Coastal Water Quality: Hydrometeorological Impact of River Overflow and High-resolution Mapping from Sentinel-2 Satellite

Abstract

The increase of human settlements and activities in coastal areas is causing a significant impact on coastal water quality. Predicting and monitoring the latter is of fundamental importance for assessing sustainable coastal engineering and ecosystem health. This trend is strongly influenced by the presence of rivers’ mouths, acting as critical links between inland and sea. Forecasting river discharges and overflows, using hydrometeorological modelling, can provide a quantitative estimate of the excessive supply of sea nutrients, favouring algal proliferation and eutrophication phenomena. The river overflow contributes to the increase of the coastal bacterial concentration, contaminating marine bioindicators, such as bivalve molluscs. Coastal water status can be monitored by satellite high-resolution optical spectroradiometers, such as Sentinel-2 constellation, capable to retrieve Chlorophyll-a concentration as well as total suspended sediments, at the resolution of about 10 meters. This remote mapping is complementary to in situ samplings, both essential for supporting decisions on the management of coastal mollusc farming and fishing. In this work, we report the recent advancements in hydrological model-based prediction of river surges and remote sensing techniques exploiting Sentinel-2 imagery as well as their implications on coastal water quality management. As a pilot area, we select the central Adriatic Sea in the Mediterranean basin and the Abruzzo region coastline in Italy.

1. Introduction

In recent decades, the increase of human settlements and activities in coastal areas is causing a significant impact on the resilience of the world’s coastal and marine natural capital. The coastal environment is a dynamic ecosystem where natural and anthropogenic processes add up and interact, modifying their geomorphological, physical, and biological characteristics. Coastal areas are also defined as ecotones, which are very important from an ecological point of view as they are a natural transition zone between two different and adjacent ecological systems.

The human pressures are different and include climate change, overfishing, offshore commerce, and land-based activities. The several pressures on the coastal ecosystem and the possible overlapping pressures can cause cumulative adverse effects [1, 2, 3]. Land-based stressors link coastal marine systems to terrestrial human activities and represent dominant stressors in coastal ecosystems [1, 4, 5, 6]. Nutrient and chemical pollution run-off create coastal eutrophication, harmful algae blooms, or hypoxic or anoxic dead zones [4, 6, 7, 8], and these impacts are able, not only to harm coastal species and ecosystems [9, 10, 11] but also affect human health [12, 13] and economic activities.

A few research is available, where the impacts of human wastewater on coastal ecosystems and community health [14, 15, 16, 17] are assessed. The combined effects from multiple pressures are not still considered in management or planning processes and this reduces the overall resilience of marine ecosystems.

According to the Water Framework Directive [18], 93% of the European marine area is under different pressures from human activities and about 28% of its coastline is affected by pressures causing changes in hydrographic conditions, for example, in seawater movement, temperature, and salinity. According to the hydromorphological pressure assessments made in coastal waters, the main sources are atmospheric deposition and discharges from urban wastewater treatment plants on the coast, or further in the catchment area [14, 19, 20, 21].

Coastal developments modify natural hydrological conditions and impact habitats where the pressure at the catchment scale is the highest on the coastline of the Mediterranean Sea. Intense human activities in regions surrounding enclosed and semi-enclosed seas, such as the Mediterranean, always produce, a strong environmental impact causing increasing coastal and marine degradation, in the long term. The sustainable development in the Mediterranean area is influenced by diverse factors, such as i) the rapid growth of the urbanisation rate; ii) the increase in tourism; iii) the rapid development that determines the degradation of coastal areas; iv) water scarcity; and v) commercial activities. This condition highlights the need to define mitigation strategies, using timely and action-oriented information.

Due to its morphology, the Italian Peninsula can be divided into two main basins that can be considered semi-enclosed. The first includes the western Mediterranean, limited eastward by the Sicilian channel, and characterised by wide abyssal plains. The second, the eastern Mediterranean, dominated by the Mediterranean ridge system, is characterised by more complex morphology. In Italy, populated areas are mainly concentrated along with coastal areas than the rest of the territory; according to the Corine Land cover data [22], the Italian coast has a length of about 8300 km: more than the 9% of the littoral is now artificially bordered by works grazing the shore (3.7%), ports (3%) and partially superimposed structures on the coast (2.4%). The artificialisation of housing and transport structures in coastal areas is gradually increasing. It has been estimated that a relative increase of the 5% in the area 10 km away from the shore was generally recorded in European countries between 2000 and 2006 [23].

The assessment of the sea and coastal systems and their interaction, based on scientific knowledge, are the indispensable basis for the management of human activities, in view of promoting the sustainable use of the seas and coasts and conserving marine ecosystems and their sustainable development.

In 1975, 16 Mediterranean states and the European Community under the auspices of the United Nations Environment Programme (UNEP) defined the Action Plan for the Mediterranean (MAP) [24, 25], aimed at protecting the environment and promoting sustainable development in the Mediterranean basin. The 19th Meeting of the Contracting Parties in 2016 agreed on the Integrated Monitoring and Evaluation Programme of the Mediterranean Sea and Coast and related evaluation criteria (IMAP) [26] which establishes the principles of integrated monitoring: for the first time, biodiversity and non-native species, pollution and marine, coastal, and hydrographic litter will be considered in an integrated way. The IMAP implementation defines 27 common indicators, foreseen in the Integrated Monitoring and Evaluation Programme, in line with the UNEP/MAP Barcelona Convention. The prediction and monitoring of water quality are among the main activities to be carried out for the protection of coastal ecosystems.

