Soil moisture profile mapping using L- and P-band radiometry

This thesis aimed to develop a multi-frequency (i.e. L-band and P-band) soil moisture profile retrieval model to provide more comprehensive soil moisture information over the soil profile. The research compared coherent and incoherent microwave emission models, revealing good consistency between model estimations and measurements for estimating both brightness temperature and soil moisture profile. Coherent models demonstrated slight advantages in dry bare soil conditions. The thesis also explored the effectiveness of machine learning algorithms including random forest, extreme gradient boosting, support vector regression, and deep learning algorithms, with random forest outperforming others in estimating soil moisture profile.