Accurate
and continuous monitoring of soil nitrogen is critical
for determining its fate and providing early warning for swift soil
nutrient management. However, the accuracy of existing electrochemical
sensors is hurdled by the immobility of targeted ions, ion adsorption
to soil particles, and sensor reading noise and drifting over time.
In this study, polyacrylamide hydrogel with a thickness of 0.45 μm
was coated on the surface of solid-state ion-selective membrane (S-ISM)
sensors to absorb water contained in soil and, consequently, enhance
the accuracy (R
2 > 0.98) and stability
(drifting < 0.3 mV/h) of these sensors monitoring ammonium (NH4
+) and nitrate (NO3
−) ions in soil. An ion transport model was built to simulate the long-term
NH4
+ dynamic process (R
2 > 0.7) by considering the soil adsorption process and
soil
complexity. Furthermore, a soil-based denoising data processing algorithm
(S-DDPA) was developed based on the unique features of soil sensors
including the nonlinear mass transfer and ion diffusion on the heterogeneous
sensor–hydrogel–soil interface. The 14 day tests using
real-world soil demonstrated the effectiveness of S-DDPA to eliminate
false signals and retrieve the actual soil nitrogen information for
accurate (error: <2 mg/L) and continuous monitoring.