As for the prediction, water quality is strongly influenced by atmospheric events that could affect the pollution management systems, such as rainfall-dependent sewage drains and tributary river flow. For this reason, river mouths act as critical links between the hinterland and the sea. The prediction of river discharges and overflows using hydrometeorological models can be fundamental for indirect estimation of water quality, given that the drainage network runoff is closely related to the supply of marine nutrients, favouring algal proliferation and eutrophication phenomena. It also contributes to the increase in the concentration of faecal bacteria, such as Escherichia coli recognised as a faecal indicator organism in the European legislation, which contaminates marine bioindicators, such as bivalve molluscs. This condition has a relevant socio-economic impact; for example, it limits the consumption of bivalve molluscs collected from contaminated waters. The reductions in water quality after high precipitation events and the subsequent increase in river discharges lead local authorities to close shellfish harvesting areas after large events. But the inability of local authorities to accurately predict these events or to immediately assess the water quality exacerbates the losses of fishing economies.

From the above premises, it is clear that knowing the ecological status of water bodies is of critical importance to monitor how human activities are impacting or, the other way around, impacting by the coastal ecosystem. Monitoring coastal waters, indeed, is fundamental for both the evaluation of ecosystem health and as a support to local fishing economies, in terms of sustainability and site selection. Member States of the European Union are required, by the EU Marine Strategy Framework Directive [6] and the Water Framework Directive [18], to preserve territorial waters within the first nautical mile and achieve good ecological status. According to the European Environment Agency water assessment [20], only 46% of water bodies are actively monitored, 23% of monitoring did not include in situ water sampling and 4% still had unknown ecological status [27]. The proportion of water bodies without observation data is much larger than the ones for which monitoring is granted and for those monitored, in most cases, the status of surface waters was not classified as “good” [28]. Satellite observations offer a solution to the current limits shown by conventional water sampling methods—they allow to achieve much wider spatial and temporal coverage and larger water bodies. Remote mapping, therefore, is complementary to in situ sampling, both essential for supporting decisions on coastal aquaculture operations. It can also provide support in quantifying elements of environmental status that are currently not reported, such as phytoplankton blooms. In this context, the European Union together with the European Space Agency has boosted the development of the most advanced satellite-based instruments to observe optical water quality. Through the Copernicus framework, the spatial sector has had significant investment in recent years, and this enhances the cost-benefit of using satellite-based technologies for monitoring surface waters, also being satellite data freely available.

As for the environmental surveillance, coastal water status can be monitored by satellite high-resolution optical spectroradiometers, capable to retrieve suspended sediments or algae presence, at spatial resolutions of up to 10 meters [23, 29].

2. Hydrometeorological impact of river mouth on coastal water quality

Most human activities are related to water and take place in coastal areas or along riverbanks. The correct management of these areas is therefore primarily referred to as the basin-scale [18] where the territorial planning activities need to consider the aggregate effects of hydrographic changes caused by human activities at sea and on land [18, 30, 31, 32].

The Marine Strategy Framework Directive [6] highlights the importance of the assessment of the hydrographical conditions through seawater physical-chemical parameters, that is, temperature, salinity, depth, currents, waves, turbulence, turbidity (from a load of suspended particulate matter), upwelling, wave exposure, mixing characteristics, residence times, the spatial and temporal distribution of nutrients, oxygen, and acidification [30, 33, 34, 35]. All these variables are essential for understanding the dynamics of marine ecosystems that can be altered by anthropic presence.

Hydrographical conditions are site-specific and depend on landscape features, morphology, and lithology and are often conditioned by large-scale forcings, such as tide, general ocean circulation and climate. Small-scale features, such as land use and human-induced pressures, are also relevant to the river dynamics, especially in coastal areas.

The offshore waters of the Mediterranean Sea are extremely oligotrophic, and the coastal areas have been historically known to be influenced by natural and anthropogenic inputs of nutrients, mainly concentrated in the Adriatic basin [36].

Monitoring of contamination in different mollusc species is a well-known methodology, applied to assess the level of sea-water contamination, which exploits the bivalve capability to accumulate and retain contaminants. Moreover, mollusc edible species contamination is constantly monitored to ensure the introduction of safe products in the food market.

The European Regulation No 627/2019 [37] assesses the official control programmes for bivalve molluscs and provides the classification of mollusc production areas based on microbiological monitoring for the bacterium E. coli in the mollusc flesh and intervalvular liquid, used as a faecal indicator organism (FIO). Since rivers are routes for the transfer of organic matter including faecal bacteria from inland to the sea, the contamination is strongly affected by the inland drainage network, which collects also the most important abiotic factors affecting bacterial contamination of molluscs [38].

Even if the influence of FIO on the quality of coastal waters is studied since the XVIII century, few studies exist that attempt to evaluate the relationship between fluvial transport and shellfish hygiene in the sea [39, 40].

