Design of a multipoint cost-effective optical instrument for continuous in-situ monitoring of turbidity and sediment

sensors Article Design of a Multipoint Cost-Effective Optical Instrument for Continuous In-Situ Monitoring of Turbidity and Sediment T. Matos 1,* , C. L. Faria 1, M. S. Martins 1, Renato Henriques 2, P. A. Gomes 3and L. M. Goncalves 1 1MEMS-UMinho, University of Minho, Campus de Azurém, 4800-058 Guimarães, Portugal; [email protected] (C.L.F.); [email protected] (M.S.M.); [email protected] (L.M.G.) 2Institute of Earth Sciences, University of Minho Pole, Campus de Gualtar, 4710–057 Braga, Portugal; [email protected] 3Centre of Molecular and Environmental Biology (CBMA), University of Minho, 4710–057 Braga, Portugal; [email protected] *Correspondence: matos.tiagoandr[email protected] Received: 29 April 2020; Accepted: 3 June 2020; Published: 4 June 2020   Abstract: A cost-effective optical instrument for continuous in-situ monitoring applications is presented. With a production cost in raw materials of 38 € , a power consumption of 300 µ A in sleep mode and 100 mA in active mode (5 ms reading), and a capacity to monitor turbidity and sedimentary displacement at eight different depths in the water column, the sensor was developed for sediment monitoring in coastal areas. Due to the extent and dynamics of the processes involved in these areas, observations require a wide spatial and temporal resolution. Each of the eight monitoring nodes uses one infrared backscatter channel, to estimate turbidity and sediment concentration, and one ultraviolet with one infrared transmitted light channels to distinguish organic/inorganic composition of the suspended material load. An in-lab calibration was conducted, using formazine to correlate turbidity with the electronic outputs of the instrument. An analysis of the influence of external light sources and correction techniques were performed. Moreover, an in-lab experiment was conducted to study the behaviour of the sensor-to-sediment transport, wash load and sediment accumulation. The device was deployed, with a water level sensor, in an estuarine area with high sediment dynamics. The monitoring data were analysed, showing the potential of the device to continuously monitor turbidity, sediment processes, and distinguish between organic and inorganic matter, at the different depths in the water column. Keywords: turbidity sensor; turbidity; sediment processes; suspended sediment; optical sensor; oceanography 1. Introduction Sediment is a set of naturally occurring particles that are broken down from rocks by weathering and erosion or formed by natural chemical processes or by biological processes. These particles are subsequently transported by the action of wind, water or ice, or by the force of gravity and form deposits when the transportation agents weaken or stop acting [ 1 – 3 ]. While the term is often used to indicate rock originated minerals with several dimensions, such as clay, silt and sand, decomposing organic substances and inorganic or biogenic materials are also considered sediment [ 4 ]. Sediment materials are typically small, with clay defined as particles less than 0.00195 mm in diameter, and coarse sand reaching up to 2 mm in diameter [ 5 ]. However, during floods or other highly energetic events, large rock fragments are also classified as sediments once they are detached from a previous rock formation and are Sensors 2020,20, 3194; doi:10.3390/s20113194 www.mdpi.com/journal/sensors Sensors 2020,20, 3194 2 of 16 carried downstream. Whenever any of these particles are carried in the course of the water (in liquid or solid-state) or any other moving fluid, such as air (wind), it is called sediment transport. Sediment transport, or sediment load, is the movement of organic and inorganic particles in a moving mean, such as water [ 6 ]. In general, the higher the flow rate, the more sediment can be transported, as well as larger particles. Sediment transportation in an aquatic environment can be strong enough to suspend particles in the water column, as they move downstream, or simply push them along the bottom of the waterway [ 7 ]. Sediment transport can be divided into three different processes: bed load, suspended load and wash load. Depending on the characteristics and properties of the streamflow, carried material, and the watershed, the sediment transport can be divided into three different types: bed load, suspended load and wash load [8]. Bedload is the portion of sediment transport that rolls, slides or bounces along the bottom of the waterway [ 9 ]. This sediment is not considered suspended, as it is in constant contact with the streambed, and its movement is neither uniform nor continuous. Bedload occurs when the force of the water flow is strong enough to overcome the weight and cohesion of the sediment, making it move [ 10 ]. In situations where the flow rate is strong enough, some of the smaller and lighter particles that are settled in the stream floor can be pushed up into the water column and become suspended. The size of the particles that can be carried as the suspended load is dependent on the flow rate (the higher the velocity of the water body, the higher the suspending force applied to the sediment material). Larger and heavier particles are more likely to fall through the upward currents to the bottom, unless the flow rate increases, increasing the turbulence at the streambed. Moreover, suspended sediment will not necessarily remain suspended if the flow rate slows [11]. The wash load is a subset of the suspended load. It is comprised of the finest suspended sediment, typically less than 2 µ m in diameter [ 12 ]. The wash load differentiates from the suspended load because it will not settle to the bottom of a waterway during a low or no flow period. Instead, these particles remain in permanent suspension as they are small enough to bounce on water molecules and stay afloat. As represented in Figure 1, suspended sediment can have different behaviours at different