Effects of variations in po river discharge on physical water characteristics and chlorophyll-a levels in the gulf of manfredonia

Rivers remarkably contribute to the transfer of fresh water and nutrients between land and marine ecosystems, underscoring their essential ecological role. The Po River contributes approximately one-third of the total freshwater discharge into the Adriatic Sea. Despite its importance, there is a lack of information regarding the impact of Po River discharge on the physical and biological characteristics of specific regions within the Adriatic Sea, such as the Gulf of Manfredonia. By using ocean model results and satellite data for 2014, 2015, 2016


Introduction
The Adriatic Sea has an elongated shape and is nearly landlocked or semi-enclosed (Artegiani, 1997), located in the northernmost part of the Mediterranean Sea (Fig. 1).While the western Adriatic coast is situated along the Italian coastline, the eastern coast aligns with the Balkan Peninsula.This water basin is subdivided into three regions: northern, middle, and southern.In the Adriatic's basins, the depth gradually increases when moving from the north to the south.
Three main factors influence the Adriatic circulation, including river runoff, wind, and inflow from the Strait of Otranto, resulting in a cyclonic current within the Adriatic Sea (Orlic et al. 1992).In the eastern part of the Adriatic Sea, the eastward current responds to inflows of water from the strait, whereas in the western part of the Adriatic Sea, the westerly current is influenced by river discharges (Ursella et al. 2006;Poulain 2001).The eastward current transports salt and warm water from the Ionian Sea into the Adriatic, while the westerly current conveys cold and less saline water from the Adriatic to the Ionian Sea through the Otranto Strait.
The discharge of rivers, particularly the Po River, has important contributions in maintaining the salinity balance of the Adriatic Sea and the Mediterranean Sea.Approximately 35 rivers discharge into the Adriatic Sea with varying discharges, totalling an average runoff of 5500-5700 m 3 /s (Sekulic 1996).With an average flow of 1500 m 3 /s, the Po River alone contributes 28% of this discharge.The Po River has experienced fluctuations on both monthly and yearly time scales, with some months exhibiting discharge levels below normal and others surpassing the normal range.Notably, the Po River discharge varied significantly, ranging from 542 m 3 /s (July 2015) to 4452 m 3 /s (November 2014) throughout the period from 2014 to 2017 (Mihanovic et al., 2021).So, selecting these years can be interesting and important to understand the effects of Po discharge fluctuations on the physical properties of water.
The fluctuation in Po River discharge can significantly impact the physical properties of water in the Adriatic Sea, influencing factors such as temperature, salinity, and circulation.The primary influence of the Po River is notable in the western region, particularly along the Italian coast.The injection of fresh water into the sea follows a predominant southward current, leading to substantial changes in the physical characteristics of water along this path.One specific area affected can be the Gulf of Manfredonia, situated in the western part of the southern Adriatic Sea (Fig. 1).The Gulf of Manfredonia is functioning as a shallow transition zone between the circulations of the northern and southern Adriatic.
Studies highlight the significant influence of wind direction on the circulation patterns within the Gulf of Manfredonia.Cyclonic and anticyclonic eddies, generated by north-northwest and south-southeast winds, respectively, play a crucial role in shaping the circulation dynamics (Focardi et al. 2009).A key question arises: How do Po River fluctuations influence the temperature, salinity, and circulation patterns in the Gulf of Manfredonia?However, our current knowledge is insufficient to provide a conclusive answer to this complex relationship.
Moreover, the Po River transports nutrients to the Gulf of Manfredonia, potentially affecting the growth Fig. 1 The Adriatic Sea map, featuring its basins, with specific markings denoting the Po River and the Gulf of Manfredonia.The western and eastern currents constitute the primary components of the Adriatic's current depicted in the figure of phytoplankton.However, our understanding of how much the Po River influences the Gulf 's biology, especially phytoplankton growth, is incomplete.This gap is crucial given the Gulf 's importance and environmental threats from waste, which can harm water quality.Elevated nutrient levels, like nitrogen and phosphorus, can increase algal biomass.We think that the impact of the Po River is considerably more pronounced in the Gulf, especially given that the Ofanto River contributes only an average flow of 13.9 m 3 /s (Simeoni 1992) and is situated to the south of the Gulf of Manfredonia.
Our motivation to work in the Gulf of Manfredonia stems from its multifaceted importance.Beyond its significance in influencing circulation patterns, the Gulf holds several key factors contributing to its environmental, ecological, and socio-economic importance.Firstly, its geographical location as a shallow transition zone within the southern Adriatic Sea renders it highly susceptible to environmental influences.Secondly, the Gulf serves as a critical habitat for diverse marine life, supporting various species of phytoplankton, zooplankton, and fish, which contribute to the overall biodiversity of the Adriatic ecosystem.Consequently, changes in environmental conditions within the Gulf can have cascading effects on marine ecosystems and biodiversity.Additionally, the Gulf of Manfredonia plays a vital role in supporting local economies and communities.It serves as a fishing ground for artisanal and commercial fishing activities, providing livelihoods for coastal communities.Moreover, the Gulf 's scenic beauty and recreational opportunities attract tourists, contributing to the region's tourism industry and economic development.These factors motivate us to work in the Gulf of Manfredonia.
This paper tries to address three pivotal scientific goals.Firstly, we study the hydrodynamics in the Gulf and how it changes from 2014 to 2017.Secondly, we aim to understand how the fluctuation in Po River discharge contributes to variations in temperature, salinity, sea level, and circulation within the Gulf of Manfredonia.This involves a comprehensive examination of the interconnected dynamics influenced by the Po River's discharge.To address the first and second scientific questions, we employ an ocean model to simulate the physical characteristics of the Gulf of Manfredonia.Thirdly, we explore the existence of correlations between Chlorophyll-a level and Po River discharge fluctuations.We know that Chlorophyll-a is a green pigment crucial for photosynthesis in plants and algae.Additionally, we investigate potential correlations between the physical properties of the Gulf of Manfredonia and the increase of Chlorophyll-a level.To answer the third scientific question, while we extract the physical property data of the Gulf of Manfredonia from the ocean model, we retrieve Chlorophyll-a levels from satellite data.
The rest of the paper is structured as follows.The second section focuses on model validation and configuration.The third section aims to investigate the hydrodynamics of the Gulf, delving into the intricate process of eddy formation within the Gulf.The fourth section delves into the impacts of the Po River on physical properties.Subsequently, the fifth section explores the effects of Po River fluctuations on Chlorophyll-a level.The final section engages in a comprehensive discussion and presents the conclusions drawn from the main results.

