This study is the first attempt to assess and enhance the applicability of exponential filter (ExpF) model to estimate rootzone soil moisture (RZSM) across different climatic and land conditions over the Tibetan Plateau (TP). To this end, in situ soil moisture (SM) profile data collected from five regional-scale monitoring networks are firstly used for the model assessment at local scale. Then a random forest (RF) approach is adopted to regionalize the critical model parameter (i.e. characteristic time length T) to enhance the estimation of RZSM at plateau scale using satellite-based surface SM (SSM) data from the Copernicus Climate Change Service (C3S). Results indicate that with increasing soil depth, the application accuracy of the ExpF model decreases due to weakened coupling strength between the SSM and RZSM, while the T values and their spatial heterogeneity increase due to greater variability in hydraulic connectivity between the SSM and RZSM across different monitoring sites. From the arid west TP to the semi-arid and humid east TP, the application accuracy of the ExpF model increases, while the variability in model accuracy across different monitoring sites shows a decreasing trend. The spatial distribution of regionalized T values at shallower depth is opposite to that of sand content, with higher values in the northwest TP and lower values in the Qaidam basin and central TP. For the deeper depth, higher values are also noted in the southeast TP due to the increasingly significant impact of meteorological and vegetation factors. The RZSM estimations based on the ExpF model with the regionalized T values capture better the observed dynamics and largely resolve the deficiencies noted for the C3S-based RZSM product. Above findings confirm the applicability of the ExpF model on the TP, offering valuable insights into enhancing the accuracy of RZSM estimations based on the satellite-based SSM products.