depths along the water column of streamflow. In estuarine and seashore areas, materials are subject not only to the strength of the downstream course but also to the maritime currents, tide and undulation, the different watershed and geological dynamics, seasonal and climatic effects and occurrence of storms, pollution and other anthropic induced processes. For these reasons, sediment processes are difficult to study, model and predict [ 13 ], to have proactive management and protection of the coastal areas as their associated societies and ecosystems. Sensors 2020, 20, x FOR PEER REVIEW 2 of 17 of the water (in liquid or solid-state) or any other moving fluid, such as air (wind), it is called sediment transport. Sediment transport, or sediment load, is the movement of organic and inorganic particles in a moving mean, such as water [6]. In general, the higher the flow rate, the more sediment can be transported, as well as larger particles. Sediment transportation in an aquatic environment can be strong enough to suspend particles in the water column, as they move downstream, or simply push them along the bottom of the waterway [7]. Sediment transport can be divided into three different processes: bed load, suspended load and wash load. Depending on the characteristics and properties of the streamflow, carried material, and the watershed, the sediment transport can be divided into three different types: bed load, suspended load and wash load [8]. Bedload is the portion of sediment transport that rolls, slides or bounces along the bottom of the waterway [9]. This sediment is not considered suspended, as it is in constant contact with the streambed, and its movement is neither uniform nor continuous. Bedload occurs when the force of the water flow is strong enough to overcome the weight and cohesion of the sediment, making it move [10]. In situations where the flow rate is strong enough, some of the smaller and lighter particles that are settled in the stream floor can be pushed up into the water column and become suspended. The size of the particles that can be carried as the suspended load is dependent on the flow rate (the higher the velocity of the water body, the higher the suspending force applied to the sediment material). Larger and heavier particles are more likely to fall through the upward currents to the bottom, unless the flow rate increases, increasing the turbulence at the streambed. Moreover, suspended sediment will not necessarily remain suspended if the flow rate slows [11]. The wash load is a subset of the suspended load. It is comprised of the finest suspended sediment, typically less than 2 µm in diameter [12]. The wash load differentiates from the suspended load because it will not settle to the bottom of a waterway during a low or no flow period. Instead, these particles remain in permanent suspension as they are small enough to bounce on water molecules and stay afloat. As represented in Figure 1, suspended sediment can have different behaviours at different depths along the water column of streamflow. In estuarine and seashore areas, materials are subject not only to the strength of the downstream course but also to the maritime currents, tide and undulation, the different watershed and geological dynamics, seasonal and climatic effects and occurrence of storms, pollution and other anthropic induced processes. For these reasons, sediment processes are difficult to study, model and predict [13], to have proactive management and protection of the coastal areas as their associated societies and ecosystems. Figure 1. Sketch of sediment transport in water. The quantification of sediment transport and deposition in the seaside area is essential to understand the evolution of the littoral coast [14]. However, it has been verified that the correlation between the modelling data and the measurements made in the field is not always satisfactory, Figure 1. Sketch of sediment transport in water. The quantification of sediment transport and deposition in the seaside area is essential to understand the evolution of the littoral coast [ 14 ]. However, it has been verified that the correlation between the modelling data and the measurements made in the field is not always satisfactory, mainly Sensors 2020,20, 3194 3 of 16 due to the great variability and complexity of sedimentary processes, the variation of sea conditions from place to place [ 15 , 16 ] and the lack of more regular observation time series. Therefore, continuous and high spatial resolution monitoring becomes essential for an effective study in each area of action. This monitoring can only be achieved by autonomous electronic sensors, able to provide reliable data of continuous measurements in the field. The current state of the art of in situ monitoring sensors, to measure sediment processes, relates mainly to acoustic and optical technologies. The acoustic ones, typically referred as Acoustic Backscatters (ABS), use a piezoelectric transducer that emits ultrasonic pulses that are reflected by the suspended material, and one or more ultrasound receivers sense the acoustic echoes, to estimate the amount of sediments, and in some cases, its size [ 17 , 18 ]. The major disadvantage of this technology is its high cost and complexity, compared to the optical technology [ 19 ]. The optical sensors aim to measure turbidity, which can provide an indirect estimation of sediment concentration and transport [ 20 ]. This type of sensor uses a light source at a certain frequency to illuminate the sample, where the suspended material will absorb and scatter the light, which is measured by optical receivers to provide the turbidity measurement. Due to its ease of use, smaller size, higher energy efficiency and lower cost, the optical turbidity sensors are the most popular instruments to measure sediment/turbidity in situ. Commercial brands such as Seabird, Seapoint Sensors, Valeport, s::can, among others, offer a wide range of turbidity sensors for in situ purposes. While these instruments offer a