Model configuration and verification
The Regional Ocean Modelling System (ROMS) is used in this paper.ROMS is a numerical ocean model widely used for simulating and studying oceanic processes at regional scales.It is a community-supported model that has been developed collaboratively by scientists and researchers.ROMS is designed to simulate the physical, chemical, and biological processes in the ocean and is particularly well-suited for modeling coastal and regional seas.ROMS stands as a nonlinear ocean model that employs free-surface equations, with its code structured in F90/F95.It forms the basis for the SCRUM (S-coordinate Rutgers University model) and has found application in diverse water basins, as indicated by studies conducted by Haidvogel et al. (2000), Marchesiello et al. (2003), and Lorenzo et al. (2004) and Babagolimatikolaei and Layeghi (2022).
This paper encompasses the entirety of the Adriatic Sea and the northern Ionian Sea.We used General Bathymetric Chart of the Oceans (GEBCO) data, with a resolution of 15 s, for generating the grid file (GEBCO Compilation Group 2022).ROMS grid spacing is approximately 2 km along latitude and longitude, with 25 vertical layers.We use hourly ERA5 data with a grid spacing of 0.25° × 0.25° for atmospheric forcing and the initial conditions are from the World Ocean Atlas 2013 (Locarnini et al. 2013;Zweng et al. 2013).A total of 30 rivers are incorporated into the model.For 29 of these rivers, the climatology data for river runoff is used, while for the Po River, monthly-average data collected at the Pontelagoscuro sampling station (Mihanovic et al., 2021) are considered.The model incorporates eight tidal forcing components (M 2 , S 2 , N 2 , K 2 , K 1 , O 1 , P 1 , Q 1 ) from the TPXO model (Egbert and Erofeeva 2002).
In configuring the ocean model, a comprehensive approach is taken to accurately simulate ocean dynamics.This involves employing various techniques and parameters to represent atmospheric forcing, frictional effects, and boundary conditions.The bulk formula is used to incorporate atmospheric forcing into the model with GLS turbulence mixing, while quadratic bottom friction is implemented to account for frictional processes at the sea bottom.Specifically at the southern boundary, radiation and nudging techniques are applied to ensure realistic conditions are maintained.Critical choices regarding the vertical coordinate system, such as the S-coordinate transformation equation (Vtransform) and stretching function (Vstretching), are made with values of 2 and 4 respectively, alongside parameters like the S-coordinate surface control parameter (theta_s) set to 5 and the S-coordinate bottom control parameter (theta_b) set to 2, all of which contribute to accurately capturing vertical ocean processes.Additionally, setting the critical depth in the vertical layer to 50 m ensures proper representation of essential oceanic phenomena.
For this study, the results of 2014, 2015, 2016, and 2017 are used after a two-year warm-up and spin-up period.The model is verified with observational data collected from 12 stations in the Adriatic Sea, chosen based on their accessibility, and which can be accessed through the website https:// map.emodn et-physi cs.eu/.All observations are conducted within the middle and southern basins of the Adriatic Sea, covering the entirety of the year 2017.The data is chosen to encompass all months of 2017.In Fig. 2, "data1" corresponds to January 2017, "data2" to February, and "index 12" denotes December 2017.The data corresponds to the average errors of each point.Validation results indicate that the model's error is generally less than 10% (Fig. 2).
In general, model-derived salinity estimates tend to be lower than observed values, especially in 2017, exhibiting errors ranging from -0.2 to 1.4 P.S.U and an average error of approximately 0.5 P.S.U (Fig. 2a).The maximum difference is observed near the surface, where the model consistently underestimates salinity.The most significant errors occur within the surface to 50 m range, corresponding to the mixed layer.Beyond a depth of 300 m, the difference between model and observation is less than 0.2 P.S.U.Subsequently, at all stations, the error remains below 0.6 P.S.U.In summary, the model tends to consistently underestimate salinity.Statistically, the bias is -0.28, with a root mean square error of 0.31 and a median absolute error of 0.24.
Temperature errors within the model exhibit a range from −1.8 to 3.5 °C, with an average error of around 1 °C (Fig. 2b).Notably, the model consistently predicts temperatures lower than those observed.The most pronounced temperature errors are observed in the upper 0-50 m layer.Beyond this depth, the model's error gradually diminishes, reaching a decrease of up to 1 °C at depths exceeding 200 m.This pattern suggests a tendency for the model to underestimate temperatures, particularly in the near-surface layers, with a gradual improvement in accuracy as the depth increases.Statistically, the model exhibits a slightly lower bias of -1.0029 in temperature prediction compared to observations, with a mean absolute error of 1.0753.The root mean square error (RMSE) for temperature is measured at 1.3, while the R-squared coefficient in ERA5 stands at 0.7369.