The land-sea-river system is extremely complex; therefore, it is not straightforward to establish a relationship between runoff, precipitation, and contamination levels. Connections are site-specific and dependent on the physiographical characteristics of each catchment and, in addition to the existing, local human pressures. Nevertheless, contamination decay is also connected to local environmental parameters which affect the bacterial dilution and the self-purification process of the bivalves.

The analysis of mollusc contamination has a socio-economic value; therefore, it is doubly important to evaluate in terms of both monitoring contamination levels and attempting to forecast possible pollution events. Nevertheless, different competencies may be required to achieve this purpose; on one hand, prediction of environmental processes requires deep knowledge of earth system modelling and data interpretation. On the other hand, environmental implications on food security monitoring are requested to adequately assess the design of useful tools or instruments able to really ameliorate the bivalves’ productions. A wide number of capabilities should be connected to work together to find ICT solutions: biologists, physicists, engineers, and economists, for example. A virtuous example of this collaboration was the CapRadNet project (http://cetemps.aquila.infn.it/capradnet/), which originated a fruitful collaboration between different institutional levels and different actors usually involved in diverse activities. The outcome of the presented work is born from the project collaboration. The main aim of the present research was to investigate the relationships between two main environmental variables (flow discharge and precipitation) and the contamination level of molluscs harvested areas in a target site. The feasibility study resulted from the capitalisation of other two previous projects, funded by the IPA-Adriatic CBC Programme. The proposed analysis has taken advantage of E. coli concentration data analysed in the framework of the CAPS2 project (www.caps2.eu) and the hydrological modelling system developed in the same area to predict possible flood events due to severe meteorological events, as an outcome of the AdriaRadNet project (http://cetemps.aquila.infn.it/adriaradnet/).

The good practices defined in the CapRadNet project are being tested in a new project financed at the regional level, which intends to create the Early Warning System as a final product to improve the economic and production efficiency of the plants through the environmental information made available to aquaculture producers, which will be described below.

2.1 Early Warning System for sanitary risk: Hydrometeorological operational forecasted tool

The outcome of the feasibility study carried out during the CapRadNet project [41] had shown that an Early Warning System (EWS) for the sanitary risk assessment could be set up operationally, given the existing relationship between discharge overflow and E. coli concentration increases in bivalves’ harvested area, in the Pescara River mouth. In general, rainfall is the most referred environmental factor influencing microbial contamination in coasts and estuaries affected by stormwater runoff and long sea and shoreline outfalls [42, 43, 44]. Rainfall-induced contamination could persist in molluscs as much as 6 days after the rainfall event [45, 46], even if a shorter lag time (<3 days) has been reported in the literature [43, 47]. The differences are often associated with the hydrogeology of the catchment (i.e., concentration time) and residence times in the receiving water [46]. Being recognised as an important predictor of microbial contamination of bivalve harvesting areas [48], river flows are explicitly mentioned in the EU legislation on official controls of bivalve molluscs intended for human consumption. In the study proposed by Campos et al. [46], the highest levels of E. coli were detected when total rainfall exceeded 2 mm and water levels in the main tributaries exceeded the mean flow.

Few published studies have considered the contribution of river flows in informing official public health controls for bivalve mollusc fisheries [48, 49], but the scientific literature has shown that catchment-scale microbial dynamics determining bathing water compliance are often determined by hydrological events [50, 51, 52].

2.2 Pilot Study

A numerical experiment was carried out, using 6 months of E. coli concentration data in the mollusc flesh and intravalvular liquid, detected in three pilot areas around the Pescara River mouth (Figure 1). Since official discharge data from hydrological annals are available until 2010, for the same 6-month period, a hydrological simulation was performed by using the CHyM distributed hydrological model [53, 54, 55], forced with observed rainfall data. The model has been extensively used in Abruzzo Region for flood forecast activities [56, 57]. The hydrometeorological conditions preceding each E. coli concentration exceedance were investigated, in terms of accumulated rainfall in the coastal area and runoff in the Pescara River mouth. A quick overview of obtained results is here given and discussed; more detailed information is available in Colaiuda et al. [41].

Figure 1.

Three pilot areas around the Pescara River mouth.

The analysed catchment originates in the inner, northern part of the Abruzzo region, draining an area of about 3147 km² before flowing into the Adriatic Sea. It is characterised by a very complex orography with altitudes spanning from zero up to almost 3000 m a.s.l. in the range of 150 km. The last ten kilometres along the river path are strongly urbanised, with a relevant solid transport amount, estimated at 106 tons/year, considering only the Pescara city urban area.

Outcomes of hydrometeorological investigations linked to the E. coli concentration peaks suggested that i) E. coli concentrations appeared to be most linked to the discharge peaks with respect to the precipitation values and ii) E. coli peaks exceeding the reported threshold occurred after 2 or 3 days after the Pescara River discharge peak, in most cases.