good precision, accuracy and the necessary conditions for continuous monitoring, their price is still a challenge for massive deployments (the low-cost series offered are typically above 1200 € ). With the challenge of reducing costs to enable the replication of these sensors and to cover a wide spatial resolution, in recent years the scientific community has been dedicated to developing low-cost oceanographic instruments. Matos et al. [ 21 ] presented a low-cost (less than 20 € ) turbidity sensor for coastal monitoring. The device was developed to continuously monitoring water parameters in situ, and is integrated in the Next-Sea Project [ 22 – 24 ], which has the main philosophy of developing cutting edge technology that is low-cost, for massive replication and deployments, low-power, to extend the lifetime of the batteries and reduce maintenance needs, low-size and light, for ease of use and installation, with no need of big ships or cranes, and fully submersible, to avoid floater restrictions. The previously developed sensor, referred to hereinafter as the SPM Sensor, is an optical single-point device that measures turbidity (in NTU or another unit, provided that calibration is established), suspended particle matter concentration (in g/L) and distinguishes between organic and inorganic suspended material (also in g/L). The presented manuscript presents an innovative device that takes advantage of the technology used in the SPM Sensor, but allows to monitor turbidity and sediment processes at different depths in the water column. While the vertical replication of the SPM Sensor could be an approach, the newly developed instrument presents advantages in ease of use and installation, energetic efficiency and cost. 2. Sensor Design The developed instrument is a continuation of the previously SPM Sensor [ 21 ] that uses three light detection techniques to measure turbidity, sediment concentration and distinguishes between organic/inorganic sediment. The SPM Sensor was developed in a radial configuration, using one infrared backscattering channel (optical receiver placed at 135 ◦ related to the emitter), one infrared nephelometric channel (optical receiver placed at 90 ◦ related to the emitter), and one ultraviolet and one infrared transmitted light channels (optical receiver placed at 0 ◦ related to the emitter). The use of multiples angles to measure turbidity allows a wide dynamic range and precision [25]. The backscattering technique relates to the measurement of scattered light by the suspended material [ 26 ]. It offers a wide range of measurement and it is the most popular technique used in commercial turbidity sensors (typically referred to as Optical Backscatter Point Sensors—OBS). On the other hand, the nephelometric technique, which also measures scatter and diffuse light but at a different angle, is typically used for high-accuracy and low-turbidity values and is mostly used for inline sensors on wastewater and water treatment plants [ 27 , 28 ]. Finally, the transmitted channels measure the attenuation Sensors 2020,20, 3194 4 of 16 of light in a straight path, by the scattering and absorption caused by suspended material. This technique is not very popular to measure turbidity, due to its high sensitivity to particle size, shape and colour. For the SPM Sensor, it was used to distinguish between organic and inorganic matter, taking advantage of the different responses of matter to different wavelengths (ultraviolet and infrared) [29]. SPM Sensor results showed that while to have a wide dynamic range to measure turbidity the backscattering, nephelometric and transmitted light techniques should be used, all three techniques have demonstrated good performance as standalone sensing systems. For this reason, a new design was adopted for the developed instrument. The ultraviolet and infrared transmitted light channels were maintained to distinguish between organic and inorganic matter. For turbidity and sediment concentration, the new device only uses the backscatter technique. Since the nephelometric technique has a high performance for low turbidity values that are not expected in coastal areas, this light detection technique was not used (see Figure 2). Sensors 2020, 20, x FOR PEER REVIEW 4 of 17 the other hand, the nephelometric technique, which also measures scatter and diffuse light but at a different angle, is typically used for high-accuracy and low-turbidity values and is mostly used for inline sensors on wastewater and water treatment plants [27,28]. Finally, the transmitted channels measure the attenuation of light in a straight path, by the scattering and absorption caused by suspended material. This technique is not very popular to measure turbidity, due to its high sensitivity to particle size, shape and colour. For the SPM Sensor, it was used to distinguish between organic and inorganic matter, taking advantage of the different responses of matter to different wavelengths (ultraviolet and infrared) [29]. SPM Sensor results showed that while to have a wide dynamic range to measure turbidity the backscattering, nephelometric and transmitted light techniques should be used, all three techniques have demonstrated good performance as standalone sensing systems. For this reason, a new design was adopted for the developed instrument. The ultraviolet and infrared transmitted light channels were maintained to distinguish between organic and inorganic matter. For turbidity and sediment concentration, the new device only uses the backscatter technique. Since the nephelometric technique has a high performance for low turbidity values that are not expected in coastal areas, this light detection technique was not used (see Figure 2). Figure 2. Design of the sensor. It uses two infrared channels: one IR LED (2) and two transducers that measure optical backscattering (3) and transmitted light (5); and one ultraviolet channel: UV emitter (1) and wideband receiver (4), adapted from [21]. The developed new instrument integrates several sensing nodes