Hydrodynamics of the gulf
In this section, we analyze the model results to achieve our initial scientific goal: investigating the hydrodynamics of the Gulf from 2014 to 2017.We discuss circulation, eddy formation, and parameters such as temperature, salinity, sea level, and vertical velocity.We aim to demonstrate whether there are fluctuations in the hydrodynamics of the Gulf or not.

Circulation in the gulf
While our primary focus is on studying the circulation of the Gulf of Manfredonia, our study area includes some portions both upstream and downstream of the gulf.This added area in our study area helps us to understand the behaviour of the Adriatic western current before the gulf and after the gulf.
In January, the behaviour of currents in the Gulf varies from year to year (Fig. 3).In January 2014, the velocity within the gulf is notably low (< 0.05 m/s), while the westerly current of the Adriatic Sea near the gulf reaches approximately 0.2 m/s.Conversely, in January 2015, an anticyclonic current dominates within the gulf.This is primarily due to the strong western Adriatic current, reaching speeds of about 0.3-0.4m/s.Further, in both January 2016 and 2017, an anticyclonic current dominates in the Gulf, maintaining a velocity of 0.05 m/s.
In April, the current speed remains relatively consistent across all years at approximately 0.05 m/s, but there are differences in its direction.In April 2014 and 2015, a cyclonic current form within the Gulf, detaching from the western Adriatic current and creating a cyclonic circulation along the western coast.During April 2016, a westward current is observed in the Gulf, encompassing a significant portion of it, with a speed of 0.03 m/s.However, in 2017, an anticyclonic current emerges after separating from the western Adriatic Sea current.
August exhibits the strongest currents in the Gulf, with speeds ranging from 0.1 to 0.5 m/s in different years.During this month, an anticyclonic current with a speed of 0.1 m/s typically cover the entire Gulf.Notably, 2015 stands out with the strongest current, boasting an anticyclonic current reaching 0.5 m/s.In 2016, the current speed decreases to 0.3 m/s, while in 2017, the anticyclonic current achieves 0.2 m/s.Among all years, the strongest Adriatic's western current is in 2014 with speed of 0.4 m/s, while 2015 has the weakest current.
In December, notable alterations in the circulation patterns of both the Gulf and the Adriatic's western current are observed.Although the western current in the Adriatic exhibits increased strength, there is a weakening in the circular circulation within the Gulf.Specifically, the circular current in the Gulf displays a uniform decline across all years, stabilizing at 0.1 m/s.Remarkably, December 2014 stands out as a pivotal month, recording the most robust westerly current among all months, reaching a peak speed of 0.5 m/s.

Circular current: eddy or gyre?
Based on the results of the "Circulation in the Gulf", the main feature of the circulation of the Gulf of Manfredonia Positive values indicate that the observed values are higher than the model calculations.Index refers to the month of 2017, where "data1" corresponds to January 2017, "data2" to February, and "index 12" denotes December 2017 is cyclonic or anticyclonic circulation.Here we can name this circulation as a gyre or eddy.In oceanography, gyres are large-scale, persistent, and more organized rotational systems of ocean currents, while eddies are smallerscale, temporary whirlpools or vortices that can exist within or near gyres.Choosing the term 'eddy' is deemed more appropriate for describing the circulation in the Gulf of Manfredonia.To understand eddies, it's essential to examine two key aspects: their spatial scale, referring to the size of the eddy in space, and their temporal scale, which relates to the duration, longevity, or time characteristics of the eddy.In the following section, we delve into a detailed analysis of both these parameters for a more comprehensive understanding.

