up by the SPANCs, which were required to have carried out a diagnosis of all ANC devices by December 31, 2012, in accordance with article L. 2224-8 of CGCT.To obtain this information for the Lys watershed, all SPANCs were contacted.Several essential items of information were requested: the communes' wastewater management zoning map, the address of each ANC facility, and its compliance status.As not all SPANCs have the same level of knowledge of their devices, nor the same database management tools, we had to do with the lowest common denominator.The number of ANC devices per commune (Figure 3) was therefore obtained from the Artois-Picardie Water Agency database.The methodology for determining ZEE will therefore not be applied by geolocating ANC devices or selecting them according to compliance criteria.

| Wastewater treatment plants in the Lys watershed
In 2016, there were 60 wastewater treatment plants (commonly WWTP, will be referred to as STEP in the following) in the Lys SAGE area, including 27 STEPs in the Nord department and 33 STEPs in the Pas-de-Calais department of France.The total treatment capacity of these STEPs is nearly 582,820 population equivalents (p.e.) and ensures the collective part of wastewater management.In the Artois-Picardie Basin, the decree of January 12, 2006, classified the entire basin as a zone sensitive to eutrophication by urban wastewater.In addition, the delimitation adopted on November 23, 2007 classifies the entire basin (except the downstream Somme) as a "vulnerable zone" to NO 3 -emitted by agricultural activities.Following a subsequent study in 2013, only the Boulonnais region was excluded from this zone.
European and national regulations impose deadlines for bringing STEP up to standard, depending on the size of the facilities and the discharge environment.In June 2016, France reported to the European Commission, in accordance with Article 17 of May 21, 1991, Wastewater Treatment Directive (EEC, 1991), a list of agglomerations of 2000 p.e. or more whose wastewater facilities are noncompliant or at saturation point or required to meet the 2017 deadline for STEP rehabilitation, as they are in eutrophication-sensitive areas.
Within the Lys SAGE perimeter, 20% of STEPs are noncompliant in terms of equipment, plant performance, and/or wastewater network.Nonconformity of performance may be linked to discharges in dry or wet weather.EPTB-Lys does not have an exhaustive census of households connected to a collective sewer network.A diagnosis of connections and an evaluation of the connection rate would enable us to assess the pressure of nonconnected sewage on the environment and to evaluate leakage volumes per network or per STEP.
In 2016, 42 STEPs treated nitrogen (15 more than in 2003) and 14 treated nitrogen and phosphorus (four more than in 2003).Nine new STEPs have been commissioned between 2010 and 2019.The installation of new STEPs is enabling a gradual reduction in the pollutant load in waterways.Mass flows in discharges from STEPs increased until 2012, when they levelled off or even decreased.At the SAGE level, the average treatment efficiency of the wastewater treatment plants is good (Table 2), with elimination ratios of over 94% for the main parameters (5-day biochemical oxygen demand (BOD 5 ), chemical oxygen demand (COD), and total suspended solids).

| Public data on monthly 5-year low flow
Flow measurements taken by the flood forecasting and hydrometry network within the regional directorates for the environment, development, housing, in French Direction régionale de l'environnement de l'aménagement et du logement (DREAL), feed the national Hydro bank of hydrometric data.This database is used for river flow monitoring, flood forecasting, statistical flow calculations, regulatory flow control, and many other purposes.The Hydro bank stores water level measurements and enables instantaneous, daily, and monthly flows to be calculated.
Data from the 83 DREAL stations in the Lys watershed were used to verify that low water had been reached during our flow measurements and to validate them (Figure 4).
The QMNA, monthly flow (QM) not exceeded (N) in a calendar year (A), is the average of the 30 daily flows in the low-water month of a given calendar year.It is commonly referred to as the monthly low-water flow.For several years of observation, statistical processing of a series of monthly low-water flows enables us to calculate a return monthly low-water flow.In particular, the 5-year monthly low-water flow (noted QMNA5, given in m 3 /s) is a monthly low-water flow value that occurs on average once every 5 years: 4 years out of 5, monthly low-water flows are higher than the QMNA5; or, every year, there is a four-in-five chance that the low-water level will be less severe than this flow value.The QMNA5 is the monthly low-water flow of reference for the application of the water policy: a typical low-water flow in a dry year, it is used in the processing of discharge and water abstraction authorization files according to the vulnerability of the environment concerned.The QMNA5 is also used to set river quality objectives.