along a one-body and compact structure controlled by a single microprocessor. The developed device is a 645 mm × 55 mm ×× 15 mm bar, with eight measuring nodes displaced vertically by 70 mm from each other (see Figure 3). Each node has a 30-mm diameter and comprises the optical transducers presented in Figure 2. The instrument was built in a scale and modular philosophy, so that similar bars can be fixed on the top of the previous one (from stream bottom to surface) and increase the number of nodes and depth monitoring along the water column. When using more than one bar, all of them are serial connect to pass information from bottom to top, and all the connected bars will be recognized as a single longer bar. Figure 2. Design of the sensor. It uses two infrared channels: one IR LED (2) and two transducers that measure optical backscattering (3) and transmitted light (5); and one ultraviolet channel: UV emitter (1) and wideband receiver (4), adapted from [21]. The developed new instrument integrates several sensing nodes along a one-body and compact structure controlled by a single microprocessor. The developed device is a 645 mm × 55 mm × 15 mm bar, with eight measuring nodes displaced vertically by 70 mm from each other (see Figure 3). Each node has a 30-mm diameter and comprises the optical transducers presented in Figure 2. The instrument was built in a scale and modular philosophy, so that similar bars can be fixed on the top of the previous one (from stream bottom to surface) and increase the number of nodes and depth monitoring along the water column. When using more than one bar, all of them are serial connect to pass information from bottom to top, and all the connected bars will be recognized as a single longer bar. The structural housing of the sensor was manufactured with epoxy, a resin material to comprise the electronic printed circuit board (PCB) and to meet the watertight needs for full submersion. This structure, installed vertically and along the water column, allows to perform the same measurements as the previous developed SPM Sensor, but at different depths and using a single apparatus, instead of using multiple sensors. One objective of the new sensor development was for it to be low-cost to allow massive replication. While the SPM Sensor has a production cost of 20 € , in raw materials, and performs a single point measurement, the new device has a production cost of 38 € and take measurements at eight different levels of depth. Sensors 2020,20, 3194 5 of 16 Sensors 2020, 20, x FOR PEER REVIEW 5 of 17 Figure 3. The developed instrument comprises 8 monitoring nodes, with the backscatter and transmitted light channels, along a 645 mm bar, to be placed vertically in the water column (the instrument is presented in this figure in the horizontal). The structural housing of the sensor was manufactured with epoxy, a resin material to comprise the electronic printed circuit board (PCB) and to meet the watertight needs for full submersion. This structure, installed vertically and along the water column, allows to perform the same measurements as the previous developed SPM Sensor, but at different depths and using a single apparatus, instead of using multiple sensors. One objective of the new sensor development was for it to be low-cost to allow massive replication. While the SPM Sensor has a production cost of 20 €, in raw materials, and performs a single point measurement, the new device has a production cost of 38 € and take measurements at eight different levels of depth. 2.1. Hardware In the hardware, changes were made concerning the previous SPM Sensor. For the optical transducers, while T1-3/4 (5 mm) packages were used previously, for this new device, T1 (3 mm) transducers were used to minimize its size. For the infrared channels, the light-emitting diode (LED) TSUS4300 (950 nm, 16° emitting angle and 18 mW/sr radiant intensity at 100 mA) and the phototransistor TEFT4300 (950 nm, view angle of 30° and 1 nA dark current) were selected, see Figure 2. For the ultraviolet pair, the LED UV3TZ-390-30 (390 nm, 15° emitting angle and 10 mW emitting power at 15 mA) and the wideband phototransistor SFH 3310 (spectral range of sensitivity from 350 to 900 nm, view angle of 75° and 3 nA dark current) were used. Notice that the use of LEDs was maintained, instead of LASERs, due to their good performance in SPM Sensor, wide commercial offer and lower price. For efficient management of all nodes and respective optical emitters and receivers, a microprocessor was carefully chosen. The smt32l496zg from the stm32 L4 family was selected due to its low power, 24 × 12-bit ADC channels (3 photodetector analogic signals × 8 measuring nodes requires 24 ADC channels) and several GPIOs (8 IR LEDs and 8 UV LEDs to be controlled). For a different approach, a simpler microprocessor could be used complemented with a multiplexer; however, the sensitivity of the instrument could be affected due to the internal resistance of the multiplexer and it would slightly increase the cost and power consumption of the sensor. The sensor was intended to be supplied by a common 3.6 V Li-Ion battery. For the power supply circuit, the TPS73630DBVT low-dropout regulator (3 V fixed output, 400 mA max current, 75 mV dropout voltage and 200 uA quiescent current) was used, as presented in Figure 4. While the whole system works with 3 V, the UV LED UV3TZ-390-30 has a forward voltage of 3.4 V so a higher supply is needed. For this purpose, the TPS61222DCKR boost convertor was used to supply the UV LEDs circuit (95% efficiency, 5.5 uA quiescent current and 200 mA switching current). Voltage regulators were used since constant voltage is mandatory to minimise LED intensity variations. The LED circuits are controlled by the microprocessor GPIOs that switches a low-on resistance N-channel MOSFET (≈30 mΩ) to turn them on and off. The phototransistor instrumentation circuited was designed with a simple current-to-voltage converter using a gain resistor. The resistor voltage potential is read by