Spatial scale
The most effective parameter for illustrating the spatial scale of an eddy is its diameter.Understanding the diameter of eddies is crucial for oceanographers as it provides valuable insights into the processes governing ocean dynamics, energy transfer, and the impact of these features on marine ecosystems and climate patterns.The Fig. 3 Surface circulation in the study area for selected months, using monthly averaged data eddy diameter is defined as the equivalent diameter (d) of a circle with the same area (A) as the closed contour outlining the eddy boundaries based on Eq. 1 (Souza et al. 2011).
We focus on the characteristics of the strongest eddy, with a particular emphasis on the month of August based on the results of "Circulation in the Gulf".Among the various years under consideration, the eddy observed in August 2015 emerges as the most formidable, while its counterpart in August 2017 displays the weakest.Using Eq. 1 reveals that eddy diameters range from 50 to 100 (km) in different years.Considering the scale of the eddy within the gulf, it conforms to the classification of mesoscale eddies.Mesoscale eddies are characterized by their circular or spiral motion, representing an intermediate size category situated between smaller-scale features like turbulence and larger-scale ocean circulation patterns.

Temporal scale
Our primary objective is to understand the timescale of eddy evolution originating from the instability in the Adriatic western current, leading to the formation of eddies.Summer proves to be the optimal season for monitoring, given that eddies reach full formation during this period, exhibiting diameters ranging 100 km.We aim to comprehend the development of the eddy between 20 June 2015, and 5 July 2015, through the analysis of daily data.The selection of this 16-day time frame results from a trial-and-error process, guided by the understanding that the timescale of mesoscale eddies in the ocean varies from several days to a few months.
On 20 June, no anticyclonic eddy is observed in the Gulf, but a weak cyclonic eddy is formed.The absence of eddies on 21 June is attributed to a north-east to southwest current, possibly influenced by the shifting Adriatic western current.However, on 23 June, an eastward current prevails in the Gulf.This occurrence, covering the southern basin, is associated with atmospheric conditions, specifically extreme winds, temporarily altering the Gulf 's circulation.On 24 June 2015, an anticyclonic eddy emerges in the Gulf, initially with a speed below 0.1 m/s.The eddy's growth continues, reaching a speed of 0.15 m/s on 25 and 26 June.By 30 June, the formation of the eddy is evident, indicating its origin from the western current.The western current, after separating from the Gargano cape (Fig. 1), becomes unstable, touches the south part of the gulf, and initiates the formation of a circular current (eddy).On 1 July, a current with a speed of 0.4 m/s attempts to penetrate the gulf.Interestingly, (1) d = 2 A the eddy's speed varies across the area, with the eastern region experiencing a higher speed than the west.By July 5th, a fully developed anticyclonic eddy becomes apparent, and the rotating current's speed stabilizes at around 0.3 m/s.
In summary, the main eddy formation at the monthly scale initiates around 29 June and undergoes a five-day evolution.This indicates that it takes approximately six days for the instability arising in the western flow of the Adriatic Sea, as it separates from the Gargano cape, to complete its formation and stabilize (Fig. 4).

Gulf's salinity fluctuation
The purpose of this section is to present a time series analysis of the variations in the physical properties of water within the Gulf of Manfredonia.To study the variability in salinity in water, the investigation distinguishes between the surface layer and the average across all layers (25 layers).The study area encompasses the Gulf of Manfredonia (Fig. 1), based on monthly salinity averages from 2014 to 2017.In the Gulf of Manfredonia, the salinity reaches its highest level in 2017 and its lowest in 2014.
Surface salinity levels exhibit variations ranging from 36.8 to 38.2 P.S.U (Fig. 5a).The minimum salinity is recorded in October 2014, while the maximum is observed in September 2017.In January 2014, the salinity stands at approximately 38 P.S.U.The value then experiences a declining trend, reaching 37.5 P.S.U in April 2014, followed by an increase to 38.2 P.S.U in August 2014.Subsequently, the Gulf undergoes significant fluctuations, reaching a sharp drop to 36.8 P.S.U in October 2014, initiating a subsequent trend of salinity increase.In 2015, the salinity peaks at 38 P.S.U in June, the highest for the year, before dropping to its lowest point of 37.5 P.S.U in October and November.The salinity trend continues its upward trend in 2017, reaching 38.25 P.S.U in September before decreasing.
However, when considering the average salinity across all layers in the Gulf, fluctuations range from 37.55 P.S.U in October to 38.2 P.S.U in August (Fig. 5b).Notably, a noteworthy aspect is the remarkable 0.6 P.S.U in salinity from August to October.In 2015, the salinity trend closely resembles that of 2014.In 2016, the highest salinity is recorded in March at 38.15 P.S.U, while the lowest is noted in September at 37.85 P.S.U.The subsequent period witnesses a rise in salinity, reaching 38.15 P.S.U in September 2017.
In conclusion, the observed fluctuations in Gulf water salinity, with the highest recorded in 2017 and the lowest in 2014, lead us to anticipate that the Po River discharge in 2014 was higher than in 2017.If this holds true, it supports our hypothesis regarding the impact of the Po River on the Gulf, given the primary role of the Po River in reducing salinity in the Adriatic Sea.