In more detail, 29 samplings were analysed, and exceeding E. coli concentrations were linked to a runoff overflow in 83% of cases and a rainfall event in 50% of cases. As for the mussel farm sampling location, in the open sea at ~5km south-east the Pescara mouth, the 100% of exceeding E. coli concentrations were linked to a river overflow, while only the 33% were preceded by rainfall in the coastal area. The case study that occurred on March 8, 2016, revealed a high peak of E. coli without any river runoff increase a few days before the event. This case study was then deepened and the hydrometeorological analysis revealed that a huge discharge peak, reaching about 400 m³/s, affected the Pescara River 7 days before, suggesting a longer river effect on bacterial transport for this case. Moreover, in some cases, a precipitation event over coastal areas and a river discharge increase occurred at the same time and the contribution of the two forcings cannot be discriminated at a first glance. The rainfall effect may also include the presence of sewer overflows (CSOs), direct land-runoff into the estuary, and re-suspension of contaminated sediments within the estuary itself. Finally, increased levels of E. coli in bivalves from all monitoring points under high river flow conditions suggest that stormwater runoff is contributing to a significant proportion of E. coli accumulation in bivalves [46]. Nevertheless, due to the catchment extension and geographical location, the coastal rainfall does not represent an environmental descriptor indicative of possible faecal contamination related to weather events. The discharge overflow estimation is indeed more representative of the hydrometeorological precursor (Figure 2).

Figure 2.

Time series showing E. coli concentrations at P1, P2, and Mussel Farm, the discharge at the mouth of the Pescara River from November 2015 to May 2016.

A significant association between E. coli concentrations and the magnitude of the antecedent discharge peak has been carried out [41]. The Spearman’s correlation coefficient r_D calculated was 0.69, and the associated p-value was low (∼4.5 × 10 5), confirming the correlation hypothesis. The correlation between rainfall maxima and E. coli concentrations resulted in a lower correlation coefficient (r_R ¼ 0.35) and the associated p-value was high (∼0.065), not confirming the correlation hypothesis.

Hydrological conditions prior to river flow peaks, such as heavy rainfall, are important in determining the presence of E. coli in seawater, but it cannot be ruled out that even low rainfall events could cause significant increases in concentrations when they follow a dry period.

Local regulations for monitoring water and molluscs have been planned regardless of weather conditions, the river flows, or other abiotic factors that can affect the concentration of FIO, and sampling intervals to detect the potential microbial contamination may, therefore, not be representative of variations in these conditions. For this reason, it is reasonable to assume that the data underestimate the strength of the correlations between bacterial concentration, precipitation, and river flow.

To overcome these limits, a holistic approach based on the correlations between data of precipitation (intensity and position) and variation in the river flow discharge is essential. This strategy is useful for predicting times and places of exposure to microbiological contamination. The combined assessment of abiotic factors (physical and chemical), hydrometeorological components and biotic factors also provides holistic information on the health of the ecosystem.

3. High-resolution coastal mapping of Chl-a concentration from Sentinel-2 satellite

As previously introduced, the use of satellite data is reaching constantly wider applicability in the earth monitoring sector. The availability of free satellite data throughout the globe, characterised by a great variety of accessible sensed data types, has sped up this process. In Europe, the raise of programmes, such as Copernicus (European Union’s Earth Observation Programme) founded by the European Commission, allowed the birth of new value-adding activities and studies on earth monitoring and services for disaster prevention [58, 59]. The whole programme is composed of seven missions, with a fleet of around ten satellites among future and operational ones. The most important thematic streams of Copernicus services are dedicated to land and marine environments. An example of marine applications is Sentinel-3 satellites that have a push-broom imaging spectrometer called OLCI (Ocean Land Color Instrument). This instrument measures solar radiation reflected by the Earth in 21 spectral bands and has a ground spatial resolution of 300 m [60].

For what concerns the marine environment, as above extensively analysed, its health status is strictly connected to many human activities, especially for coastal waters which represent a vital asset. The possibility to monitor it constantly and rapidly, covering large portions of territory, plays a crucial role and perfectly fits the enhancement introduced using satellite data for monitoring.

The symptoms to be monitored for marine environment’s health status evaluation are several and an example is the detection of algae presence. Under certain conditions, indeed, algae can reproduce in an accelerated way, giving birth to what is called an “algal bloom”. Some kinds of algal bloom can be toxic, may cause skin rashes or illness in humans as well as can be poisoning for some marine species (e.g., shellfish). A possible parameter to measure algae presence in a water body is chlorophyll-a (Chl-a), the pigment that is used by algae for photosynthesis which constitutes the part that mostly interacts with solar radiation. The concentration of Chl-a contributes to the so-called particulate organic matter present in a water body. On the other hand, water clarity is also connected to all the possible suspended matter that contributes, when in high concentrations, to increase its turbidity [61, 62]. The concentration of all possible suspended particles in a water body is defined as Total Suspended Matter (TSM) concentration. Coastal waters, moreover, operate as a link between land and ocean systems. Rivers physically allow this connection, acting as a conduit for delivering significant amounts of dissolved and particulate materials from terrestrial environments to the coastal ocean, increasing TSM concentrations. In some cases, part of this TSM can be composed of soil particles detached from the coastline and dragged away from waterpower [63].