de microprocessor ADCs to be processed (Figure 4). Figure 3. The developed instrument comprises 8 monitoring nodes, with the backscatter and transmitted light channels, along a 645 mm bar, to be placed vertically in the water column (the instrument is presented in this figure in the horizontal). Hardware In the hardware, changes were made concerning the previous SPM Sensor. For the optical transducers, while T1-3/4 (5 mm) packages were used previously, for this new device, T1 (3 mm) transducers were used to minimize its size. For the infrared channels, the light-emitting diode (LED) TSUS4300 (950 nm, 16 ◦ emitting angle and 18 mW/sr radiant intensity at 100 mA) and the phototransistor TEFT4300 (950 nm, view angle of 30 ◦ and 1 nA dark current) were selected, see Figure 2. For the ultraviolet pair, the LED UV3TZ-390-30 (390 nm, 15 ◦ emitting angle and 10 mW emitting power at 15 mA) and the wideband phototransistor SFH 3310 (spectral range of sensitivity from 350 to 900 nm, view angle of 75 ◦ and 3 nA dark current) were used. Notice that the use of LEDs was maintained, instead of LASERs, due to their good performance in SPM Sensor, wide commercial offer and lower price. For efficient management of all nodes and respective optical emitters and receivers, a microprocessor was carefully chosen. The smt32l496zg from the stm32 L4 family was selected due to its low power, 24 × 12-bit ADC channels (3 photodetector analogic signals × 8 measuring nodes requires 24 ADC channels) and several GPIOs (8 IR LEDs and 8 UV LEDs to be controlled). For a different approach, a simpler microprocessor could be used complemented with a multiplexer; however, the sensitivity of the instrument could be affected due to the internal resistance of the multiplexer and it would slightly increase the cost and power consumption of the sensor. The sensor was intended to be supplied by a common 3.6 V Li-Ion battery. For the power supply circuit, the TPS73630DBVT low-dropout regulator (3 V fixed output, 400 mA max current, 75 mV dropout voltage and 200 uA quiescent current) was used, as presented in Figure 4. While the whole system works with 3 V, the UV LED UV3TZ-390-30 has a forward voltage of 3.4 V so a higher supply is needed. For this purpose, the TPS61222DCKR boost convertor was used to supply the UV LEDs circuit (95% efficiency, 5.5 uA quiescent current and 200 mA switching current). Voltage regulators were used since constant voltage is mandatory to minimise LED intensity variations. The LED circuits are controlled by the microprocessor GPIOs that switches a low-on resistance N-channel MOSFET ( ≈ 30 m Ω ) to turn them on and off. The phototransistor instrumentation circuited was designed with a simple current-to-voltage converter using a gain resistor. The resistor voltage potential is read by de microprocessor ADCs to be processed (Figure 4). For the communications, two low power RS485 drivers (LTC1480CS8) were used for two UARTs of the microprocessor. These two communication channels were designed by anticipating the modular integration of multiple bars along the water column, allowing to work as a single longer bar. One RS385 bus was used to connect to the bar above, and the other to the under one. The developed hardware has a power consumption of approximately 100 mA during the readings (5 ms) and 300 µ A in sleep mode. This means that using a common mobile-phone 3000 mA × 3.6 V lithium battery, the sensor has an autonomy of more than 1 year to take continuous measurements at a 1/min sample rate. Sensors 2020,20, 3194 6 of 16 Sensors 2020, 20, x FOR PEER REVIEW 6 of 17 Figure 4. Schematics of power supply (top section), LEDs circuit (bottom left section) and phototransistors circuit (bottom right section). For the communications, two low power RS485 drivers (LTC1480CS8) were used for two UARTs of the microprocessor. These two communication channels were designed by anticipating the modular integration of multiple bars along the water column, allowing to work as a single longer bar. One RS385 bus was used to connect to the bar above, and the other to the under one. The developed hardware has a power consumption of approximately 100 mA during the readings (5 ms) and 300 µA in sleep mode. This means that using a common mobile-phone 3000 mA × 3.6 V lithium battery, the sensor has an autonomy of more than 1 year to take continuous measurements at a 1/min sample rate. 3. In-lab Calibration To provide reliable data in situ measurements, the sensors must be calibrated. In the previous work [21], the SPM Sensor was calibrated with formazine (NTU), seashore sand (concentration of suspended particles in g/L) and organic matter using a phytoplankton solution (to demonstrate the efficiency in distinguishing organic from inorganic material). For this new device, and since the major objective is the measurement of sediment circulation at different depths in the water column, a calibration with seashore sand should be used. However, in past experiments calibrating the SPM Sensor, to get a homogeneous solution and keep the particles in suspension, a mixer needed to be used. For this new device and due to its “long” size, it was unpractical to use the same experimental design to calibrate the sensing nodes. The alternative was to perform the calibration only with formazine, which is a homogeneous solution, and no mixer was needed. Despite turbidity units and suspended particles concentration are not interchangeable, a change in turbidity is directly related to a change in suspended particle concentration. 3.1. Turbidity Calibration (NTU) To perform the turbidity calibration, the procedure described in [21] was used. Starting with a commercial 4000NTU Formazine Standard solution, the sample was diluted in deionized water following the methods and procedures of the Hach Water Analysis Guide [30]: 𝐷𝑖𝑙𝑢𝑡𝑖𝑜𝑛𝑓𝑎𝑐𝑡𝑜𝑟 = 𝑣𝑜𝑙𝑢𝑚𝑒𝑡𝑜𝑡𝑎𝑙 𝑣𝑜𝑙𝑢𝑚𝑒 𝑠𝑎𝑚𝑝𝑙𝑒 = 𝑣𝑜𝑙𝑢𝑚𝑒 𝑑𝑒𝑖𝑜𝑛𝑖𝑧𝑒𝑑_𝑤𝑎𝑡𝑒𝑟 + 𝑣𝑜𝑙𝑢𝑚𝑒 𝑁𝑇𝑈_𝑠𝑎𝑚𝑝𝑙𝑒 𝑣𝑜𝑙𝑢𝑚𝑒 𝑁𝑇𝑈_𝑠𝑎𝑚𝑝𝑙𝑒 (1) Figure 4. Schematics of power supply ( top section ), LEDs circuit ( bottom left section ) and phototransistors circuit (bottom right section). 