Gulf's temperature changes
Similar to salinity, the investigation into temperature changes in the Gulf of Manfredonia involve analysing both the mean surface layer and total layers.The average temperature is calculated in the Gulf of Manfredonia using monthly averages from 2014 to 2017.The most remarkable fluctuations are observed in the surface layer.In August 2015, the highest temperature reaches 30 ℃ in the surface layer, while in September 2016, it records 21.2 ℃ across all layers.
Considering the surface layer in 2014, the lowest temperature of 13.8 ℃ occurs in February, with the highest temperature in August reaching approximately 26 ℃ (Fig. 6a).In 2015, the lowest temperature in the surface layer is observed in March at 13 ℃ while the highest temperature in August 2015, which peaked at 30 ℃the highest recorded temperature and notably 2 to 3 ℃ higher than August in other years.In 2016, the minimum temperature in March is about 0.5 °C higher than in Fig. 4 Evolution of the anticyclonic eddy in the Gulf from 20 June 2015, to 5 July, 2015, plotted using daily average data March 2015, while the maximum temperature in August is 2.5 °C lower.Furthermore, in 2017, the maximum and minimum temperatures occur in the same months as in 2016, but the minimum temperature is 0.5 ℃ cooler, and the maximum temperature is 0.5 ℃ higher.
Contrastingly, when considering the average across all layers, temperature fluctuations are less pronounced (Fig. 6b).In 2014, the lowest temperature in March is 13.5 °C, and the maximum temperature occurs in September, reaching 20.8 °C.In 2015, March records a temperature of 12.8 °C, with the maximum temperature reaching 21 °C in September.In 2016, both the minimum and maximum temperatures occur in the same months as in 2014, but the minimum temperature is 0.5 °C warmer, and the maximum temperature is 0.2 °C higher.However, in 2017, the minimum and maximum temperatures in March and September are 1 °C cooler than corresponding months in 2016.

Gulf's sea level variation
In this section, we analyse the fluctuations in the surface level of the Gulf of Manfredonia.The sea level holds significance, especially if the hypothesized influence of Po River discharge fluctuations on the Gulf is valid.This is primarily since variations in Po's discharge leads to changes in transport, potentially impacting the sea surface.To explore these effects, we use Zeta as the optimal parameter for depicting and studying sea level variations.Overall, the surface level is at its peak in 2014 and reaches its lowest point in 2017, with the highest recorded in April 2014 and the lowest documented in September 2017 (Fig. 7a).
In 2014, the surface level, averaging about 1 cm in the Gulf, increase to 3.2 cm in April, marking the highest value.Subsequently, it exhibits a declining trend, reaching -1 cm in August, which is the lowest level.
The trend then reverses, showing an upward trend to 1.5 cm in December.Moving on to 2015, the surface level starts at 0.5 cm in January, escalated to 2.5 cm in February-its peak for the year-before experiencing a descending trend, reaching its lowest at −1.5 cm in September.
In 2016, the surface level is at 0.5 cm in January, reached its maximum at 1.5 cm in April, and then fluctuates to its lowest point at −1.5 cm in November-the lowest value for 2016.For 2017, the surface level begins at −1.5 cm in January, displays an ascending trend to zero in April, and then descends to −3 cm

Gulf's vertical velocity fluctuation
Another important parameter for understanding the Gulf 's hydrodynamics and the potential influence of Po River fluctuations is vertical velocity near surface.This parameter is significant not only for comprehending physical properties such as downwelling and upwelling but also plays a crucial role in understanding biological features.Downwelling, indicated by negative vertical velocity, transports oxygen to the deep water, while upwelling (positive vertical velocity) brings nutrients from the deep water to the surface.The model directly calculates this quantity and the average vertical speed in the Gulf has been computed using monthly data (Fig. 7b).
In general, the highest positive vertical speed is recorded in February, while the lowest (most negative) vertical speed occurs in June 2014.In 2014, the vertical velocity begins at 0.6 × 10 -4 cm/s, increases to 0.7 × 10 -4 cm/s in February-its highest value for the year.Subsequently, it decreases to −1.2 × 10 -4 cm/s in June, representing the most negative vertical speed.After fluctuating, it reaches −0.3 × 10 -4 cm/s in December 2014.
Moving on to 2015, the vertical velocity starts at −0.5 × 10 -4 cm/s in January, approached zero in February, and then, with a decreasing trend, reaches −1.3 × 10 -4 cm/s in July, eventually transitioning to a positive vertical velocity.In 2016, the vertical velocity begins at about 0.2 × 10 -4 cm/s in January, shifting towards negative velocities, reaching −1 × 10 -4 cm/s in August-the most negative value for the year.Afterward, the velocity shifts towards positive values.In 2017, it is −0.5 × 10 -4 cm/s in January, transitioning towards positive speeds and reaching 0.2 × 10 -4 cm/s in March.Subsequently, it reaches the most negative vertical speed in August, which is −1.2 × 10 -4 cm/s.