Several approaches have been followed for years to perform marine monitoring through satellite observations, depending on the specific parameter to be estimated (Sea Surface Height, Wind Speed, Sea Ice, just to mention a few) [64]. For what concerns the detection of water quality, the estimated parameters are connected to its bio-optical properties. At some specific wavelengths, indeed, the suspended particles inside water can interact with radiation incoming from the atmosphere, giving back in return an upwelling radiant flux that has some characteristic responses (it can be absorbed in certain wavelengths more than at others). Optical sensors, therefore, have been widely used to identify the so-called spectral marine inherent optical properties (IOPs, e.g., absorption and scattering) [65]. Usually, the water-leaving signal is quite low (sometimes 1% or less of downwelling irradiance) and requires the sensors to work in a set of narrow, sensitive spectral channels and to remove the atmospheric effects. Those sensors are usually referred to as “ocean color” sensors. The spectral signals received can be used to estimate phytoplankton abundance and other radiatively active constituents.

Usually, the main approaches followed to retrieve IOPs from satellite measurements are two [66]—the first applies atmospheric correction (AC) algorithms to remove the contribution of the atmosphere from the signal received at the top of the atmosphere (TOA) by the sensor and leads to the estimate of the bottom of atmosphere (BOA) reflectance (calculated as the ratio of water-leaving radiance to downwelling irradiance just above the air-sea interface). Different kinds of algorithms can be then applied to AC reflectance values to produce estimates of geophysical properties (e.g., inversion model [67, 68]). The second approach tries to find a direct relationship between the spectral radiance at the top of the atmosphere and IOPs [69, 70]. This one is more immediate and removes the AC step that can sometimes lead to misinterpretation of the atmospheric contribution in presence of optically complex water masses.

However, the resolution of the satellite images used remains the main constraint for accuracy and precision obtained through monitoring developed solutions. It is important to note that coastal areas are also spatially and optically complex and would require more frequent spatial and spectral sampling to enhance the monitoring capability [71].

3.1 Advantages of Sentinel-2 satellite data for coastal water remote sensing

As said, satellite data can boost the realisation of more effective environment monitoring algorithms. Among the most used satellite data for coastal water studies, we can find several studies that employ high-resolution optical data obtained from the OLI (Operational Land Imager) sensor on board of Landsat-8 satellite and MERIS (Medium Resolution Imaging Spectroradiometer) on the Envisat satellite [72, 73]. Additionally, to the OLI sensor, Landsat 8 satellite payload is also made of Thermal Infrared Sensor (TIRS). These two sensors are characterised by a spatial resolution of 30 meters (visible, NIR, SWIR); 100 meters (thermal); and 15 meters (panchromatic). MERIS, which is a push broom radiometer, reaches a spatial resolution of 300 m at nadir (for full resolution products) and 1200 m for reduced resolution data and its spectral range varies from 390 nm to 1040 nm.

In parallel with the aforementioned missions, and despite being built mainly as a land monitoring mission, also Sentinel-2 satellite has gained popularity for marine applications. Indeed, thanks to its high spatial resolutions together with a high revisit frequency, Sentinel-2 allowed to overcome several limitations of existing missions. Sentinel-2 is equipped with a MultiSpectral Instrument (MSI) with 13 spectral bands from the visible and near-infrared to the short-wave infrared (from 443 to 2190 nm). The spatial resolution varies from 10 m to 60 m, depending on the spectral band, with a 290 km field of view [74]. The MSI sensor is made of a three-mirror, 150 mm aperture telescope which collects light and focuses it into two separate focal planes—one for visible (VIS) and near-infrared (NIR), and the other for short-wave infrared (SWIR) wavelengths, respectively. Each focal plane is composed of 12 detectors staggered in two rows.

Its revisit frequency is increased with respect to other missions thanks to the simultaneous operations of two identical satellites: Sentinel-2A and Sentinel-2B, launched in 2015 and 2017, respectively. This more frequent data availability offers several benefits—from a higher probability of finding imagery clear of cloud and sun glint; to more effective applicability of change detection algorithms [75]. Moreover, the free availability of its data allowed to facilitate the spread of their usage.

3.2 Retrieval algorithms for Chl-a concentrations through Sentinel-2 data

In the studies conducted by Marzano et al. [76], Sentinel-2 data played a crucial role in the detection of water quality. Their study was focused on Case-II waters, as per Morel and Prieur water classification [77]. In coastal areas, indeed, water quality is mainly conditioned by Chl-a and TSM concentrations variations and their study was focused on different retrieval approaches to these quantities.

The region of interest in this study is central-northern Italy over Tyrrhenian and the Adriatic Sea. The in situ observations of Chl-a concentrations were provided by Italian regional environmental agencies named ARPA (Agenzia Regionale per la Protezione Ambientale), for the regions of Tuscany, Lazio, Abruzzo, and Veneto, and covered the time period between 2016-08-04 and 2018-04-19. In Figure 3, the spatial distribution of the dataset along the Italian coasts is shown.

Figure 3.

In situ observation points locations of ARPAs’ dataset.

They analysed the use of both empirical and model-based regressive algorithms to retrieve IOPs. More in detail:

• the first method is based on the use of atmospheric correction for the retrieval of BOA reflectance from TOA radiances; values obtained using maximum band ratio (MBR) model on satellite data were put in comparison with observations provided by ARPA’s for the evaluation of the empirical regressive algorithm, realized through models defined in literature.

• in the latter was developed a radiative transfer equation (that uses observations to determine absorption and scattering coefficients) through which synthetic reflectance values are retrieved. Those synthetic values are then used to evaluate the model-based regressive algorithm.