| Location of sampling and measurement points
The Lys watershed comprises 12 main subwatersheds.The study focused on five of these, for which the river length and watershed surface area are given here: the Lys River (36.0 km; 312 km 2 ); the Laquette River (23.4 km; 111 km 2 ); the Clarence River (32.8 km; 105 km 2 ); the Lawe River (41.1 km; 176 km 2 ); the Loisne River (28.0 km; 45 km 2 ).The Aire Node is associated with the Lys and Laquette watersheds, with the sampling points for the outlets located at Aire-sur-la-Lys.
In addition, two tributaries were included in the study: the Nave (21.9 km; 101 km 2 , a tributary of the Clarence) and the Surgeon (14.4 km; 37 km 2 , a tributary of the Lys).The Guarbecque subwatershed is not included in the study, as its hydrographic network is highly disturbed by drainage.The study excludes five other subwatersheds, including the Lys plain.The ZEE was therefore determined for a set of seven rivers.
The sampling and gauging campaign focused on springs and confluences.Two criteria were met: 1. Accessibility: Making field equipment easier to carry and safer from vandalism.2. The characteristics of the watercourse: Sufficient flow and section allowing sampling and gauging.This work was carried out in the field, the last week of May 2017, with the support of an EPTB-Lys technician.

| Types of samples and parameters analyzed
Samples and measurements were taken during the lowwater period, after about 10 days of dry weather.The assumption made in the field was that the parameters analyzed for nitrogen-Kjeldahl nitrogen (NTK), NH 4 + , nitrite (NO 2 -), NO 3 -, total nitrogen (NGL)-and P tot were essentially due to wastewater discharges (ANC and STEP) and not to the leaching of agricultural inputs into the soil.For lack of information, we did not consider the "proven risk" of subsoil contamination by leaking sewage networks.Three types of sampling were carried out (Figure 5): 1.A sample taken at a confluence is a pollutant flow accumulated over one day, obtained by a 24-h assessment in a single container using a battery-powered, time-servocontrolled portable sampler.2. Sampling at a spring provides an initial analysis of the pollution contributed by the water table; it is a one-off event, since it is assumed that the chemical variability of the water table is low over a day.3. Sampling at the STEP outlet gives an idea of the impact of the STEP discharge on the watercourse; it is a oneoff, considering that the chemical variability of the discharge is low over a day.
Samples were sent to the departmental laboratory in Arras (Pas-de-Calais).Herbicide analyses (glyphosate, AMPA, atrazine, desethylatrazine) for source and confluence points were carried out by the Carso laboratory (Lyon city).The parameters analyzed at STEP outlets were restricted-as herbicides are not eliminated by STEP treatment, they were not measured-to the contribution of NGL and phosphorus to the watercourse.
Temperature and oxygen gas concentration were measured with a digital multiparameter hand-held field instrument.
This provides an initial source of data for approaching pollution by ANC.Other parameters could have been studied (Escherichia coli, total coliforms, COD, BOD 5 , etc.), but a technical and economic compromise was made when designing this sampling campaign.

| In situ flow measurements
In situ flow measurements were carried out during lowwater periods.These measurements were compared with data from DREAL monitoring stations for the QMNA5.
The flow rate was determined using an electromagnetic current meter, a measuring device used to determine the velocity at a point in the flow section.It is used to measure water velocity point by point across the section, using a graduated pole.Here, 284 measurements were taken on 39 cross-sections of a watercourse, that is, around seven measuring points per section.

| Determination of subwatersheds
The identification of 36 subwatersheds associated with the sampling points make it possible to assign the data acquired at the nodes (flow, concentration, and flux for each physicochemical parameter) to sections, contributing surfaces, and finally zones, and to compare them to argue for classification as a ZEE or not.EPTB-Lys has an airborne Lidar (light detection and ranging) system with a topographical accuracy of 0.5 m.Lidar provides georeferenced point clouds that can be used to produce a digital terrain model (DTM).At our study scale, with a sufficient resolution of 20 m, each pixel represents an area of 400 m 2 .DTM processing to determine subwatersheds (Figure 6) was carried out according to a standardized protocol using Arcgis software (version 10.2).

| Methodology for determining ZEE
To define an environmental impact, using a classic risk assessment approach, we need to cross-reference a stress or pressure-a hazard characterized by its intensity-with the vulnerability of the chosen environment, characterized by indicators and quality classes.This cross-referencing is done using ratios or scale intervals.

