Efficient and accurate monitoring of seawater temperature and salinity is crucial for marine resource exploitation, ecosystem protection, and assessing concrete structure durability. Compared to conventional conductivity-based sensors, optical fiber sensors are rapidly advancing in marine environmental exploration due to their advantages, such as corrosion resistance, compact size, resistance to electromagnetic interference, and ease of long-distance signal transmission. Previous optical fiber thermohaline sensors often have limited sensitivity, complex preparation processes, and assume a linear, non-interfering relationship between interference angle wavelength and temperature/salinity, with data decoupling achieved through matrix equations. This assumption, however, leads to significant calculation deviations. Therefore, developing a high-sensitivity fiber optic sensor for temperature and salinity measurement, along with a high-precision decoupling algorithm, is essential. In this paper, we propose a semi-open cavity Mach?Zehnder interferometer (MZI) sensor, whose sensing path is in direct contact with seawater. Variations in temperature and salinity influence the refractive index of seawater, altering the phase difference between two interfering light beams. By tracking the drift in interference angle wavelength, both temperature and salinity can be measured with high sensitivity. To address the crosstalk between temperature and salinity, a quadratic polynomial surface fitting nonlinear decoupling algorithm is applied, effectively eliminating crosstalk and reducing measurement deviations.

A segment of single-mode fiber (SMF) is placed between two coaxial multi-mode fibers (MMF 1 and MMF 2) through lateral offset splicing. MMF 1 and MMF 2 are spliced with input and output SMF sections, respectively. Light introduced into the SMF expands within MMF 1 before encountering the first offset weld. It then divides into two parts: one path propagates in seawater as the sensing arm, while the other travels through the SMF envelope as a reference arm. MMF 2 then couples the light from seawater and the reference arm back into the outgoing SMF. Variations in seawater temperature and salinity alter the phase difference between the two interference paths, shifting the MZI spectrum. By monitoring this interference angle wavelength, the sensor can measure temperature and salinity with high sensitivity. The simultaneous measurement of temperature and salinity is performed in real-time using the quadratic polynomial surface fitting nonlinear decoupling algorithm. To assess the accuracy of this decoupling approach, results are compared to both the transfer matrix method and a nonlinear decoupling algorithm without interaction terms. This comparison demonstrates the importance of interaction terms and validates the effectiveness of the quadratic polynomial surface fitting-based decoupling approach, which minimizes both maximum and average temperature and salinity errors.

Under a fixed salinity of 30‰, the interference inclination positions (Dip 1 and Dip 2) at temperatures of 18, 20, 22, 25, 27, and 32 ℃ are determined to the MZI’s temperature response characteristics (Fig. 5). As temperature increases, the MZI spectrum’s interference inclination shifts towards longer wavelengths. The relationship between interference inclination position and seawater temperature is best described by a quadratic polynomial, with a maximum temperature sensitivity of 2.1636 nm/℃ for Dip 1 and 1.8997 nm/℃ for Dip 2. For salinity response testing, seawater samples with salinities of 30‰, 33‰, 35‰, 37‰, and 40‰ are prepared and tested at a constant temperature of 18 ℃ (Fig. 6). With increasing salinity, the MZI’s spectrum’s interference inclination moves towards short wavelengths. Linear fitting of the interference inclination wavelength against salinity results in salinity sensitivities of -2.65 nm/‰ for Dip 1 and -2.5948 nm/‰ for Dip 2, respectively. To address temperature?salinity crosstalk, a quadratic polynomial surface fitting decoupling algorithm is used, achieving a maximum temperature deviation of -0.4031 ℃ and a salinity deviation of -0.1242‰, with an average deviation of 0.1599 ℃ and 0.0779‰, respectively.

In this paper, we propose an all-fiber MZI structure based on core offset, proving that a nonlinear decoupling algorithm using quadratic polynomial surface fitting is effective for simultaneous seawater temperature and salinity measurement. The sensor’s semi-open cavity serves as the sensing path, in direct contact with seawater, while the biased SMF envelope acts as the reference path. Changes in seawater properties alter the phase difference between these transmission paths. By recording two selected interference inclination wavelengths in the transmission spectrum, temperature and salinity can be measured with high precision. The experimental and correlation analyses show that the quadratic function more accurately models the relationship between the interference angle wavelength and seawater temperature, while both quadratic and linear functions can describe the relationship with seawater salinity. Due to the cross-interference between temperature and salinity, a quadratic polynomial surface fitting algorithm is applied to demodulate these parameters, effectively eliminating crosstalk and reducing measurement deviation. The sensor demonstrates strong repeatability, good stability, and high precision, providing a valuable reference for detecting seawater environmental parameters.