Influence of po river discharge on hydrodynamic fluctuations in the gulf
In the preceding section, we presented evidence indicating changes in the physical properties of water in the Gulf during 2014-2017.Firstly, the observed lowest salinity in 2014 and the highest salinity in 2017 are likely linked to fluctuations in the Po River discharge.Secondly, the elevated water level of the Po River in 2014 and its reduced level in 2017 could be influenced by the Po River discharge fluctuation.If we show that the Po River experienced fluctuations during 2014-2017, and these changes are consistent with the Gulf 's hydrodynamics, our primary hypothesis that the Po River discharge affects the Gulf 's hydrodynamics becomes more robust.In this section, we seek to substantiate the validity of our hypothesis.

Po river discharge fluctuation
We analyse the variations in Po River discharge using averaged data from 2014 to 2017 (Fig. 8).The maximum discharge from the Po River is in November 2014 while the minimum discharge is in August 2017.
In January 2014, the discharge begins at 2,750 m 3 /s, gradually increasing to 3,750 m 3 /s in February.Subsequently, with a declining trend, it reaches its lowest value of 1,300 m 3 /s in both June and September of 2014.However, a significant surge occurs in the river's flow from October to November, tripling to 4,500 m 3 /s.In 2015, the river's discharge fluctuates between 500 and 1800 m 3 /s, with the lowest recorded in July and the highest in December.Moving to 2016, the lowest value is noted in July at 600 m 3 /s, while the highest is in June at 1,900 m 3 /s.Notably, in 2017, the Po River exhibits its lowest discharge, reaching 500 m 3 /s in August, and subsequently rising to 1100 m 3 /s.
If we observe the discharge trend from 2014 to 2017, we see a downward trend.This implies that the highest discharge in 2017 corresponds to the lowest salinity in the Gulf in 2014, and conversely, the highest salinity in the Gulf aligns with the lowest Po River discharge in 2017.This indicates the validity of our hypothesis.

Net effects of po
An alternative approach to enhance our comprehension of the Po River's impacts and validate our hypothesis is to illustrate the net effects of the Po River on the temperature, salinity, and circulation of the gulf.We designed two model scenarios-one incorporating the Po River and another omitting it.Subsequently, we computed the temperature, salinity, and velocity differences by subtracting the outputs of the Po River-inclusive model from the Po River-excluded model.We replicate the model for the years 2016 and 2017, limiting our focus only to the outcomes of the year 2017, specifically in seasonal averages.The simulations considered identical boundary conditions for both scenarios, focusing solely on the Po and no-Po scenarios.
Temperature fluctuations vary in the Gulf depending on the presence or absence of the Po Rive (Fig. 9a).During the winter of 2017, the Po River contributes to a surface temperature increase of 1 ℃, whereas in spring, its presence leads to a decrease of 2 ℃.In summer, the Po River presence raises the Gulf temperature by up to 1.5 ℃, while fall witnesses a significant increase of up to 3 ℃.Conversely, the Po River has a notable impact on the salinity of the Gulf (Fig. 9b).Salinity decreases to 0.2 P.S.U in the Gulf during winter, rises to 0.4 P.S.U in spring, and experiences its most substantial decrease in summer, reaching 1 P.S.U salinity.This salinity difference peaks at 0.8 P.S.U in fall.However in the circulation, in the winter of 2017, the westerly current experiences an increase of approximately 0.04 to 0.1 m/s upstream of the gulf (Fig. 9c).However, the westerly current path's orientation results in a limited impact on the gulf circulations during this season.Moving into spring, the presence of the Po River strengths the western current of the Adriatic Sea by approximately 0.14 m/s.This enhancement is attributed to the Po River's increased discharge during the spring, leading to more pronounced effects on the gulf, particularly with a flow increase of around 0.04 m/s in the eastern parts of the gulf.
As we transition to summer, although the westerly current weakens compared to spring, it gains strength of about 0.1 m/s due to the influence of the Po River.Notably, the formation of anticyclonic eddies in the gulf contributes to a current strength of approximately 0.14 m/s, resulting in the most significant impact observed during the summer.In the autumn, the current upstream and preceding the Gargano cape strengthens by approximately 0.1 m/s in the presence of the Po River.Simultaneously, the current within the gulf experiences a boost of Hence, this simulation supports our hypothesis regarding the impact of Po River discharge fluctuations on the Gulf.Now, let's delve into the precise correlation between Po River discharge effects and the physical variability in the Gulf.