| OSS (ANC) environmental impact indicators
Current regulations define a "proven risk" as the result of a malfunction or nonconformity of an ANC device.As we do not wish to use this type of data from SPANC since confidential, we will refer in our methodology to pressure zones associated with ANC in general, that is, the surface or underground discharge of wastewater treated by the full range of devices.Indeed, it is known that devices recognized as compliant and/or operating correctly also contribute to the diffuse source of pollution, particularly NO 3 -.These zones will be crossreferenced with vulnerability zones.
A prerequisite is to specify how the impact of ANC is expressed on the scale of a subwatershed on the corresponding section of a watercourse.Its seasonal expression at low water is a priori more significant, as in rainy weather, the impact of ANC could be masked by dilution and pollution due to agricultural soil leaching.The source hydraulic impact is defined as the total daily flow from ANC discharges (source of contamination), diluted (divided) by the monthly low-water flow of the receiving watercourse.This first ratio is a unitless rate.The source concentration impact is then defined as the total mass flow (g/day) of a chemical element from the ANC discharges (the source), diluted (divided) by the monthly low-water flow of the receiving watercourse.This second ratio has the unit of a concentration.For both ratios, it is assumed that the totality of the discharges (the source) could directly influence the watercourse (the target).They are often referred to as "dilution rates" (without or with chemical parameters); while it is undoubtedly an inappropriate expression since the indicator means … the inverse of dilution, this usage will be taken up in the discussion (Section 4.2) of this paper.The second ratio was calculated and analyzed but was not ultimately used to select the ZEEs for the Lys watershed.
The target flow impact is then defined as the mass flow differential (g/day) of a chemical element between the downstream and upstream points of a river section.This third indicator, with unity of a mass flow, does not know the source.Its effect must be examined within a flow scale to assign a quality class to the watercourse.On a section used as an outlet by a STEP, the flow contributed by collective discharges is to be subtracted from the target flow impact to isolate the flow impact of diffuse pollution (ANC and agriculture together).Based on the impact calculated in this way, an effective number of ANC discharges can be induced, provided that the contribution of agriculture is low.AMPA and glyphosate values are used to validate or qualify the origin of the NO 3 -input to the section.