The atmospheric correction software used is ACOLITE, with the Dark Spectrum Fitting (DSF) enabled. Figure 4 reports an example of RGB composite images for Top Of Atmosphere and Bottom Of Atmosphere reflectance before and after atmospheric correction, respectively.

Figure 4.

RGB Sentinel-2 remote-sensing reflectance images over the Adriatic coast in the Marche region before (a) and after (b) the atmospheric correction using the ACOLITE software.

Focusing on the Empirical Regressive algorithm (EmpReg), developed in [76], the retrieval of Chl-a concentrations was defined through the following:

$$r_{MBR}=\frac{max\left(R_{wlB1},R_{wlB2}\right)}{R_{wlB3}}\text{,}$$E1

where the numerator is the maximum between B1 and B2 (blue bands) water-leaving reflectance and the denominator is the water-leaving reflectance for B3 (green band). Indeed, the bands more sensitive to chlorophyll presence were considered to retrieve Chl-a concentrations. This blue-to-green reflectance maximum band ratio (MBR) model is among the most used ones in literature [78, 79].

The empirical regressive retrieval algorithm, which is the optimal regressive formula found with respect to the area of analysis and dataset used in the paper, is defined as:

$${\hat{C}}_{\mathit{Chla}}=a_{1}exp\left(-a_2{r}_{MBR}\right)$$E2

where a₁=59.795 mg/m³ and a₂ = 4.559.

In the following Figure 5, is reported the scatterplot of chlorophyll-a (Chl-a) in situ measurements (mg m⁻³) with respect to Sentinel-2 MSI water-leaving blue-to-green maximum band ratio (MBR) in the Tyrrhenian and the Adriatic Sea. Figure 5 highlights the non-linearity that characterises the relationship between Chl-a concentrations and MBR.

Figure 5.

Measured MBR with respect to in situ Chl-a concentration for the whole training dataset.

However, statistical regression algorithms show limitations in handling non-linearity and non-monotonicity, and to overcome these limitations, Marzano et al. [76] used also neural networks. This allowed the exploitation of data contained in several MSI spectral channels of Sentinel-2 products (from B1 to B8A) and spatio-temporal information. In the same way as the previously described methods, also in this case, two neural network-based algorithms were tested:

• empirically trained algorithms, for which the inputs were constituted by atmospherically corrected satellite data, extracted in the point closest to observation, for bands B1 to B8A, together with latitude-longitude information;

• model-trained algorithms, for which the input was made of results obtained from the model-based regressive method (radiative transfer).

The experiments conducted in the work lead to observing better results for NN algorithms when trained with empirical data, rather than with synthetic ones. Although a test to be performed on a wider dataset would be needed. The results obtained through the empirical regressive algorithms with MBR, instead, did not always provide an accurate estimation of the Chl-a concentration, depending on the higher turbidity of Tyrrhenian coastal waters. This was probably related to turbidity conditions of water, which can impact the effectiveness of estimation.

4. Implications for farmed and wild bivalves

Coastal water quality is strictly correlated to the food safety of bivalves. In fact, mollusc bivalves are filter feeders, and they can accumulate microbiological and chemical contaminants from surrounding water. Bivalves are generally cultivated or collected along with the coastal areas where their nutrients are abundant and flow from inland water. But the rivers can also discharge faecal bacteria, mostly from untreated wastewater [80]. The malfunctioning of urban waste water treatment plants or their by-pass during the severe rain events can contribute to the release faecal bacteria into the river. This can pose a potential risk to the consumers of bivalves that can accumulate these bacteria.

To avoid any health human risks, according to the EU Regulation No 627/2019 [37] the competent authorities firstly classify the production areas (Class A, B, and C) through specific monitoring campaigns. Then, they continue to control the level of faecal contamination of the bivalves according to the specific surveillance monitoring plan. The bacterium E. coli is used as a faecal indicator organism. For the Class A assignment, for example, the samples (80%) shall not exceed 230 E. coli per 100 g of flesh and intravalvular liquid. From this Class, molluscs can be collected for direct human consumption. During the sanitary control, if the mollusc health standards are not met, the competent authorities shall close the production areas and/or reclassify them [37].

As a decision support system for the competent authority, several studies have investigated the correlation between the increase of bacterial concentration in molluscs and weather conditions [40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 81, 82, 83]. The prediction of local precipitation and river discharges have been used as early warning signals for mollusc bacterial contamination [84, 85]. The advantages are multiple—i) to avoid the collection of the potentially contaminated product; ii) to avoid any temporary closing of production areas; iii) to optimise the monitoring surveillance programme; iv) to ensure the health of consumers.

Generally, results demonstrated that the correlation is site-specific and it depends on numerous factors, such as the geographical location, land use, and catchment size.

From ancient times, the Adriatic basin is particularly devoted to the bivalve farming and fishing in the lagoons and along the coasts. Here, the influence of the weather condition and river run-off on the bivalve hygiene condition has been investigated [41, 86, 87].