Correlation
In the "Circular Current: Eddy or Gyre?", we discussed variations in water temperatures, salinities, vertical velocities near the surface, and sea surface water levels in the study area.To what extent can these fluctuations be attributed to changes in the Po River's discharge fluctuation?For this reason, we examined the correlation between variations in Po River discharge and physical variables in the Gulf.We normalize salinity, temperature, water level changes, and vertical velocity near the surface by their maximum values.
There is a notable inverse relationship between the discharge of the Po River and the distribution of temperature and salinity in the Gulf.The salinity of the study area decreases with an increase in Po River discharge, with a correlation coefficient of −0.38 (Fig. 10a).Furthermore, a correlation of approximately −0.33 exists between temperature variations and Po River discharge (Fig. 10b), indicating a decrease in temperature in the studied area with an increase in Po River discharge.
In contrast, the Po River discharge shows a direct relationship with changes in sea surface level and vertical velocity near the surface (Fig. 10c).Sea surface fluctuation has a correlation coefficient of 0.59 with Po River discharge, while vertical velocity has a correlation coefficient of 0.54 with sea surface fluctuation (Fig. 10d).In the Gulf, Zeta and W increase when the Po River discharge increases, as observed in 2014, and decrease when the Po River discharge decreases, as observed in 2017.
Fluctuations in the Po River discharge affect the physical characteristics of the water in the Gulf of Manfredonia.Zeta and W have higher correlation coefficients compared to salinity and temperature, indicating a greater impact on sea surface fluctuation and vertical velocity.

River discharge fluctuation on chlorophyll-a concentration
In this section, our focus is on the biological aspects of the Gulf, specifically evaluating chlorophyll-a concentration using Aqua/Modis sensor data.Chlorophyll-a, a vital green pigment for photosynthesis in plants and algae, stands as a crucial indicator of phytoplankton biomass in aquatic ecosystems (Focardi et al. 2009).The algorithm employed calculates the near-surface concentration of chlorophyll-a in mg m −3 , relying on an empirical relationship derived from in situ measurements of chlorophyll-a and the blue-to-green band ratios of in situ remote sensing reflectance (Rrs).We use monthly averaged data with a grid spacing of approximately 4 km.

Chlorophyll-a changes
In January 2014, the chlorophyll-a concentration is approximately 4 mg m −3 , decreasing to 2-3 mg m −3 in 2015, rising to 4-5 mg m −3 in 2016, and remaining below 5 mg m −3 in 2017 (Fig. 11).In April 2014, most areas exhibit chlorophyll-a levels below 1 mg m −3 , but a few Fig. 11 Chlorophyll-a concentration in the study area (mg/m 3 ) for selected months across different years parts in the northern region reach about 2 mg m −3 .In 2015, chlorophyll-a is around 2 mg m −3 along the coast and less than 1 mg m −3 farther from the Gulf coast.
Both in 2016 and 2017, chlorophyll-a near the coast is approximately 2.5 mg m −3 , with the highest concentration observed in April.However, in August, chlorophyll-a levels are generally less than 2 mg m −3 , slightly higher in 2017.
In December, there is a noticeable increase in chlorophyll-a throughout the entire the gulf.Notably, in December 2014, chlorophyll-a concentration is unusually five times higher than December in other years, measuring around 10 mg m −3 .In December 2015, chlorophyll-a typically registers at 2 mg m −3 but is concentrated along coastal areas.For December 2016 and 2017, chlorophylla levels around 2 mg m −3 cover most of the gulf.
The crucial question arising here is whether these fluctuations are related to the Po River discharge fluctuation.To address this, we calculated the average chlorophyll-a concentration in the gulf and compared it with the time series of Po River discharge changes (Fig. 12).The analysis reveals a noteworthy observation: the peak in chlorophyll-a concentration occurs in December 2014, while the peak in Po River discharge is in November 2014, indicating a one-month time lag.Consequently, it can be a corelation between Po's discharge and the chlorophyll-a concentration in the gulf.