| Pressure zones: Source hydraulic impact on a subwatershed
The number of ANC devices per subwatershed is a basic datum for assessing the source hydraulic impact and the source concentration impact.To avoid being intrusive in the use of data of a private nature, the geolocation of ANC devices was not used.First, for each subwatershed, we considered the total number of ANC devices in the communes intersected by the subwatershed.The number of ANC devices is overestimated in this first calculation, which is therefore on the safe side.Then, the number of ANC devices was refined: the number of ANC devices, known per commune, was distributed to the subwatersheds covering part of the commune's territory in proportion to the area of intersection determined by geoprocessing with the ArcGis software.
The source hydraulic impact index of ANC discharges is calculated for each subwatershed.The numerator of this ratio is 315 L/day/household, the reference value for calculating discharge rates from ANC (average discharge rate for 1 p.e.: 105 L/day/p.e., 1 ANC is equivalent to 3 p.e., which is the average number of people per household according to Insee 1 ).This reference value is multiplied by the number of ANC devices in the subwatershed.In the denominator: the stream flow is the low-water flow with a 5-year return period (QMNA5) at the subwatershed outlet; when the QMNA5 is not available, the flow measured during the field campaign is used as a default.
This indicator denoted as I was proposed by the Artois-Picardie Water Agency, based on the STEP impact assessment.It is largely overestimated since a significant proportion of hydraulic solicitation disappears through evapotranspiration in the soil.This is why the Water F I G U R E 6 Outlets and associated subwatersheds-calculation based on digital terrain model.
Agency has set the limit quite high, advising that subwatersheds with an I of over 10% should be placed in the ZEE.Other studies have chosen different values (2%, 3%, or 5%).The appropriateness of these values depends on the study area and the degree of selectivity required.In the Lys watershed, calculations were carried out for these four possible threshold values.
We proceed from downstream to upstream: the subwatershed defined by an outlet point is tested for each hydraulic impact threshold.If the I value of the subwatershed in question is greater than the threshold value, then we are in a pressure zone.However, if subwatersheds included upstream of the subwatershed under consideration have an I value below the threshold, they are excluded from the final pressure envelope.
3.4.3| Areas where watercourses are vulnerable to nitrogen pollution: Target flow impact NGL is the sum of NTK, which quantifies the reduced fraction of nitrogen (sum of organic nitrogen such as proteins, and mineralized, ammoniacal nitrogen) and oxidized nitrogen (NO 2 -and NO 3 - ).This parameter provides an overall indicator of nitrogen pollution.The contributing surface area of a watercourse section is the surface area obtained by subtracting the watershed area at the upstream point from the watershed area at the downstream point.Pollution vulnerability zones were determined in two stages.
As a first stage, to express the pollutant contribution of the contributing surface, the difference between upstream and downstream flows (in kg/day of NGL) is related to (divided by) the difference between upstream and downstream flows (in m 3 /day).This ratio is the concentration of NGL in the water flowing over the surface contributing to this section.This content is then classified in the Water Quality Assessment System (SEQ-Eau) (Table B1); this operation is straightforward, as this assessment system is built on content intervals.In so doing, we are conscious of diverting the SEQ from its normal use, which is to classify the point content at a watercourse junction and not the content in the water running off the contributing surface at a section.These two concentrations are not additive.
In the second stage, the NO 3 flux was calculated at each node and classified in the SEQ-Eau transformed into flux; the SEQ-Flux is specific to each node since it is obtained by multiplying the intervals of the SEQ-Eau content by the flow at the node.We preferred to convert to flow because by calculating the flow differential over the section, we can induce an effective number of ANC on the surface contributing to the section.
These two stages contribute to the decision by defining the criteria for classification as ZEE.
At the end of the first stage, calculation of the NGL content of the water flowing over the contributing surface, the sections receiving water of average, mediocre, or poor quality are classified as zones of proven pollution and therefore enter the potential ZEEs.Only the pressure zones containing these sections will ultimately be ZEEs.
At the end of the second stage, the NO 3 flux calculated at each node made it possible to classify all river nodes in the SEQ-Eau, and moreover to measure the flux differential between upstream and downstream of each contributing surface.This highlights the change in quality of a section between its upstream and downstream nodes.A section that improves from upstream to downstream, or that deteriorates while remaining in the good or very good classes, is not considered for classification as ZEE.A change from good to average or from average to poor will trigger a classification as ZEE.If the quality class remains the same in the section, either average or poor, we examine whether the quality is deteriorating.It is not always justified to classify this section as ZEE for ANC.Indeed, ANC is not the only or even the main impact on the stretch.The presence of agricultural pollution or a STEP can be invoked and must be studied.We must ask ourselves whether it is realistic to assume that the rehabilitation of the ANC devices would significantly contribute to preventing the deterioration of the section, or even enable it to move into the lower SEQ-Flux class (better quality).This is why we test whether the downstream flow is closer to the lower SEQ-Flux class or to the upper SEQ-Flux class: in the first case, the section is classified in the ANC impact zone.A section in the poor class, even if it is deteriorating from upstream to downstream, could not be restored by priority action on ANC and is therefore not classified as ZEE.

| Results on indicators and ZEE
The calculation of hydraulic impact, using the 10% threshold recommended in the Water Agency's initial method, results in only one watershed being classified as a pressure zone (and three very small subwatersheds).It is both poor and too selective.To make it more flexible, simply lower the threshold and cross it with a vulnerability criterion.Several values of the hydraulic impact threshold were tested and the pressure zone map for the 2% value was chosen (Figure 7).
The results for NGL flows from contributing surfaces to watercourse sections (Figure 8) are classified as average, mediocre, and poor.Only the Nave River is classified as being in very good condition according to SEQ-Eau, except for the section from the source Rimbert Burbure to Le Rimbert Busnettes.The results of the differential NO 3 fluxes (Table 3) were placed on a map to show the sequence of the sections and locate any notable points of pollution.Three sections were thus identified: 1. Loisne upstream-Loisne downstream: There is a large number of induced ANC spots and heavy agricultural pollution with AMPA, which has gone from good to mediocre status.2. Nave Brassarderie-Nave confluence Clarence: Change from very good condition to good condition due to mixing with the Clarence, which is in poor condition along its entire length.3. Laquette confluence Quernes-Laquette Aire-sur-la-Lys: Upgraded from mediocre to average due to dilution in the Surgeon Enquin/Laquette confluence Quernes section.
This makes the differential flow method a more selective tool for prioritizing action.Application of the criterion based on the differential NO 3 -flow between upstream and downstream points of a river section-the second stage in Section 3.4.3,carried out after the decision by the local water commission (CLE)-led to identify two areas more vulnerable than others: one in the Lawe valley and the other in the Laquette valley.These areas are covered by SPANC of the Béthune-Bruay Artois-Lys Romane conurbation.This selection is represented in overlaid hatching on the watersheds classified as ZEE at the CLE meeting (in red in Figure 9), which were defined by a hydraulic impact threshold of 2% and an average to poor NGL class of the contributing surface (first stage of Section 3.4.3).By the prefectorial decree of September 20, 2019, the revised SAGE has been approved and its report and map atlas were published.En 2020, this zonation ZEE has been published on https://aeap.maps.arcgis.com/apps/mapviewer/index.html?webmap=28e2c12daa 9b420f8b43062e212aa0ad (Lys subwatersheds in green).

