ABSTRACT Soil moisture plays a significant role in the global hydrological cycle in the field of agriculture, environment science, urban planning and construction, forestry and many more. Soil moisture can be determined using various approaches, however some of the limitations in these approaches are longer time duration, labor intensive, require installation and maintenance, expensive, lower resolution, etc. This research work proposes a soil moisture detection method which is portable, economic and time efficient. Spectroscopic sensor enables rapid soil moisture estimation using photodiodes of varying sensitivities and wavelengths across a broad spectrum of light, allowing precise identification and analysis of various samples based on their unique spectral fingerprints. In this study, a relation between sensor output data and soil moisture percentage value obtained from gravimetric method has been established. Using various regression methods, the highest value of correlation coefficient was found in case of polynomial function of order 3. Residual values of all wavelengths were then calculated by comparing the calculated with the expected moisture values obtained using the polynomial function. Result showed that 705 nanometers (nm) wavelength has the least residual value, thus signifying that the polynomial equation of order 3 at 705 nm wavelength perfectly demonstrates the relation between soil moisture and sensor reading. Furthermore, the results were verified using optimization method which showed that 705 nm wavelength has the highest closeness coefficient, thus verifying this model and observations. This model can be helpful in determining the soil moisture instantaneously instead of hours as seen in traditional gravimetric methods.
Read full abstract