Correlation
To enhance understanding, our objective is to calculate the correlation between fluctuations in Po River discharge and chlorophyll-a concentration.We use two distinct methods: one without shifting months and the other incorporating a one-month shift.This comes from that in the "River discharge fluctuation on chlorophyll-a concentration".
In employing the first method (without shifting), simultaneous evaluation reveals a direct correlation (0.27) between increased Po River discharge and elevated chlorophyll-a levels (Fig. 13a).The use of the second method (with one-month shifting) results in a doubled correlation coefficient, underscoring the direct influence of rising Po River discharge on chlorophyll-a levels within a month of its occurrence (Fig. 13b).This impact is notably evident, for instance, in February 2014, reflecting the effect of river discharge in January 2014 on chlorophyll-a levels.
As discussed in "Gulf 's Vertical Velocity Fluctuation", the vertical speed emerges as a critical factor influencing biological features, such as chlorophyll-a levels in the Gulf.Considering this, we conducted calculations to determine the correlation between vertical speed and chlorophyll-a levels.The exploration revealed a positive correlation (0.39), indicating that elevated surface vertical speed aligns with increased chlorophyll-a levels in the Gulf (Fig. 13c).
In summary, a connection exists between the fluctuations in Po River discharge and chlorophyll-a levels in the Gulf.Furthermore, an increase in vertical speed correlates with an elevation in chlorophyll-a concentration, with the time lag being of negligible significance in this corelation.noteworthy that the conditions upstream of the current play a pivotal role in shaping the eddy, as indicated by laboratory experiment results presented by Babagolimatikolaei and Bidokhti (2019).
We proved the Po River's impact on the Gulf based on three important facts.Initially, the salinity variations, ranging between 36.8 and 38.2 P.S.U, showcased the lowest levels in 2014 and the highest in 2017, establishing an inverse correlation with the fluctuations in Po River discharge (correlation coefficient: −0.38).Secondly, we saw changes in surface levels ranging from -3 to 3 during the study period.The maximum sea level was recorded in 2014, while the minimum occurred in 2017.These alterations are attributed to increased water volume from the Po River entering the Adriatic Sea, influencing eddy formation, and subsequently affecting water levels, correlating with the Po River discharge fluctuations (correlation coefficient: 0.59).Thirdly, these empirical findings align with results from limited simulations, indicating a westerly current strengthened by 0.14 m/s and a 20% contribution to the strength of the summer eddy due to the presence of the Po River.
To consider the impact of the Po River, satellite data was employed as the ocean model lacked the capability to simulate chlorophyll-a.Analysis using satellite data highlighted the sensitivity of chlorophyll-a to the Po River's discharge.The results demonstrated a correlation of 0.27 when considering Po River discharge without a one-month time lag.Considering the time lag increased the correlation to 0.55, indicating that an increase in Po River discharge in the subsequent month corresponds to a rise in chlorophyll-a levels in the gulf.The time lag is likely due to the geographical distance of approximately 500 km between the Po River and the Gulf, where nutrients require time to travel or facilitate growth.Furthermore, it's essential to acknowledge that the discharge from the Po River can create favorable physical conditions for phytoplankton growth beyond just transporting nutrients.
However, considering constraints due to inadequate data and the model's incapacity to simulate chlorophylla, definitive conclusions cannot be drawn with high certainty.Also, we should consider that MODIS might overestimate chlorophyll-a levels in coastal waters due to the influence of turbidity, which could potentially introduce errors in our analysis.In coastal regions, where sediment runoff, river discharge, and other particulate matter are common, turbidity levels can vary widely and significantly impact the accuracy of chlorophyll-a estimates.As a result, MODIS may mistakenly interpret the increased scattering caused by turbidity as higher chlorophyll-a concentrations, leading to overestimation in these areas.Therefore, definitive conclusions regarding chlorophyll-a cannot be confidently drawn, necessitating further research.
The core message conveyed by this paper underscores the sensitivity of the Gulf 's hydrodynamic changes to the variability of the Po River.These fluctuations exert influence not only on the physical attributes but also on the biological aspects of the Gulf, albeit with a need for further in-depth investigation into the latter about chlorophyll-a.In the future, expected climate changes suggest that this gulf and its ecosystems will become more sensitive, potentially affecting the organisms living in its waters.

Fig. 2 a
Fig. 2 a Model error at various depths for salinity and (b) temperature.Errors are calculated by subtracting observed values from model values.Positive values indicate that the observed values are higher than the model calculations.Index refers to the month of 2017, where "data1" corresponds to January 2017, "data2" to February, and "index 12" denotes December 2017

Fig. 5 a
Fig. 5 a Fluctuations in surface salinity, b average salinity across all layers.The monthly-averaged data is used for plotting

Fig. 6 a
Fig. 6 a Changes in surface temperature, b average temperature across all layers.The monthly-averaged data is used for plotting

Fig. 7 a
Fig. 7 a The fluctuation of Zeta and b Vertical Velocity (W) using monthly-averaged data in the Gulf

Fig. 8
Fig. 8 Variation in Po River discharge from 2014 to 2017, along with its trend line

Fig. 9 a
Fig. 9 a Net Impact of the Po River on the temperature, b salinity (c) Circulation Patterns in the Study Area.The Po River model outputs are subtracted from the results obtained without the Po River

Fig. 10 a
Fig. 10 a Correlation of salinity, b temperature, c Zeta, d W with discharge.All values are non-dimension with its maximum value

Fig. 12
Fig. 12 Time series of chlorophyll-a and discharge during various months of the study period, presented in a non-dimensional number with their respective maximum values