| Discussion of the methodology
Feedback from other watersheds in Artois and Boulonnais regions since 2012 has highlighted the "dilution by chemical parameter" method or simple "dilution rate" initially proposed by the Seine-Normandie Water Agency to characterize the impact of STEPs (Figuet & Frangi, 2005;Morize, 1984).Indeed, ZEEs in some watersheds were determined using the "dilution rate" method: the Authie watershed SAGE used a threshold of 2%, while the Canche watershed SAGE used a threshold of 0.2%.For the Sambre watershed SAGE, the ZEEs were determined by cross-referencing zones with a "dilution rate" of over 3% with the number of zones of ecological interest.In the Liane watershed, the SAGE (said SAGE of Boulonnais) crossreferenced zones of ecological interest (zones grouping at least four ecological interests were retained) with zones with a "dilution rate" greater than 10% and then applied the "dilution by chemical parameter" method.The Somme watershed SAGE used the same criteria but went into greater F I G U R E 7 Pressure zones (in red) are defined by a hydraulic impact threshold of 2% as a necessary condition for environmental issue zones; no-pressure zones are in grey.
detail in selecting areas containing homes equipped with ANC, with the support of the watershed's SPANC.In the ANC zones at the head of the watershed, a 100 m buffer was drawn by coupling watercourses and wet zones, and then the land registry parcels containing buildings were identified from the National Institute for Geographical and Forest Information database.
Two other cases were found in other French regions: the method for the Vaucluse department was the subject of the prefectorial decree of July 25, 2014, defining the ZES and ZEE; on the Tille watershed (a tributary of Saône River), the areas retained as ZEE are at the intersection of safeguard zones for drinking water resources, catchment supply areas, and ANC zones (from the dilution rate method).
In line with this feedback, the "dilution by chemical parameter" method was therefore tested by Arondel (2018) for the Lys SAGE.In this way, a potential content linked to ANC discharges was obtained on the Lys subwatersheds.The method was evaluated for phosphorus and forms of nitrogen, as the determination of NH 4 + and NO 3 -in a discharge makes it possible, where appropriate, to identify whether we are dealing with a well-functioning ANC installation (discharge rich in NO 3 -and poor in NH 4

+
) or a poor one (discharge poor in NO 3 -but rich in NGL).For each chemical parameter, the threshold between good ecological status and average ecological status was taken as the discharge limit.According to WFD criteria, the limit value is 0.5 mg/L for NH 4 + , 0.2 mg/L for P tot , and 50 mg/L for NO 3 -(Table A1).If the potential content due to discharges exceeds this limit, it is assumed that this leads to a downgrading to average ecological status for the given parameter, and the subwatershed is classified in a potential pollution zone.
In the Lys watershed, for NO 3 -, the 50 mg/L limit value is exceeded in four cases: 1.The two Surgeon subwatersheds: Probably due to the STEP already mentioned in the flow method.2. Loisne subwatershed 2: As with flows, the cause is ANC, STEP, and agriculture (indicator AMPA).3. Nave subwatershed 3: The number of ANCs in this subwatershed is high (1368) compared to the subwatershed's 5-year low-flow monthly flow (424 m 3 /day).
The NH 4 + criterion per exceedance of the limit value results in classifying 75% of the Lys watershed area as ZEE, and P tot even more so, which does not make them effective prioritization criteria.What is more, the hydraulic and chemical impacts on watercourses of leaking wastewater F I G U R E 8 Contributing surface water by section-Overall nitrogen content and classification in the SEQ-Flux water quality assessment system; colors: poor (red), mediocre (orange), average (yellow), and very good (blue).
T A B L E 3 Flux method applied to NO 3 .| 277 networks are not considered, even though they are of the same order of magnitude as those of ANC.In fact, in the absence of any permanent diagnosis of the Lys watershed area, there is even less data on the state of the networks than on the state of ANC devices.Even the connection rate is unknown, not to mention the compliance rate for connections.So, we stopped there with the "dilution by chemical parameter" method.