3. In-Lab Calibration To provide reliable data in situ measurements, the sensors must be calibrated. In the previous work [ 21 ], the SPM Sensor was calibrated with formazine (NTU), seashore sand (concentration of suspended particles in g/L) and organic matter using a phytoplankton solution (to demonstrate the efficiency in distinguishing organic from inorganic material). For this new device, and since the major objective is the measurement of sediment circulation at different depths in the water column, a calibration with seashore sand should be used. However, in past experiments calibrating the SPM Sensor, to get a homogeneous solution and keep the particles in suspension, a mixer needed to be used. For this new device and due to its “long” size, it was unpractical to use the same experimental design to calibrate the sensing nodes. The alternative was to perform the calibration only with formazine, which is a homogeneous solution, and no mixer was needed. Despite turbidity units and suspended particles concentration are not interchangeable, a change in turbidity is directly related to a change in suspended particle concentration. 3.1. Turbidity Calibration (NTU) To perform the turbidity calibration, the procedure described in [ 21 ] was used. Starting with a commercial 4000NTU Formazine Standard solution, the sample was diluted in deionized water following the methods and procedures of the Hach Water Analysis Guide [30]: Dilutionfactor =volumetotal volume sample = volume deionized_water +volume NTU_sample volume NTU_sample (1) For each calibration sample, 10 measurements were recorded for each photodetector (backscatter, IR transmitted, and UV transmitted) for every eight nodes. The sensor performed with a high accuracy, with a major reading error of 2 mV. This value is related to the ADC resolution. Figure 5shows the calibration results of the developed device, as well as the previous calibration curves of the SPM Sensor. The behaviour of the different techniques was as expected, according to the theory presented in SPM Sensor Manuscript [21]. Sensors 2020,20, 3194 7 of 16 In the backscattering technique, for the maximum turbidity value (4000NTU), the maximum optical scattering by suspended particles was achieved, similarly to the photodetector, which results in a higher electrical output. As the sample is diluted, the number of suspended particles decreases, as the optical scattering and electrical output of the backscatter photodetector. For the transmitted light techniques (both infrared and ultraviolet) the opposite happens. For the 4000NTU solution, the emitting light is highly reflected and absorbed in its direct path, so the sensed light by the photodetector is minimum. As the turbidity decreases, the passage of light increases as the electric output of the receiver transducer. The output voltages of the sensors, as a function of turbidity of the sample solution, is presented in Figure 5. The output voltages obtained in the three measuring techniques (IR backscattering, IR transmission, and UV transmission) are presented in three separate graphs. Each graph presents the output voltage of the eight nodes of the sediment bar, from top to bottom node, as well as the output voltage of the previously developed SPM sensor [21]. Sensors 2020, 20, x FOR PEER REVIEW 8 of 17 Figure 5. Formazine calibration results for the three light detection techniques of each of the 8 nodes and comparison with SPM Sensor curves. The developed device can detect changes in turbidity above 10NTU, which underperforms against the SPM Sensor, but is still adequate to the coastal areas where higher turbidity values are expected (as an example, water for human consumption are limited to 5NTU, which is clear water). At the upper limits of the scale, the backscattering sensor can measure higher turbidity values, since only 200–400 mV were obtained with 4000NTU. The calibration values from Figure 5 are used to feed a linearization table in the microprocessor of the sensor, used to calculate the turbidity in real time from the sensors output voltage during insitu deployments. 3.2. External Light Calibration Previous calibrations with the SPM Sensor showed that this type of optical device is influenced by the external light. This is extremely important for in situ monitoring, as the solar light can produce interferences in the measurements. To reduce the external light interference, the instrument was submerged in a water solution and exposed to an external light source, while measurements with the LEDs on and off were recorded. The graph in Figure 6 represents the difference between the voltage obtained when LEDs are turned on, and the voltage obtained when the LEDs are turned off, in the three proposed sensing technologies, when ambient light changes from dark (0%) to a predefined value of ambient light is a sunny day (100%). 0 200 400 600 800 0 500 1000 1500 2000 2500 3000 3500 0,01 0,1 1 10 100 1000 0 500 1000 1500 2000 2500 3000 Backscattering (mV) node1 (bottom) node2 node3 node4 node5 node6 node7 node8 (top) SPM Sensor Formazine Calibration IR Transmitted (mV) UV Transmitted (mV) Turbidity (NTU) Figure 5. Formazine calibration results for the three light detection techniques of each of the 8 nodes and comparison with SPM Sensor curves. Comparing the curves of the developed instrument with the SPM Sensor, the new device presents a lower sensitivity, mainly in IR and UV transmitted channels. This happened due to the change for smaller transducers (3 mm in the step of 5 mm used in SPM Sensor), since the new emitters have less radiant power and photodetectors less sensitivity. Note that the gain resistor used was the same for Sensors 2020,20, 3194 8 of 16 both devices (1 M Ω in Figure 4). It is important to notice that both transducers and gain resistors can be easily changed during the manufacture of a new sensor. The developed device can detect changes in turbidity above 10NTU, which underperforms against the SPM Sensor, but is still adequate to the coastal areas where higher turbidity values are expected (as an example, water for human consumption are limited to 5NTU, which is clear water). At the upper limits of the scale, the backscattering sensor can measure higher turbidity values, since only 200–400 mV were obtained with 4000NTU. The calibration values from Figure 5are used to feed a linearization table in the microprocessor of the sensor, used to calculate the turbidity in real time from the sensors output voltage during in-situ deployments. 