Remote sensing has emerged as a critical tool for soil moisture estimation [4]. Our initial work employed drone-based optical remote sensing, which provided high-resolution imagery but was constrained by limited spatial coverage and operational scalability. To overcome these limitations, this study utilizes Synthetic Aperture Radar (SAR) data from the RADARSAT Constellation Mission (RCM) satellites. Unlike optical sensors, SAR can penetrate cloud cover and vegetation, facilitating all-weather soil moisture monitoring at large spatial scales. However, SAR-derived soil moisture retrieval is influenced by surface roughness, vegetation density, and soil dielectric properties, necessitating calibration with in situ ground measurements. At the farm scale, additional challenges arise due to sensor noise and environmental variability. To mitigate these issues, reference soil moisture sensors are employed as ground truth feedback to perform systematic bias correction within a model. This approach has the potential to enhance the reliability of soil moisture estimates by reducing noise and improving spatial accuracy, thereby supporting closed-loop irrigation applications.
This study investigates the integration of RCM SAR data with in situ soil moisture sensors to improve soil moisture sensing accuracy. Key SAR-based features, including backscatter intensity, polarization ratios, and coherence metrics, are extracted and utilized in a neural network model for soil moisture inference. To improve model accuracy, we compare the soil moisture values from the model with ground point sensor measurements, quantifying systematic bias. The calculated bias is then spatially propagated to refine the sensed soil moisture values across the field. The methodologies developed contribute to the advancement of large-scale closed-loop irrigation systems for agriculture applications.
References
1. UN Water. World Water Development Report—Water for a Sustainable World. United Nations, 2015.
2. World Economic Forum. The Global Risks Report 2019, 14th Edition. World Economic Forum, Geneva, Switzerland, 2019.
3. Shah, S. L., Bakshi, B. R., Liu, J., Georgakis, C., Chachuat, B., Braatz, R. D., & Young, B. R. (2021). Meeting the challenge of water sustainability: The role of process systems engineering. AIChE Journal, 67(2), e17113.
4. Abbes, A. B., Jarray, N., & Farah, I. R. (2024). Advances in remote sensing-based soil moisture retrieval: Applications, techniques, scales, and challenges for combining machine learning and physical models. Artificial Intelligence Review, 57(9), 224.