Watercourses
In the NO 3 -flux method for river sections, a change from good to average or from average to poor should trigger classification as ZEE.The fact that these criteria are not encountered in the 19 subwatersheds of this study (except in one case, which changes from average to poor, but where the section is very short) does not make these two cases unusable criteria in general.The case exists and could exist more in another SAGE or if we had a greater number of nodes in the Lys watershed.
Some areas have zero (or negative) differential NO 3 flux.In this case, there is no reason to incriminate the ANC -the contribution of pollutants is nonexistent, either  because of the low number of ANC devices, or because of the purifying power of the contributing zone, which naturally attenuates pollutants from ANC, or because the watercourse in this section is itself totally resilient to these pollutants, or because there is a good-performance STEP, which discharges a significant quantity of water cleaner than the water in the upstream node (this is possible when the section is bad).The same reasoning applies if we observe an improvement in the quality of a section.
Of course, the results on flow differentials are only punctual in time.In fact, there was only one sampling campaign for this study.To confirm the methodology, other sampling campaigns could be launched, multiplying the number of measurement points to obtain shorter sections and thus a more precise mapping of pollution linked to ANC.It would then be necessary to geolocate ANC devices in the Lys study area by subwatershed, so that the hydraulic impact approach could provide a sufficient approximation.
The Lys SAGE territory is affected by environmental zoning covering an area of around 23,600 ha, or 13% of the territory.Beyond the vulnerability of aquatic environments, the territory's vulnerability would be characterized more comprehensively by also considering ZIE, which presents a notable ecological challenge (biodiversity).This issue was integrated into the sustainable development and management plan via the provisions and regulations of the SAGE, which were presented to the CLE at its meeting on October 18, 2017, to adopt the Lys SAGE project.A trial was carried out by Arondel (2018), which mapped the intersection of watercourses, vectors of pollution, with ZIEs.To the map of ZEE and ZIE could also be added a criterion of the landscape or tourist value of the sites.

| Perspective on the environmental impact of OSS
To further explore the notion of environmental impact, data is beginning to emerge in France on the source of ANC's impact, thanks to unilateral approaches by some SPANCs to carry out enhanced diagnostics that include analysis of discharge quality, and a study of the comparative quality of in situ discharges (Olivier et al., 2019), which showed that the source of pollution does not lie solely in faulty or old devices.Their findings, Gallagher and Gill (2021) as well, highlighted the variation in environmental performance of different ANC configurations and indicated opportunities for design improvements to reduce their life cycle impacts.Several countries, faced with the same problem, have undertaken research projects (Lusk et al., 2017).In the same viewpoint, Nasri and Fouché (2019) monitored the intermittent flux from a sand filter for household wastewater to the vadose zone, and from this in situ experiment, integrated solute transfer to the water table, especially NO 3 -and organic micropollutants such as parabens.Alternatively, they studied the impact of leachate from waterless toilets (Nasri et al., 2019).In the absence of a significant filtration potential of the underlying soil and/ or absence of a vadose zone with high thickness, by multiplying the amount of pollutant flux by the number of the OWTS in a region, OSS can be a considerable source of pollution for the environment.Evacuating treated water directly to any hydraulic way or aquatic medium is common with recently approved systems.In France, more than 210 ETA were published in 10 years, and the new standard acknowledges this large variety of systems.While the soil is still considered by the law as the better way for the release of treated water, these recently agreed systems allow for (and owners are authorized to) evacuate treated water directly to any drainage network, watercourse, or aquatic medium.For easiness, in 80% of cases in setting recently agreed systems, this solution is chosen over soil irrigation or infiltration, and this solution is more hazardous from a sustainable development perspective (Carroll et al., 2006).In addition, more and more sophisticated treatment systems or OWTS are installed with no warranty on their maintenance and performance.Their effect on water quality is also uncertain because current nutrient abatement policies ignore the temporal variation in nutrient loading that can influence ecological response in streams and connecting ditches.Between the source and the target, if the water coming from the ANC devices is the vector of contamination, the conditions of natural attenuation along the flow trajectories are little studied and the delay in which this contamination can generate pollution in the environment is unknown.It highlights the need for further studies of additional classes of contaminants of emerging concern in OWTS effluents, particularly in regions impacted by failing OWTS and high groundwater exchange with surface water.