3.2. External Light Calibration Previous calibrations with the SPM Sensor showed that this type of optical device is influenced by the external light. This is extremely important for in situ monitoring, as the solar light can produce interferences in the measurements. To reduce the external light interference, the instrument was submerged in a water solution and exposed to an external light source, while measurements with the LEDs on and offwere recorded. The graph in Figure 6represents the difference between the voltage obtained when LEDs are turned on, and the voltage obtained when the LEDs are turned off, in the three proposed sensing technologies, when ambient light changes from dark (0%) to a predefined value of ambient light is a sunny day (100%). Sensors 2020, 20, x FOR PEER REVIEW 9 of 17 Figure 6. External light calibration results. The device was submerged in a sample of water and exposed to an external light source while measurements with LEDs on and off were recorded. Since the instrument was submerged in a sample with constant turbidity, an increase of the electrical voltage output of the photodetectors (both for LEDs on and off measurements) is expected with the increase of the power luminosity of the external light. In this instrument (since the emitters/receivers pairs have lower sensitivity than previous SPM instrument), for backscatter and transmitted infrared channels, the difference between the output voltage with the LEDs on and the LEDs off remained constant, when ambient light changes from 0% to 100% (see Figure 6). This means that this difference (also called ON-OFF technique) can be used to measure turbidity, avoiding the interference of ambient light. However, the same did not happen with the ultraviolet transmitted channel, and calibration of the sensor as a function of ambient light is still needed. As for the SPM Sensor [21], a mathematical expression was calculated to eliminate the external light effect, with 𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑖𝑜𝑛𝑓𝑎𝑐𝑡𝑜𝑟 corresponding to the photodetector correction factor and 𝑜𝑓𝑓𝑣𝑎𝑙𝑢𝑒 the voltage measurement of the external light influence (measurement with IR LEDs off). The on-off measurement of the UV channel will be divided by this factor. 𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑖𝑜𝑛𝑓𝑎𝑐𝑡𝑜𝑟 = 0.0000002 ∗𝑜𝑓𝑓𝑣𝑎𝑙𝑢𝑒2− 0.005 ∗𝑜𝑓𝑓𝑣𝑎𝑙𝑢𝑒 + 1.0174 (2) Using the mathematical equation above, the turbidity measurements can be corrected as shown in Figure 7.The equations are used to process the in-situ data and to eliminate the offset caused by any kind of external light. The factor is applied to the electrical on-off value that afterwards is used to estimate the turbidity value using the linearization table of the NTU calibration. 0100 0 100 200 300 400 500 600 700 800 On-off measurement (mV) External light (%) backscatter ir transmitted uv transmitted External light calibration Figure 6. External light calibration results. The device was submerged in a sample of water and exposed to an external light source while measurements with LEDs on and offwere recorded. Since the instrument was submerged in a sample with constant turbidity, an increase of the electrical voltage output of the photodetectors (both for LEDs on and offmeasurements) is expected with the increase of the power luminosity of the external light. In this instrument (since the emitters/receivers pairs have lower sensitivity than previous SPM instrument), for backscatter and transmitted infrared channels, the difference between the output voltage with the LEDs on and the LEDs offremained constant, when ambient light changes from 0% to 100% (see Figure 6). This means that this difference (also called ON-OFF technique) can be used to measure turbidity, avoiding the interference of ambient light. However, the same did not happen with the ultraviolet transmitted channel, and calibration of the sensor as a function of ambient light is still needed. As for the SPM Sensor [ 21 ], a mathematical expression was calculated to eliminate the external light effect, with correctionfactor corresponding to the photodetector correction factor and of fvalue the voltage measurement of the external light influence (measurement with IR LEDs off). The on-off measurement of the UV channel will be divided by this factor. Sensors 2020,20, 3194 9 of 16 correctionfactor =0.0000002 ∗o f fvalue2−0.005 ∗o f fvalue +1.0174 (2) Using the mathematical equation above, the turbidity measurements can be corrected as shown in Figure 7.The equations are used to process the in-situ data and to eliminate the offset caused by any kind of external light. The factor is applied to the electrical on-offvalue that afterwards is used to estimate the turbidity value using the linearization table of the NTU calibration. Sensors 2020, 20, x FOR PEER REVIEW 10 of 17 Figure 7. Demonstration of the effectiveness of the developed technique to correct external light influence. The blue squares and the straight line show the on-off measurement curve of Figure 6. In green squares and dashed line, the on-off measurement curve but with the correction factor of (2) applied. 