In the central Adriatic coast of the Marche region, recently, Ciccarelli et al. [88] published the correlation between the concentration of E. coli in the natural banks of Chamelea gallina and the local precipitation from 2016 to 2020. The results showed that the rainfall events were significant for the increase of E. coli (> 230 MPN/100 g) in the molluscs collected from the south sampling points. In the same region, the increase of Salmonella spp. detected in bivalves was reconducted in 2015 and 2016 to the severe meteorological events [89].

In the Northern Adriatic Sea, the CADEAU project [90] developed specific indexes to evaluate the potential microbial pollution impact of urban waste water treatment plants on the farmed molluscs in the Municipality of Chioggia (Venice, Italy). It provides indexes of dilution for E. coli based on the bacterial decay due to salinity, temperature, and solar radiation [90].

In the following sections, we report some “site-specific” study cases carried out in the Abruzzo region, on the central Adriatic coast of Italy.

4.1 The study case of wild clams and farmed mussels in the Pescara province

In 2021, Colaiuda et al. [41] published the case study in the Pescara province (Abruzzo Region, Italy) that was already detailed in the paragraph 2.2. Here, two production areas of wild clams (C. gallina) and one farm of mussels (Mytilus galloprovincialis) facing the Pescara River were investigated (Figure 1). In Figure 1, Pescara 1 and 2 are the production areas of clams, the other is the farm of mussels. Thanks to the CapRadNet project, this study executed a correlation analysis between river discharge trough to the CHyM model, precipitation in the catchment area, and the concentrations of E. coli detected in the bivalves during the official monitoring programme. The referring period was from August 1, 2015 to July 31, 2016. The EU reference method to detect E. coli was ISO-16649-3 [91]. Results were expressed as the most probable number – MPN per 100 g of flesh and intravalvular liquid of mollusc. Microbiological data were downloaded from the database of the project CAPS2 developed also the informative tool “CAPS2 WEB GIS” useful for the management of the production areas. The classification of the three production areas (Class A) was viewable in the CAPS2 WEB GIS [92]. The competent authority was the unique authorised user to modify the classification and the boundaries of the production areas in the CAPS2 WEB GIS.

The results showed that the concentration of E. coli in molluscs increased within 6 days of a river discharge peak (Figure 2). Moreover, 87% of cases of high concentration of E. coli were consequent to the increased river flow, while 60% of cases to the precipitation. These results suggested that the Pescara River discharge was the potential hydrometeorological driver of E. coli in facing molluscs to be further evaluated with specific sampling before and after discharge peak at the river mouth.

4.2 The study case of mussel farm in the Teramo province

The research project FORESHELL was funded by the FLAG Costa Blu through the 2014-20 EMFF programme of the Abruzzo Region. It is aimed at developing sanitary/weather-environmental predictive technological tools to enhance the efficiency and sustainability of a mussel farm in the Teramo province (Giulianova city, Abruzzo region, Italy) [93]. This production area of M. galloprovincialis was classified as Class A, and it is facing the Salinello and Vibrata Rivers far away almost 3 miles from the coast (Figure 6).

Figure 6.

Sampling points at river mouths and at the farm in the Abruzzo region.

The hydrological model (CHyM) has analysed the hydrographic basins of the rivers and it has been forecasting the discharge peaks. Before and after these events, a sample of freshwater at the river mouths, and of molluscs and sea water at the farm have been collected for the E. coli detection [91]. Preliminary results showed that until September 2021, there were four meteorological events (Table 1) that did not cause a peak discharge at the river mouth. Results did not register a significant increase of E. coli in the mussels (Figure 7). At the same time, the environmental parameters such as sea water temperature, salinity, Chl-a, sea currents, and wave motion are acquired by the satellites and in situ probes.

Date of meteorological event	Description of the event
21/09/2020	Scattered rain in the internal area
10/10/2020	Severe event in the northern Adriatic Sea
17/07/2021	Rainfall in the coastal area
27/08/2021	Storm at the coastal area

Table 1.

FORESHELL project: Description of meteorological events.

Figure 7.

FORESHELL project: E. coli concentrations before and after meteorological events.

The web application for data visualisation is under construction, as well as the early warning signalling to the farmer by mail/SMS/WhatsApp. The alerts are referred to the potential faecal contamination of molluscs predicted through hydrological data and other parameters that can damage the farm, such as high temperature and wave motion.

Furthermore, the growth of mussels is constantly monitored with biometric controls.

In conclusion, the first period of project execution was characterised by precipitation scarcity that did not cause any discharge peaks at the river mouths without any presence of E. coli in the molluscs. Further analyses are expected to be executed during the rainy period of the autumn and winter seasons.

4.3 Satellite data for bivalves

Recently, satellite data and maps are increasingly used for the identification of areas intended for aquaculture, for the knowledge of the environmental conditions useful for shellfish farming and fishing, for the prediction of potentially harmful events, etc. The knowledge of parameters such as temperature, salinity, and turbidity give important information to managing bivalve production. For example, data on Chl-a could be useful to understand the food disposal for molluscs or the prediction of algal bloom potential toxic.

The projects AQUACULTURE2000 and VALUE-SHELL analyse satellite data, such as pH, temperature, and CO₂, to assess the possible contribution of mussel farming in sequestering carbon from seawater through the biocalcification processes in the northern Adriatic Sea by mitigating the effects of climate change [94].