| CONCLUSION
It is at the watershed scale so at the SAGE level that the priority zones for bringing ANC devices into compliance must be located.The water local commission (CLE) of the Lys SAGE has expressed the wish that these environmentally sensitive areas should be identified and referred to as ZEE, so that the actions to be implemented as part of the SAGE revision can be guided and evaluated.
Pressure zones (hydraulic impact) and vulnerability zones (nitrogen pollution flow differential) have been determined.Cross-referencing these zones has enabled us to map the ZEE of the Lys watershed area.These zones will be considered by SPANCs to prioritize the work required to rehabilitate defective devices in the frame of the ongoing SDAGE Artois-Picardie, the regional Water Development and Management Scheme, which has been renewed for the period 2022-2027.
As the regulations do not propose a set of criteria for identifying ZEE, it was useful to draw on feedback from other territories in France.In consultation with the CLE members, the identification criteria most relevant to the territory were selected.The method implemented in the Lys watershed is intended to be adaptable to other watersheds, depending on their characteristics, based on the same key issues-raw water supply and recharge zones for drinking water and contributing zones of the hydrographic network-and on detailed information on on-site domestic wastewater treatment devices.APPENDIX: WFD and SEQ-EAU STATUS CLASS LIMITS Tables A1 and B1.

F
I G U R E 1 Scheldt International River Basin District and water management structures linked to the Scheldt-Lys Basin.The Lys watershed is shaded in light blue.

F
I G U R E 2 Drinking water-sensitive zones in the Lys (Lily) watershed.Blue area: Zone of drinking water issue.Brown to yellow area: Catchment protection perimeter (when large enough).Triangles: Water intake point-blue = active; green = project; grey = considered for abandonment.

3. 2 |
Public data on sanitation and river flows 3.2.1 | Sanitation zoning and OSS ANC (OSS) concerns dwellings that cannot be connected to a public wastewater collection network.It is generally recommended in scattered housing areas.The compliance rate for ANC devices, estimated in 2010, was 21.1% for the Nord and Pas-de-Calais, and 19.4% for the Pas-de-Calais department alone.Wastewater management zoning, provided for under the French Water Act of January 1992 and mandatory since 2005, is required under article L. 2224-10 of the French General Code for Local Authorities (CGCT).The Lys watershed comprises 222 communes, 95% of which have finalized their wastewater management zoning: five are in full collective management (AC); 25 communes are in full OSS (ANC); 181 communes use both modes, AC or ANC, by zone; and 11 have not defined it.

F
I G U R E 3 Number of on-site sanitation devices per municipality in the Lys (Lily) watershed.Noncollective domestic wastewater management (ANC) metadata from SPANC.Triangle (STEP): Wastewater treatment plant.
Abbreviations: BOD5, 5-day biochemical oxygen demand; COD, chemical oxygen demand; TSS, total suspended solids.a Gross flow is the pollutant load at the STEP inlet.b Net flow is the pollutant load at the STEP outlet after treatment.c Minimum efficiency is 70% for total nitrogen, based on BOD 5 loads >600 kg/day.d Minimum efficiency for total phosphorus is based on BOD 5 loads >600 kg/day.

F
I G U R E 5 Sampling points for the field campaign: yellow = 24 h cumulative sample; green = instantaneous sample; red = wastewater treatment plant (STEP).

F
I G U R E 9 Zones with environmental issues environmental issue zonation, in French Zones à enjeu environnemental (in red) validated by the local water commission (CLE) and refinement (hatched) proposed post-CLE, based on the differential flow criterion.
T A B L E 1Abbreviations: MESU, surface water body (with a European identifier); WFD, Water Framework Directive.
: NO 3 , nitrate; SMAEL, Joint Association for water supply of Lys watershed, in French Syndicat mixte d'adduction des eaux de la Lys. Abbreviations T A B L E A1 WFD ecological status classes for the chemical parameters studied.A B L E B1 SEQ-Eau ecological status classes for the chemical parameters studied.
T -, nitrate; SEQ-Eau, System for the Evaluation of Water Quality.