4. In-lab Evaluation with Seashore Sand To test the behaviour of the developed instrument, an in-lab evaluation was performed with seashore sand, collected from the local where the device was intended to be deployed. The instrument was submerged in a container full of deionized water (minimum turbidity value), and seashore sand was gradually released from the top until it settles. This movement was repeated until the container became saturated with sand (see Figure 8). Figure 8. In-lab experiment design. The sensor was submerged in a container with deionized water (1) and seashores sand was released from the top (2). The release was repeated (3) until the container became saturated with sand (4). During the experiment, the electrical output of the photodetectors of all nodes was recorded with a sample period of 200 ms. The graph in Figure 9. shows the measurement of the backscatter channels of the eight nodes of the sensor. 0100 0 100 200 300 400 500 600 700 800 Output (mV) External light (%) on-off using correction factor External light correction techniques Figure 7. Demonstration of the effectiveness of the developed technique to correct external light influence. The blue squares and the straight line show the on-offmeasurement curve of Figure 6. In green squares and dashed line, the on-offmeasurement curve but with the correction factor of (2) applied. 4. In-Lab Evaluation with Seashore Sand To test the behaviour of the developed instrument, an in-lab evaluation was performed with seashore sand, collected from the local where the device was intended to be deployed. The instrument was submerged in a container full of deionized water (minimum turbidity value), and seashore sand was gradually released from the top until it settles. This movement was repeated until the container became saturated with sand (see Figure 8). Sensors 2020, 20, x FOR PEER REVIEW 10 of 17 Figure 7. Demonstration of the effectiveness of the developed technique to correct external light influence. The blue squares and the straight line show the on-off measurement curve of Figure 6. In green squares and dashed line, the on-off measurement curve but with the correction factor of (2) applied. 4. In-lab Evaluation with Seashore Sand To test the behaviour of the developed instrument, an in-lab evaluation was performed with seashore sand, collected from the local where the device was intended to be deployed. The instrument was submerged in a container full of deionized water (minimum turbidity value), and seashore sand was gradually released from the top until it settles. This movement was repeated until the container became saturated with sand (see Figure 8). Figure 8. In-lab experiment design. The sensor was submerged in a container with deionized water (1) and seashores sand was released from the top (2). The release was repeated (3) until the container became saturated with sand (4). During the experiment, the electrical output of the photodetectors of all nodes was recorded with a sample period of 200 ms. The graph in Figure 9. shows the measurement of the backscatter channels of the eight nodes of the sensor. 0100 0 100 200 300 400 500 600 700 800 Output (mV) External light (%) on-off using correction factor External light correction techniques Figure 8. In-lab experiment design. The sensor was submerged in a container with deionized water ( 1 ) and seashores sand was released from the top ( 2 ). The release was repeated ( 3 ) until the container became saturated with sand (4). Sensors 2020,20, 3194 16 of 16 18. Crawford, A.M.; Hay, A.E. Determining suspended sand size and concentration from multifrequency acoustic backscatter. J. Acoust. Soc. Am. 1993,94, 3312. [CrossRef] 19. Gartner, J.W. Estimation of Suspended Solids Concentrations Based on Acoustic Backscatter Intensity: Theoretical Background. In Proceedings of the Turbidity and Other Sediment Surrogates Workshop, Reno, NV, USA, 30 April–2 May 2002. 20. Davies-Colley, R.J.; Smith, D.G. Turbidity, suspended sediment, and water clarity: A review. J. Am. Water Resour. Assoc. 2001,37, 1085–1101. [CrossRef] 21. Matos, T.; Faria, C.L.; Martins, M.S.; Henriques, R.; Gomes, P.A.; Goncalves, L.M. Development of a Cost-Effective Optical Sensor for Continuous Monitoring of Turbidity and Suspended Particulate Matter in Marine Environment. Sensors 2019,19, 4439. [CrossRef] [PubMed] 22. Baptista, J.P.; Matos, T.; Lopes, S.F.; Faria, C.L.; Magalhaes, V.H.; Vieira, E.M.F.; Martins, M.S.; Goncalves, L.M.; Brito, F. A four-probe salinity sensor optimized for long-term autonomous marine deployments. In Proceedings of the OCEANS 2019, Marseille, France, 17–20 June 2019; pp. 1–5. [CrossRef] 23. Faria, C.L.; Martins, M.S.; Lima, R.A.; Goncalves, L.M.; Matos, T. Optimization of an Electromagnetic Generator for Underwater Energy Harvester. In Proceedings of the OCEANS 2019, Marseille, France, 17–20 June 2019; pp. 1–6. [CrossRef] 24. Martins, M.S.; Faria, C.L.; Matos, T.; Goncalves, L.M.; Silva, A.; Jesus, S.M.; Cruz, N. Performance evaluation of a PVDF hydrophone for deep sea applications. In Proceedings of the OCEANS 2019, Marseille, France, 17–20 June 2019; pp. 1–5. [CrossRef] 25. Ronald, J.; Zaneveld, V.; Spinrad, R.W.; Bartz, R. Optical Properties Of Turbidity Standards. In Proceedings of the Ocean Optics VI, Monterey, CA, USA, 26 March 1980; Volume 208, pp. 159–169. [CrossRef] 26. Bin Omar, A.F.; Bin MatJafri, M.Z. Turbidimeter design and analysis: A review on optical fiber sensors for the measurement of water turbidity. Sensors 2009,9, 8311–8335. [CrossRef] [PubMed] 27. Khouja, H. Turbidimetry & Nephelometry. Cell Biochem. Funct. 2010,11, 1–3. 28. US EPA. Method 180.1: Determination of Turbidity by Nephelometry; US EPA: Cincinnati, OH, USA, 1993. 29. Snyder, W.A.; Arnone, R.; Davis, C.O.; Goode, W.; Gould, R.W.; Ladner, S.; Lamela, G.; Rhea, W.J.; Stavn, R.; Sydor, M.; et al. Optical scattering and backscattering by organic and inorganic particulates in U.S. coastal waters. Appl. Opt. 2008,47, 666–677. [CrossRef] [PubMed] 30. Hach. Water Analysis Guide 09/2013; Hach: Loveland, CO, USA, 2013. 31. Matos, T.; Faria, C.L.; Martins, M.; Henriques, R.; Goncalves, L. Optical device for in situ monitoring of suspended particulate matter and organic/inorganic distinguish. In Proceedings of the OCEANS 2019, Marseille, France, 17–20 June 2019; pp. 1–6. [CrossRef] © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).