In 2021, the total suspended matter, temperature, and Chl-a estimated from satellite acquisitions have been used to predict the presence of radioactivity in molluscs [95].

Along the Adriatic coast facing the Abruzzo region where several farms are placed, two pilot studies were conducted. The goal was to calibrate algorithm coefficients, at a local scale, to set up a processing chain that derives accurate concentration maps of chlorophyll and suspended solids from the satellite, as said, taking advantage of the high frequency of revisit time and high spatial resolution of the satellite acquisitions.

The first study estimated Chl-a and sediment dispersions in the sea, derived from Sentinel-2 images, compared with in situ data acquired by means of a multiparametric probe in the monitoring stations that Agenzia Regionale per la Tutela dell’Ambiente (ARTA) Abruzzo monthly checks [96]. The Case-2 Regional Coast Color processor in ESA SNAP software was used, applying the C2RCC-Nets algorithm [97], whose parameters have been set using in situ measurements, specifically the salinity and temperature variables. This preliminary study provided encouraging results with only four sampling dates—for example, the concentration map of TSM of 09/03/2018 is reported (Figure 8).

Figure 8.

Concentration maps of Total Suspended Matter along Abruzzo coast in the Adriatic Sea, as elaborated with the C2RCC processor from Sentinel-2 imagery of 09/03/2018.

In the second study [98], the authors developed the Water Color Data Analysis System (WC-DAS), a tool for the operational generation of maps and indicators useful in the monitoring of water quality. The tool, in its first release, allows the processing of satellite optical multispectral data acquired by Sentinel-3 OLCI, Sentinel-2 MSI, and Landsat-8 OLI sensors, using the algorithms Case 2 Regional CoastColour (C2RCC) [97] and Atmospheric Correction for OLI “lite” (ACOLITE) [99].

The tool was tested in the central Adriatic coastal zone, setting the local parameters according to the ad hoc in situ sampling campaign that ARTA Abruzzo carried out along the Abruzzo coast, simultaneously with the satellites overpasses (example in Figure 9).

Figure 9.

Turbidity, Total Suspended Matter, Chlorophyll-a maps along Abruzzo coast, facing Pescara river mouth, as derived from Water Colour Data Analysis System Tool.

The results show performances of calibrated algorithms and the data system’s suitability to contribute to the production of monitoring maps and indicators, informing domain-specific decision-making and supporting services for integrated coastal zone management.

5. Conclusions

The human pressure on coastal areas is causing a significant impact on the resilience of the marine ecosystem in which natural and anthropogenic processes interact, modifying their geomorphological, physical, and biological characteristics. The combined effects from multiple pressures are not still considered in management or planning processes and this reduces the overall resilience of marine ecosystems. The assessment of the sea and coastal systems and their interaction, based on scientific knowledge, are the indispensable basis for the management of human activities, in view of promoting the sustainable use of the seas and coasts and conserving marine ecosystems and their sustainable development. A few research is available, where the impacts of human wastewater on coastal ecosystems and community health are assessed.

To overcome these limitations, an Early Warning System for health risk assessment should be optimised through a holistic approach based on the combined analysis of abiotic factors (physical and chemical), hydrometeorological components, and biotic factors also provide health information ecosystem.

The prediction and monitoring of water quality are among the main activities to be carried out for the protection of coastal areas. As for the prediction, water quality is strongly influenced by atmospheric events that could affect the pollution management systems, such as rainfall-dependent sewage drains and tributary river flow. For this reason, river mouths act as critical links between the hinterland and the sea. The prediction of river discharges and overflows using hydrometeorological models can be fundamental for indirect estimation of water quality, given that the drainage network runoff is closely related to the supply of marine nutrients and faecal bacteria. These can be accumulated by the farmed and wild bivalves in the coastal areas posing a risk for the human consumer. Therefore, the development of Early Warning Systems integrating predictive satellite data, could improve both the sanitary surveillance by competent authorities and the daily farming/fishing operations by workers. The economic loss could be reduced by improving the protection of consumer health.

The use of satellite data for water quality monitoring is gaining high importance, see the limitations that remote sensing techniques allow to overcome with respect to in situ observations (wide-area coverage, more frequent data availability). Several approaches have been followed for years to perform marine monitoring through satellite observations, depending on the specific parameter to be estimated. For what concerns the detection of quality of water, the parameters estimated are connected to its bio-optical properties. The Sentinel-2 data played a crucial role in the detection of water quality in the coastal areas, indeed, water quality is mainly conditioned by Chl-a and TSM.

A wide number of capabilities should be connected to work together to find ICT solutions: biologists, physicists, engineers, and economists, for an example. Different competencies may be required to achieve this purpose; on one hand, prediction of environmental processes requires deep knowledge of earth system modelling and data interpretation. On the other hand, environmental implications on food security monitoring are requested to adequately assess the design of useful tools or instruments able to really ameliorate the bivalves’ productions.

The study cases detailed in this chapter have been conducted by a multidisciplinary team that has developed several tools, also predictive, useful for stakeholders, such as competent authorities, farmers, researchers, veterinary services, and fishermen, consumers. The overall aim was to acquire knowledge and develop innovative technological tools focused on the enhancement of regional services.

Conflict of interest

The authors declare no conflict of interest.