A Salinity-Temperature Sensor Based on Microwave Resonance Reflection

Abstract

We developed and tested a microwave in situ salinity sensor (MiSSo) to simultaneously measure salinity and temperature within the same water sample over broad ranges of salinity (S) (3−50 psu) and temperature (T) (3−30 °C). Modern aquatic S sensors rely on measurements of conductivity (C) between a set of electrodes contained within a small volume of water. To determine water salt content or S, conductivity, or C, measurements must be augmented with concurrent T measurements from the same water volume. In practice, modern S sensors do not sample C and T within the same volume, resulting in the S determination characterized by measurement artifacts. These artifacts render processing vast amounts of available C and T data to derive S time-consuming and generally preclude automated processing. Our MiSSo approach eliminates the need for an additional T sensor, as it permits us to concurrently determine the sample S and T within the same water volume. Laboratory trials demonstrated the MiSSo accuracy of S and T measurements to be <0.1 psu and <0.1 °C, respectively, when using microwave reflections at 11 distinct frequencies. Each measurement took 0.1 μs. Our results demonstrate a new physical method that permits the accurate S and T determination within the same water volume.

Keywords: aquatic salinity measurements; aquatic temperature measurements; environmental monitoring.

Figure 1

(A) Schematics of the microwave reflection measurement system: TX, transmitter; RX, receiver and the probe containing the antenna. (B) Experimental setup and sensor location: (1) MiniSonde; (2) fast thermistors FP07; (3–5) 1/2 wavelength monopole antennas; (6) submerged part of the sensor. (C) Example of sample reflection measurement $S_{11}\left(f\right)$ with $S=10$ psu and $T=20$ °C as a function of frequency f. The measured $S_{11}\left(f\right)$ (black line) can be decomposed into the Signal (1) reflected off the sample (red line) and the signal caused by TX/RX impedance mismatch (not shown here). $f_0$ is the main resonance frequency.

Figure 2

(A) An example of temperature and salinity determination for four frequencies. The color-coded least-squares deviation of a point located at ($S_{MiSSo}$, $T_{MiSSo}$) from minimum value of $\Delta S_{11}(S,T,f_i)$, for four selected $f_i$ frequencies 2.1222, 2.3411, 2.7454, 2.9255 GHz. In this example, the MiSSo-found S and T was $S_{MiSSo}=10.05$ psu and $T_{MiSSo}=15.02$ °C. Concurrently, sample T and S values were independently measured by the CT sensor (MiniSonde 4a) and were $T_{MiniSonde}=15.06$ °C and $S_{MiniSonde}=11.00$ psu. (B) The histogram of deviations $\delta (S_{MiSSo}-S_{MiniSonde})$ and $\delta (T_{MiSSo}-T_{MiniSonde})$ of the measured T and S from the seawater S and T at selected four frequencies: 2.1222, 2.3411, 2.7454, 2.9255 GHz. Here, the 0.05 dB noise was added to the received signal.

Figure 3

Time series of MiSSo-retrieved temperature and salinity compared to readings of commercial CTD sensor MiniSonde 4a. Red denotes the MiSSo data, and black the MiniSonde 4a measurements. In each experiment of ≈120 min duration, salinity was approximately constant, while the temperature was computer-controlled between 12 and 30 °C. The abrupt increases in salinity reflect the change in sample salinity between (120 min) experiments. (A) Time series of T and S data spanning 4 days of salinity and temperature measurements (9 experiments), without added noise. (B) Subset of the time series collected over a day (2 experiments) without added noise. (C) The same subset of the time series as in (B), but before data processing, we added a noise of 0.05 dB to the measured $S_{11}$ to simulate, for example biofouling effects. An increase in MiSSo-derived S coincided with an increase in external noise. The MiSSo time series were obtained using 11 $f_i$ frequencies: 2.5166, 2.6124, 2.6170, 2.7355, 2.7401, 2.7454, 2.7918, 2.7940, 2.8427, 2.9331, and 2.9339 GHz.